diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,47898 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 7978,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.3223,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 4.127,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-05,
+      "loss": 3.8926,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 4.0557,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0833333333333333e-05,
+      "loss": 0.2629,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 2.4141,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9166666666666666e-05,
+      "loss": 2.4883,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 2.0317,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.75e-05,
+      "loss": 1.8884,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1666666666666665e-05,
+      "loss": 2.0662,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 2.0518,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 1.8203,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.416666666666667e-05,
+      "loss": 2.2256,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.833333333333333e-05,
+      "loss": 2.2903,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-05,
+      "loss": 2.3442,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 1.8438,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.083333333333334e-05,
+      "loss": 2.0557,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-05,
+      "loss": 2.0825,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.916666666666666e-05,
+      "loss": 2.2671,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-05,
+      "loss": 2.1284,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.75e-05,
+      "loss": 2.3442,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 2.2158,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.583333333333334e-05,
+      "loss": 2.1316,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 2.0645,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010416666666666667,
+      "loss": 2.7192,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010833333333333334,
+      "loss": 1.793,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 1.9534,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011666666666666667,
+      "loss": 1.8796,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012083333333333333,
+      "loss": 1.896,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000125,
+      "loss": 2.5396,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012916666666666667,
+      "loss": 1.7197,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 2.1172,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001375,
+      "loss": 2.1733,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014166666666666668,
+      "loss": 2.103,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014583333333333335,
+      "loss": 1.3877,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015,
+      "loss": 1.9116,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015416666666666668,
+      "loss": 1.9229,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015833333333333332,
+      "loss": 2.0737,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 2.2156,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 2.1489,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017083333333333333,
+      "loss": 1.9587,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000175,
+      "loss": 1.5857,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017916666666666667,
+      "loss": 2.7075,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.2922,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001875,
+      "loss": 1.582,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019166666666666667,
+      "loss": 1.9648,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019583333333333334,
+      "loss": 2.2886,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.9338,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020416666666666668,
+      "loss": 1.926,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020833333333333335,
+      "loss": 1.9324,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002125,
+      "loss": 2.1191,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021666666666666668,
+      "loss": 2.1582,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022083333333333333,
+      "loss": 1.9312,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 2.0001,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022916666666666666,
+      "loss": 1.7617,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023333333333333333,
+      "loss": 2.0029,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002375,
+      "loss": 1.9915,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024166666666666667,
+      "loss": 2.064,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002458333333333333,
+      "loss": 1.9434,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025,
+      "loss": 2.0388,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025416666666666665,
+      "loss": 2.1049,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025833333333333334,
+      "loss": 1.8016,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026250000000000004,
+      "loss": 1.8149,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002666666666666667,
+      "loss": 1.9885,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002708333333333333,
+      "loss": 1.8438,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000275,
+      "loss": 2.1909,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027916666666666666,
+      "loss": 1.5735,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028333333333333335,
+      "loss": 1.9688,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002875,
+      "loss": 2.2919,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002916666666666667,
+      "loss": 1.8003,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029583333333333333,
+      "loss": 2.2383,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003,
+      "loss": 2.2507,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030416666666666667,
+      "loss": 1.9038,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030833333333333337,
+      "loss": 1.938,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003125,
+      "loss": 2.3398,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031666666666666665,
+      "loss": 2.0679,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032083333333333334,
+      "loss": 0.0775,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032500000000000004,
+      "loss": 1.8989,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003291666666666667,
+      "loss": 1.9902,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 1.8933,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003375,
+      "loss": 1.7324,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034166666666666666,
+      "loss": 2.0344,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034583333333333335,
+      "loss": 1.9165,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035,
+      "loss": 1.9918,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003541666666666667,
+      "loss": 1.7848,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035833333333333333,
+      "loss": 0.0726,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003625,
+      "loss": 1.8579,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036666666666666667,
+      "loss": 1.6489,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037083333333333337,
+      "loss": 1.8225,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000375,
+      "loss": 1.8835,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037916666666666665,
+      "loss": 0.0617,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038333333333333334,
+      "loss": 1.8457,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038750000000000004,
+      "loss": 1.6519,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003916666666666667,
+      "loss": 1.6523,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003958333333333333,
+      "loss": 1.8003,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 1.9146,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040416666666666666,
+      "loss": 1.8191,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040833333333333336,
+      "loss": 1.8691,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004125,
+      "loss": 1.9321,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004166666666666667,
+      "loss": 1.6443,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042083333333333333,
+      "loss": 1.9814,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000425,
+      "loss": 2.0627,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042916666666666667,
+      "loss": 1.7913,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043333333333333337,
+      "loss": 1.8848,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004375,
+      "loss": 1.7351,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044166666666666665,
+      "loss": 1.7881,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044583333333333335,
+      "loss": 1.5391,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 2.3174,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004541666666666667,
+      "loss": 1.9868,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004583333333333333,
+      "loss": 2.1895,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004625,
+      "loss": 1.7886,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046666666666666666,
+      "loss": 0.0535,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047083333333333336,
+      "loss": 1.9224,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000475,
+      "loss": 1.7793,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004791666666666667,
+      "loss": 1.8384,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00048333333333333334,
+      "loss": 1.5626,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004875,
+      "loss": 1.9985,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004916666666666666,
+      "loss": 1.7812,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004958333333333334,
+      "loss": 1.9089,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.0747,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005041666666666667,
+      "loss": 1.9824,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005083333333333333,
+      "loss": 1.5085,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005124999999999999,
+      "loss": 1.7146,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005166666666666667,
+      "loss": 1.8816,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005208333333333334,
+      "loss": 2.1106,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005250000000000001,
+      "loss": 0.0483,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005291666666666667,
+      "loss": 1.5497,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005333333333333334,
+      "loss": 1.8591,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005375,
+      "loss": 1.7373,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005416666666666666,
+      "loss": 1.938,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005458333333333333,
+      "loss": 1.7251,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00055,
+      "loss": 1.6508,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005541666666666667,
+      "loss": 1.8608,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005583333333333333,
+      "loss": 2.1145,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005625000000000001,
+      "loss": 2.1765,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005666666666666667,
+      "loss": 1.7561,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005708333333333333,
+      "loss": 1.7012,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000575,
+      "loss": 1.9019,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005791666666666667,
+      "loss": 1.7783,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005833333333333334,
+      "loss": 2.1406,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005875,
+      "loss": 1.8555,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005916666666666667,
+      "loss": 2.0938,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005958333333333333,
+      "loss": 1.9917,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006,
+      "loss": 0.0415,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006041666666666666,
+      "loss": 1.6831,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006083333333333333,
+      "loss": 1.8027,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006125000000000001,
+      "loss": 1.7866,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006166666666666667,
+      "loss": 2.0618,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006208333333333334,
+      "loss": 1.9504,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000625,
+      "loss": 2.1223,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006291666666666667,
+      "loss": 1.6636,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006333333333333333,
+      "loss": 1.6157,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006374999999999999,
+      "loss": 2.0474,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006416666666666667,
+      "loss": 1.9028,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006458333333333334,
+      "loss": 2.0981,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 1.6831,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006541666666666667,
+      "loss": 1.7529,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006583333333333334,
+      "loss": 2.0774,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006625,
+      "loss": 1.8049,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 1.9082,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006708333333333333,
+      "loss": 1.9712,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000675,
+      "loss": 1.5513,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006791666666666667,
+      "loss": 1.8667,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006833333333333333,
+      "loss": 1.8022,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006875,
+      "loss": 1.9116,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006916666666666667,
+      "loss": 1.634,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006958333333333334,
+      "loss": 1.897,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007,
+      "loss": 2.1929,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007041666666666667,
+      "loss": 1.8545,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007083333333333334,
+      "loss": 1.9246,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007125,
+      "loss": 1.6248,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007166666666666667,
+      "loss": 1.9287,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007208333333333333,
+      "loss": 1.7083,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000725,
+      "loss": 1.7501,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007291666666666666,
+      "loss": 1.7725,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007333333333333333,
+      "loss": 2.123,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007375000000000001,
+      "loss": 1.7395,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007416666666666667,
+      "loss": 2.0637,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007458333333333334,
+      "loss": 1.8148,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00075,
+      "loss": 1.9844,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007541666666666667,
+      "loss": 2.0479,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007583333333333333,
+      "loss": 1.7937,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007624999999999999,
+      "loss": 1.9899,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007666666666666667,
+      "loss": 1.9817,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007708333333333334,
+      "loss": 1.7769,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007750000000000001,
+      "loss": 1.7415,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007791666666666667,
+      "loss": 1.9185,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007833333333333334,
+      "loss": 1.7571,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007875,
+      "loss": 1.6211,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007916666666666666,
+      "loss": 0.0876,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007958333333333333,
+      "loss": 1.7236,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008,
+      "loss": 2.0649,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008041666666666667,
+      "loss": 1.7312,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008083333333333333,
+      "loss": 1.8079,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008125000000000001,
+      "loss": 1.5933,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008166666666666667,
+      "loss": 1.7231,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008208333333333334,
+      "loss": 2.1333,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000825,
+      "loss": 1.6195,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008291666666666667,
+      "loss": 1.9077,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008333333333333334,
+      "loss": 1.9771,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008375,
+      "loss": 1.6016,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008416666666666667,
+      "loss": 1.6119,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008458333333333333,
+      "loss": 1.7527,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00085,
+      "loss": 1.8054,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008541666666666666,
+      "loss": 1.2957,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008583333333333333,
+      "loss": 1.9165,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008625000000000001,
+      "loss": 1.8892,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008666666666666667,
+      "loss": 1.6238,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008708333333333334,
+      "loss": 1.6209,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000875,
+      "loss": 1.6309,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008791666666666667,
+      "loss": 1.4824,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008833333333333333,
+      "loss": 0.1401,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008874999999999999,
+      "loss": 1.7722,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008916666666666667,
+      "loss": 1.9277,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008958333333333334,
+      "loss": 1.8037,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 1.6921,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009041666666666667,
+      "loss": 1.853,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009083333333333334,
+      "loss": 1.5697,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009125,
+      "loss": 0.1616,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009166666666666666,
+      "loss": 1.7329,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009208333333333333,
+      "loss": 1.5236,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000925,
+      "loss": 1.8635,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009291666666666667,
+      "loss": 1.9521,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009333333333333333,
+      "loss": 1.9097,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009375,
+      "loss": 1.871,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009416666666666667,
+      "loss": 1.5881,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009458333333333334,
+      "loss": 1.6685,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00095,
+      "loss": 1.4482,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009541666666666667,
+      "loss": 1.6606,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009583333333333334,
+      "loss": 1.4314,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009625,
+      "loss": 1.7632,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009666666666666667,
+      "loss": 1.6663,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009708333333333333,
+      "loss": 1.939,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000975,
+      "loss": 1.7214,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009791666666666666,
+      "loss": 0.0847,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009833333333333332,
+      "loss": 1.6309,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009875,
+      "loss": 1.6685,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009916666666666667,
+      "loss": 1.6516,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009958333333333334,
+      "loss": 1.6868,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 1.6003,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999999958791928,
+      "loss": 2.0054,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999998351677183,
+      "loss": 1.9526,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999996291273914,
+      "loss": 1.8474,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999993406709813,
+      "loss": 1.7935,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999989697985357,
+      "loss": 1.6775,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999985165101157,
+      "loss": 1.686,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999979808057958,
+      "loss": 1.6436,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999973626856644,
+      "loss": 1.7859,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999966621498234,
+      "loss": 1.7202,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999958791983882,
+      "loss": 1.9204,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999995013831488,
+      "loss": 0.0677,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999940660492654,
+      "loss": 0.0599,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999930358518766,
+      "loss": 1.708,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999919232394914,
+      "loss": 1.689,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999990728212293,
+      "loss": 1.3828,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999989450770479,
+      "loss": 1.4258,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999880909142592,
+      "loss": 1.7661,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999866486438583,
+      "loss": 1.832,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999851239595138,
+      "loss": 1.6843,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999835168614769,
+      "loss": 1.6294,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999818273500125,
+      "loss": 1.6528,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999800554253994,
+      "loss": 1.6147,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999782010879296,
+      "loss": 1.5176,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999762643379086,
+      "loss": 1.5469,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999742451756556,
+      "loss": 1.5005,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999721436015037,
+      "loss": 1.3838,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999969959615799,
+      "loss": 1.4685,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999676932189016,
+      "loss": 1.5271,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999653444111853,
+      "loss": 1.5613,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999629131930369,
+      "loss": 1.6445,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999603995648572,
+      "loss": 1.7358,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999578035270607,
+      "loss": 1.3657,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999955125080075,
+      "loss": 1.7734,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999523642243422,
+      "loss": 1.7703,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999949520960317,
+      "loss": 1.5586,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999465952884679,
+      "loss": 1.5767,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999435872092772,
+      "loss": 1.582,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999940496723241,
+      "loss": 1.4988,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999373238308684,
+      "loss": 1.7017,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999340685326826,
+      "loss": 1.6499,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999307308292203,
+      "loss": 1.8,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999273107210314,
+      "loss": 1.7495,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999238082086797,
+      "loss": 1.5054,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999202232927424,
+      "loss": 1.7073,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999165559738108,
+      "loss": 1.4745,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999912806252489,
+      "loss": 1.6069,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999089741293953,
+      "loss": 1.3877,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999050596051612,
+      "loss": 1.3289,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999010626804323,
+      "loss": 1.5205,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999896983355867,
+      "loss": 1.4475,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998928216321376,
+      "loss": 1.8267,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998885775099307,
+      "loss": 1.5017,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998842509899456,
+      "loss": 1.4438,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998798420728949,
+      "loss": 1.457,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998753507595062,
+      "loss": 1.5503,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998707770505191,
+      "loss": 1.4534,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999866120946688,
+      "loss": 1.5859,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998613824487802,
+      "loss": 1.5491,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998565615575768,
+      "loss": 1.4456,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998516582738722,
+      "loss": 1.4233,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999846672598475,
+      "loss": 1.5161,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998416045322064,
+      "loss": 1.6433,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998364540759026,
+      "loss": 1.6572,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998312212304119,
+      "loss": 1.2319,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998259059965972,
+      "loss": 1.3123,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998205083753342,
+      "loss": 1.8081,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998150283675135,
+      "loss": 1.4368,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998094659740373,
+      "loss": 1.224,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999803821195823,
+      "loss": 1.2954,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997980940338012,
+      "loss": 1.4701,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997922844889155,
+      "loss": 1.3218,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997863925621238,
+      "loss": 1.6989,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997804182543973,
+      "loss": 1.5339,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997743615667205,
+      "loss": 1.4172,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999768222500092,
+      "loss": 1.4321,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997620010555235,
+      "loss": 1.2856,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997556972340404,
+      "loss": 1.4012,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997493110366824,
+      "loss": 1.4788,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997428424645015,
+      "loss": 1.5518,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997362915185643,
+      "loss": 1.5645,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997296581999503,
+      "loss": 1.3779,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999722942509753,
+      "loss": 1.4329,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997161444490795,
+      "loss": 1.2688,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997092640190502,
+      "loss": 1.2819,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997023012207995,
+      "loss": 1.3628,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996952560554747,
+      "loss": 1.4634,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996881285242371,
+      "loss": 1.4856,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996809186282618,
+      "loss": 1.4377,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999673626368737,
+      "loss": 1.6587,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999666251746865,
+      "loss": 1.314,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996587947638611,
+      "loss": 1.4458,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996512554209544,
+      "loss": 1.5623,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996436337193879,
+      "loss": 1.5623,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996359296604177,
+      "loss": 1.3721,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996281432453138,
+      "loss": 1.3765,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996202744753594,
+      "loss": 1.3271,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999612323351852,
+      "loss": 1.397,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996042898761017,
+      "loss": 1.542,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999596174049433,
+      "loss": 1.3748,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995879758731836,
+      "loss": 1.5161,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995796953487046,
+      "loss": 1.5959,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995713324773613,
+      "loss": 1.4731,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995628872605316,
+      "loss": 1.3839,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995543596996081,
+      "loss": 1.2695,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999545749795996,
+      "loss": 1.3162,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999537057551115,
+      "loss": 1.4072,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995282829663975,
+      "loss": 1.2915,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995194260432898,
+      "loss": 1.3657,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995104867832518,
+      "loss": 1.3098,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999501465187757,
+      "loss": 1.519,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994923612582929,
+      "loss": 1.4912,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994831749963596,
+      "loss": 1.3411,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994739064034712,
+      "loss": 1.3555,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999464555481156,
+      "loss": 1.2939,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999455122230955,
+      "loss": 0.0867,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999445606654423,
+      "loss": 1.2161,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994360087531285,
+      "loss": 0.0579,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994263285286538,
+      "loss": 1.2944,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994165659825945,
+      "loss": 1.4592,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994067211165595,
+      "loss": 1.1677,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993967939321718,
+      "loss": 1.396,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993867844310675,
+      "loss": 1.1058,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993766926148966,
+      "loss": 1.3843,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993665184853226,
+      "loss": 1.4592,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993562620440226,
+      "loss": 1.1797,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999345923292687,
+      "loss": 1.3883,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993355022330202,
+      "loss": 1.4988,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993249988667398,
+      "loss": 1.2104,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993144131955768,
+      "loss": 1.3317,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993037452212766,
+      "loss": 1.2798,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992929949455972,
+      "loss": 1.3218,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992821623703108,
+      "loss": 1.2412,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992712474972028,
+      "loss": 1.2795,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992602503280726,
+      "loss": 1.2668,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992491708647325,
+      "loss": 0.0533,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992380091090093,
+      "loss": 1.215,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999226765062742,
+      "loss": 1.1584,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999215438727785,
+      "loss": 1.2629,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992040301060043,
+      "loss": 1.182,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999192539199281,
+      "loss": 1.302,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991809660095093,
+      "loss": 0.0514,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991693105385962,
+      "loss": 1.1018,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991575727884632,
+      "loss": 1.3418,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991457527610452,
+      "loss": 1.3536,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991338504582905,
+      "loss": 1.1804,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991218658821608,
+      "loss": 1.2324,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991097990346316,
+      "loss": 1.1973,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999097649917692,
+      "loss": 1.3335,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990854185333445,
+      "loss": 1.2549,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990731048836053,
+      "loss": 1.0865,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999060708970504,
+      "loss": 1.208,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990482307960838,
+      "loss": 1.3105,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990356703624018,
+      "loss": 1.3601,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999023027671528,
+      "loss": 1.0736,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990103027255463,
+      "loss": 1.2974,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989974955265546,
+      "loss": 1.1685,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989846060766635,
+      "loss": 1.4841,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989716343779982,
+      "loss": 1.1438,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989585804326961,
+      "loss": 1.063,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989454442429094,
+      "loss": 1.1113,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989322258108031,
+      "loss": 1.2773,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989189251385562,
+      "loss": 1.1787,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998905542228361,
+      "loss": 1.4731,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988920770824237,
+      "loss": 0.0479,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988785297029633,
+      "loss": 1.2544,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988649000922131,
+      "loss": 1.3342,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988511882524198,
+      "loss": 1.1631,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988373941858435,
+      "loss": 0.9307,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988235178947577,
+      "loss": 1.1365,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099880955938145,
+      "loss": 1.2656,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998795518648221,
+      "loss": 1.1172,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998781395697385,
+      "loss": 1.323,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987671905312702,
+      "loss": 1.2374,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987529031522178,
+      "loss": 1.1912,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987385335625829,
+      "loss": 1.5164,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987240817647342,
+      "loss": 1.1277,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987095477610537,
+      "loss": 1.3643,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998694931553937,
+      "loss": 1.2125,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986802331457933,
+      "loss": 1.1604,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986654525390457,
+      "loss": 1.3206,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986505897361302,
+      "loss": 1.0945,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986356447394967,
+      "loss": 1.3306,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986206175516088,
+      "loss": 1.1538,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986055081749434,
+      "loss": 1.0852,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985903166119907,
+      "loss": 1.271,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985750428652554,
+      "loss": 0.0459,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985596869372545,
+      "loss": 1.1234,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985442488305194,
+      "loss": 1.1619,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985287285475947,
+      "loss": 1.0275,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998513126091039,
+      "loss": 1.0642,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984974414634236,
+      "loss": 0.0456,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984816746673342,
+      "loss": 0.9827,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984658257053693,
+      "loss": 1.1487,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984498945801417,
+      "loss": 1.2043,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984338812942773,
+      "loss": 1.269,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984177858504155,
+      "loss": 1.1458,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984016082512093,
+      "loss": 0.991,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983853484993252,
+      "loss": 1.1412,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983690065974438,
+      "loss": 1.0627,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983525825482582,
+      "loss": 1.157,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998336076354476,
+      "loss": 1.0964,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998319488018818,
+      "loss": 1.0618,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983028175440181,
+      "loss": 1.1675,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982860649328242,
+      "loss": 1.1512,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998269230187998,
+      "loss": 0.0434,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982523133123145,
+      "loss": 1.0127,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982353143085617,
+      "loss": 1.1509,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982182331795416,
+      "loss": 1.2061,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099820106992807,
+      "loss": 1.0491,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981838245569758,
+      "loss": 1.1091,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981664970691018,
+      "loss": 1.1204,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981490874673038,
+      "loss": 0.0424,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981315957544518,
+      "loss": 0.9866,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981140219334288,
+      "loss": 1.1033,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980963660071317,
+      "loss": 1.0938,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980786279784703,
+      "loss": 1.1982,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998060807850369,
+      "loss": 1.0563,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980429056257647,
+      "loss": 0.8628,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980249213076085,
+      "loss": 1.0812,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980068548988649,
+      "loss": 1.1807,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979887064025114,
+      "loss": 0.9387,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979704758215399,
+      "loss": 1.2581,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997952163158955,
+      "loss": 0.9902,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979337684177755,
+      "loss": 0.0413,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979152916010335,
+      "loss": 1.0525,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978967327117744,
+      "loss": 1.0327,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978780917530572,
+      "loss": 1.012,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978593687279547,
+      "loss": 0.8512,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997840563639553,
+      "loss": 1.5005,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997821676490952,
+      "loss": 0.0407,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978027072852648,
+      "loss": 0.0407,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977836560256179,
+      "loss": 1.2268,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997764522715152,
+      "loss": 1.0901,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977453073570202,
+      "loss": 0.9287,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977260099543906,
+      "loss": 1.1189,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977066305104434,
+      "loss": 0.9897,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976871690283735,
+      "loss": 0.9998,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976676255113884,
+      "loss": 1.259,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976479999627096,
+      "loss": 0.9282,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976282923855719,
+      "loss": 0.9941,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976085027832242,
+      "loss": 1.01,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975886311589278,
+      "loss": 1.1167,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975686775159586,
+      "loss": 1.0452,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997548641857606,
+      "loss": 1.0179,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975285241871714,
+      "loss": 1.0415,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997508324507972,
+      "loss": 0.9438,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974880428233366,
+      "loss": 1.2441,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974676791366086,
+      "loss": 1.2214,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974472334511445,
+      "loss": 1.2759,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974267057703146,
+      "loss": 1.0214,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974060960975023,
+      "loss": 1.1372,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973854044361046,
+      "loss": 0.994,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973646307895325,
+      "loss": 1.1084,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099734377516121,
+      "loss": 0.884,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973228375545749,
+      "loss": 1.116,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973018179730783,
+      "loss": 1.0356,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972807164201848,
+      "loss": 0.0412,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972595328993728,
+      "loss": 1.1167,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997238267414134,
+      "loss": 0.9502,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972169199679735,
+      "loss": 0.9351,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099719549056441,
+      "loss": 0.9344,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971739792069762,
+      "loss": 1.0518,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971523858992174,
+      "loss": 0.9856,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997130710644693,
+      "loss": 0.9089,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997108953446976,
+      "loss": 0.9895,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970871143096525,
+      "loss": 1.0176,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997065193236322,
+      "loss": 0.887,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970431902305984,
+      "loss": 0.9613,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997021105296108,
+      "loss": 1.0724,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969989384364915,
+      "loss": 0.9077,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969766896554023,
+      "loss": 1.0305,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969543589565084,
+      "loss": 1.0232,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969319463434898,
+      "loss": 1.1052,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969094518200413,
+      "loss": 1.0364,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968868753898706,
+      "loss": 0.8582,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996864217056699,
+      "loss": 0.9165,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968414768242615,
+      "loss": 0.9827,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968186546963063,
+      "loss": 1.1433,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996795750676595,
+      "loss": 0.8981,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967727647689033,
+      "loss": 0.809,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967496969770198,
+      "loss": 0.9922,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996726547304747,
+      "loss": 0.0834,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967033157559004,
+      "loss": 0.9431,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966800023343096,
+      "loss": 0.8623,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966566070438172,
+      "loss": 0.8716,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966331298882798,
+      "loss": 0.8468,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996609570871567,
+      "loss": 0.8716,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996585929997562,
+      "loss": 1.0408,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965622072701614,
+      "loss": 1.1611,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996538402693276,
+      "loss": 0.8606,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965145162708292,
+      "loss": 0.9196,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964905480067585,
+      "loss": 0.8282,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964664979050144,
+      "loss": 0.9237,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964423659695612,
+      "loss": 0.8632,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964181522043767,
+      "loss": 0.8751,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996393856613452,
+      "loss": 1.0089,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963694792007919,
+      "loss": 0.8801,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963450199704145,
+      "loss": 1.1189,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963204789263517,
+      "loss": 0.1014,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962958560726482,
+      "loss": 0.7834,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962711514133631,
+      "loss": 0.9054,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996246364952568,
+      "loss": 0.9358,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962214966943493,
+      "loss": 0.9166,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961965466428054,
+      "loss": 0.8441,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961715148020492,
+      "loss": 0.7448,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961464011762067,
+      "loss": 0.967,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961212057694173,
+      "loss": 0.8617,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960959285858339,
+      "loss": 0.3604,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960705696296236,
+      "loss": 1.4731,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996045128904966,
+      "loss": 0.7899,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960196064160542,
+      "loss": 0.8646,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959940021670958,
+      "loss": 0.8923,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959683161623106,
+      "loss": 0.9165,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995942548405933,
+      "loss": 0.7933,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959166989022099,
+      "loss": 0.8956,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958907676554025,
+      "loss": 0.9802,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995864754669785,
+      "loss": 0.9521,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995838659949645,
+      "loss": 0.8789,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995812483499284,
+      "loss": 0.7645,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957862253230166,
+      "loss": 0.9594,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995759885425171,
+      "loss": 0.9314,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957334638100888,
+      "loss": 0.8412,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957069604821253,
+      "loss": 0.9204,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995680375445649,
+      "loss": 0.6946,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956537087050421,
+      "loss": 0.8702,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956269602647,
+      "loss": 0.9398,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956001301290316,
+      "loss": 0.8048,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955732183024595,
+      "loss": 0.8748,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955462247894199,
+      "loss": 1.791,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955191495943617,
+      "loss": 0.8732,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954919927217482,
+      "loss": 0.8567,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954647541760553,
+      "loss": 0.8839,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995437433961773,
+      "loss": 0.7505,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995410032083405,
+      "loss": 0.8604,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953825485454672,
+      "loss": 0.8147,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953549833524905,
+      "loss": 0.9001,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995327336509018,
+      "loss": 0.858,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952996080196072,
+      "loss": 0.885,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952717978888282,
+      "loss": 0.8464,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952439061212655,
+      "loss": 0.749,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952159327215162,
+      "loss": 0.8544,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951878776941912,
+      "loss": 0.8839,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951597410439153,
+      "loss": 0.797,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951315227753259,
+      "loss": 0.8123,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951032228930744,
+      "loss": 0.8385,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995074841401826,
+      "loss": 0.833,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950463783062577,
+      "loss": 0.8967,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950178336110625,
+      "loss": 0.7747,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949892073209446,
+      "loss": 0.8344,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949604994406228,
+      "loss": 0.7568,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949317099748293,
+      "loss": 0.8228,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949028389283092,
+      "loss": 0.9071,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948738863058213,
+      "loss": 0.7562,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948448521121384,
+      "loss": 0.8135,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994815736352046,
+      "loss": 0.7341,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947865390303435,
+      "loss": 0.7385,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947572601518432,
+      "loss": 1.0645,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947278997213711,
+      "loss": 0.8505,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946984577437674,
+      "loss": 0.6611,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946689342238849,
+      "loss": 0.6714,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946393291665895,
+      "loss": 0.7489,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946096425767616,
+      "loss": 0.0578,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945798744592942,
+      "loss": 0.8787,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945500248190943,
+      "loss": 0.7007,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994520093661082,
+      "loss": 0.7874,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994490080990191,
+      "loss": 0.6423,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994459986811368,
+      "loss": 0.8314,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944298111295737,
+      "loss": 0.7676,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943995539497823,
+      "loss": 0.7705,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943692152769809,
+      "loss": 0.0612,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943387951161702,
+      "loss": 0.7032,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943082934723646,
+      "loss": 0.9469,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942777103505917,
+      "loss": 0.8187,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942470457558928,
+      "loss": 0.9288,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942162996933218,
+      "loss": 0.6572,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941854721679474,
+      "loss": 0.722,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941545631848502,
+      "loss": 0.7384,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941235727491259,
+      "loss": 0.7482,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994092500865882,
+      "loss": 0.6576,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940613475402404,
+      "loss": 0.8513,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940301127773362,
+      "loss": 0.6309,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939987965823177,
+      "loss": 0.6951,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939673989603473,
+      "loss": 0.657,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939359199165998,
+      "loss": 0.5944,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939043594562643,
+      "loss": 0.8179,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938727175845428,
+      "loss": 0.8198,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993840994306651,
+      "loss": 0.6672,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938091896278178,
+      "loss": 0.7524,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937773035532857,
+      "loss": 0.6716,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993745336088311,
+      "loss": 0.5999,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993713287238162,
+      "loss": 0.8099,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936811570081224,
+      "loss": 0.6697,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936489454034877,
+      "loss": 0.8048,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936166524295673,
+      "loss": 0.6462,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935842780916847,
+      "loss": 0.7137,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935518223951757,
+      "loss": 0.5479,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935192853453903,
+      "loss": 0.6101,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934866669476915,
+      "loss": 0.8945,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934539672074563,
+      "loss": 0.6177,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934211861300742,
+      "loss": 0.6194,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933883237209487,
+      "loss": 0.9369,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933553799854966,
+      "loss": 0.6594,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933223549291482,
+      "loss": 0.5992,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993289248557347,
+      "loss": 0.6891,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099325606087555,
+      "loss": 0.7119,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932227918892276,
+      "loss": 0.6395,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931894416038635,
+      "loss": 0.562,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931560100249553,
+      "loss": 0.6277,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993122497158013,
+      "loss": 0.7152,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930889030085612,
+      "loss": 0.7003,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930552275821368,
+      "loss": 0.5974,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993021470884291,
+      "loss": 0.7859,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929876329205875,
+      "loss": 0.7692,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929537136966046,
+      "loss": 0.5959,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929197132179327,
+      "loss": 0.6725,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928856314901763,
+      "loss": 0.656,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928514685189533,
+      "loss": 0.6824,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928172243098949,
+      "loss": 0.6903,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927828988686453,
+      "loss": 0.6143,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927484922008629,
+      "loss": 0.7892,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927140043122188,
+      "loss": 0.7964,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926794352083977,
+      "loss": 0.5833,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926447848950977,
+      "loss": 0.6472,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926100533780304,
+      "loss": 0.7209,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925752406629205,
+      "loss": 0.8071,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925403467555063,
+      "loss": 0.6554,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925053716615397,
+      "loss": 0.7891,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924703153867856,
+      "loss": 0.5647,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992435177937022,
+      "loss": 0.5863,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923999593180414,
+      "loss": 0.0529,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923646595356482,
+      "loss": 0.7268,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923292785956616,
+      "loss": 0.7675,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992293816503913,
+      "loss": 0.6976,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922582732662484,
+      "loss": 0.5105,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922226488885257,
+      "loss": 0.7208,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921869433766172,
+      "loss": 0.6841,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921511567364085,
+      "loss": 0.7794,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921152889737985,
+      "loss": 0.619,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920793400946988,
+      "loss": 0.6909,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920433101050356,
+      "loss": 0.9028,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920071990107474,
+      "loss": 0.4569,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919710068177865,
+      "loss": 0.5696,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919347335321186,
+      "loss": 0.8322,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918983791597227,
+      "loss": 0.5846,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918619437065913,
+      "loss": 0.5913,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918254271787299,
+      "loss": 0.5651,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917888295821577,
+      "loss": 0.6622,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917521509229075,
+      "loss": 0.5518,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917153912070245,
+      "loss": 0.5869,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916785504405683,
+      "loss": 0.5817,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916416286296114,
+      "loss": 0.6609,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916046257802398,
+      "loss": 0.4675,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915675418985526,
+      "loss": 0.5707,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915303769906622,
+      "loss": 0.4331,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914931310626952,
+      "loss": 0.7073,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914558041207904,
+      "loss": 0.6759,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914183961711005,
+      "loss": 0.618,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991380907219792,
+      "loss": 0.6208,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913433372730437,
+      "loss": 0.7748,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913056863370487,
+      "loss": 0.0491,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991267954418013,
+      "loss": 0.5248,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991230141522156,
+      "loss": 0.5968,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911922476557107,
+      "loss": 0.7867,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991154272824923,
+      "loss": 0.5016,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911162170360523,
+      "loss": 0.7037,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910780802953718,
+      "loss": 0.678,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910398626091672,
+      "loss": 0.6472,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910015639837384,
+      "loss": 0.5229,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990963184425398,
+      "loss": 0.5422,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909247239404725,
+      "loss": 0.5376,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908861825353007,
+      "loss": 0.552,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908475602162365,
+      "loss": 0.7676,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908088569896454,
+      "loss": 0.7411,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990770072861907,
+      "loss": 0.6343,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907312078394146,
+      "loss": 0.6885,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990692261928574,
+      "loss": 0.5884,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906532351358047,
+      "loss": 0.0521,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906141274675397,
+      "loss": 0.5806,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905749389302254,
+      "loss": 0.5978,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905356695303212,
+      "loss": 0.6398,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904963192743,
+      "loss": 0.6042,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904568881686478,
+      "loss": 0.5896,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904173762198642,
+      "loss": 0.4543,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903777834344622,
+      "loss": 0.7919,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903381098189678,
+      "loss": 0.4966,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902983553799207,
+      "loss": 0.6486,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902585201238734,
+      "loss": 0.0516,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902186040573924,
+      "loss": 0.7756,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901786071870567,
+      "loss": 0.704,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901385295194595,
+      "loss": 0.7562,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900983710612067,
+      "loss": 0.7064,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900581318189178,
+      "loss": 0.7601,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900178117992257,
+      "loss": 0.5723,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899774110087758,
+      "loss": 0.5294,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989936929454228,
+      "loss": 0.874,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989896367142255,
+      "loss": 0.6027,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898557240795424,
+      "loss": 0.7151,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898150002727898,
+      "loss": 0.6012,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897741957287098,
+      "loss": 0.5143,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897333104540282,
+      "loss": 0.754,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989692344455484,
+      "loss": 0.658,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896512977398302,
+      "loss": 0.6863,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896101703138322,
+      "loss": 0.5546,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895689621842695,
+      "loss": 0.6515,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895276733579341,
+      "loss": 0.582,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989486303841632,
+      "loss": 0.7549,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894448536421822,
+      "loss": 0.6365,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989403322766417,
+      "loss": 0.5587,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893617112211822,
+      "loss": 0.5292,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893200190133364,
+      "loss": 0.47,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989278246149752,
+      "loss": 0.5106,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892363926373145,
+      "loss": 0.5386,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989194458482923,
+      "loss": 0.4763,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989152443693489,
+      "loss": 0.7772,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891103482759383,
+      "loss": 0.6663,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890681722372096,
+      "loss": 0.5859,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890259155842546,
+      "loss": 0.8274,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988983578324039,
+      "loss": 0.5955,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988941160463541,
+      "loss": 0.6124,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888986620097523,
+      "loss": 0.6027,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888560829696786,
+      "loss": 0.6119,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888134233503377,
+      "loss": 0.762,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887706831587615,
+      "loss": 0.7119,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988727862401995,
+      "loss": 0.6958,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886849610870965,
+      "loss": 0.5985,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886419792211375,
+      "loss": 0.0513,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885989168112027,
+      "loss": 0.7419,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885557738643902,
+      "loss": 0.6583,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885125503878115,
+      "loss": 0.63,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988469246388591,
+      "loss": 0.6124,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884258618738666,
+      "loss": 0.4539,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883823968507895,
+      "loss": 0.6968,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883388513265243,
+      "loss": 0.7157,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882952253082488,
+      "loss": 0.5875,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882515188031535,
+      "loss": 0.5625,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882077318184434,
+      "loss": 0.6571,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988163864361335,
+      "loss": 0.6577,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881199164390598,
+      "loss": 0.4906,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880758880588616,
+      "loss": 0.6361,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988031779227998,
+      "loss": 0.7323,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879875899537391,
+      "loss": 0.7021,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987943320243369,
+      "loss": 0.6533,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878989701041846,
+      "loss": 0.5173,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878545395434965,
+      "loss": 0.6166,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987810028568628,
+      "loss": 0.6057,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987765437186916,
+      "loss": 0.5055,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987720765405711,
+      "loss": 0.5747,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876760132323757,
+      "loss": 0.5754,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876311806742873,
+      "loss": 0.5115,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875862677388352,
+      "loss": 0.6588,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875412744334226,
+      "loss": 0.5066,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874962007654661,
+      "loss": 0.7047,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874510467423952,
+      "loss": 0.7911,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874058123716526,
+      "loss": 0.5764,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873604976606944,
+      "loss": 0.6101,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098731510261699,
+      "loss": 0.6746,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872696272480222,
+      "loss": 0.6471,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872240715612863,
+      "loss": 0.6848,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871784355642917,
+      "loss": 0.5504,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871327192645604,
+      "loss": 0.6233,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870869226696285,
+      "loss": 0.7566,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987041045787044,
+      "loss": 0.7509,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869950886243694,
+      "loss": 0.6266,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869490511891798,
+      "loss": 0.7794,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869029334890638,
+      "loss": 0.5414,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868567355316227,
+      "loss": 0.4448,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868104573244718,
+      "loss": 0.5345,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867640988752392,
+      "loss": 0.6251,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986717660191566,
+      "loss": 0.7649,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986671141281107,
+      "loss": 0.613,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866245421515298,
+      "loss": 0.7483,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986577862810516,
+      "loss": 0.597,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865311032657591,
+      "loss": 0.6497,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864842635249672,
+      "loss": 0.5511,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864373435958607,
+      "loss": 0.6693,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863903434861735,
+      "loss": 0.0508,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863432632036529,
+      "loss": 0.563,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986296102756059,
+      "loss": 0.8492,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986248862151166,
+      "loss": 0.6343,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862015413967598,
+      "loss": 0.0508,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861541405006408,
+      "loss": 0.7087,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861066594706225,
+      "loss": 0.5841,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860590983145306,
+      "loss": 0.4662,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860114570402056,
+      "loss": 0.5746,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859637356554993,
+      "loss": 0.6549,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859159341682787,
+      "loss": 0.634,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858680525864224,
+      "loss": 0.6753,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985820090917823,
+      "loss": 0.6257,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857720491703863,
+      "loss": 0.516,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857239273520307,
+      "loss": 0.5455,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856757254706888,
+      "loss": 0.6936,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856274435343054,
+      "loss": 0.679,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985579081550839,
+      "loss": 0.4371,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855306395282616,
+      "loss": 0.4534,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854821174745574,
+      "loss": 0.4688,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854335153977247,
+      "loss": 0.5604,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853848333057748,
+      "loss": 0.6177,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985336071206732,
+      "loss": 0.4196,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852872291086338,
+      "loss": 0.6304,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985238307019531,
+      "loss": 0.4905,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851893049474874,
+      "loss": 0.682,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851402229005804,
+      "loss": 0.6494,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850910608869,
+      "loss": 0.4883,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098504181891455,
+      "loss": 0.527,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849924969916467,
+      "loss": 0.5477,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849430951263203,
+      "loss": 0.5591,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848936133267138,
+      "loss": 0.5527,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848440516009832,
+      "loss": 0.6525,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847944099572977,
+      "loss": 0.5664,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847446884038403,
+      "loss": 0.7172,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846948869488067,
+      "loss": 0.6871,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846450056004053,
+      "loss": 0.5328,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845950443668585,
+      "loss": 0.835,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845450032564015,
+      "loss": 1.0186,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844948822772827,
+      "loss": 0.502,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844446814377638,
+      "loss": 0.6454,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843944007461192,
+      "loss": 0.6858,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843440402106367,
+      "loss": 0.4691,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842935998396179,
+      "loss": 0.6594,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842430796413767,
+      "loss": 0.5483,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841924796242403,
+      "loss": 0.5109,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841417997965493,
+      "loss": 0.6204,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840910401666574,
+      "loss": 0.5729,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840402007429317,
+      "loss": 0.5562,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839892815337516,
+      "loss": 0.6046,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839382825475107,
+      "loss": 0.7278,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838872037926153,
+      "loss": 0.6818,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838360452774843,
+      "loss": 0.6017,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837848070105508,
+      "loss": 0.63,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837334890002605,
+      "loss": 0.6298,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836820912550721,
+      "loss": 0.7139,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836306137834575,
+      "loss": 0.5371,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983579056593902,
+      "loss": 0.5752,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983527419694904,
+      "loss": 0.5001,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834757030949747,
+      "loss": 0.6259,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834239068026388,
+      "loss": 0.6865,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983372030826434,
+      "loss": 0.5968,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983320075174911,
+      "loss": 0.5977,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832680398566339,
+      "loss": 0.6681,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832159248801798,
+      "loss": 0.6924,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983163730254139,
+      "loss": 0.5078,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831114559871147,
+      "loss": 0.6982,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830591020877234,
+      "loss": 0.5389,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830066685645946,
+      "loss": 0.0498,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829541554263715,
+      "loss": 0.5243,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829015626817094,
+      "loss": 0.5779,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982848890339278,
+      "loss": 0.6997,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827961384077583,
+      "loss": 0.5499,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827433068958467,
+      "loss": 0.6063,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826903958122506,
+      "loss": 0.6799,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826374051656922,
+      "loss": 0.5376,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825843349649055,
+      "loss": 0.6322,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825311852186386,
+      "loss": 0.7278,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982477955935652,
+      "loss": 0.8173,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824246471247199,
+      "loss": 0.5519,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823712587946289,
+      "loss": 0.5765,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823177909541795,
+      "loss": 0.488,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822642436121845,
+      "loss": 0.5541,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822106167774707,
+      "loss": 0.6033,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821569104588774,
+      "loss": 0.7223,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982103124665257,
+      "loss": 0.5009,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820492594054752,
+      "loss": 0.632,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819953146884108,
+      "loss": 0.6974,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819412905229554,
+      "loss": 0.6093,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818871869180144,
+      "loss": 0.4677,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818330038825053,
+      "loss": 0.4812,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817787414253595,
+      "loss": 0.5826,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817243995555213,
+      "loss": 0.5493,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816699782819475,
+      "loss": 0.4825,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981615477613609,
+      "loss": 0.6909,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981560897559489,
+      "loss": 0.6649,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815062381285842,
+      "loss": 0.7512,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814514993299044,
+      "loss": 0.0495,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813966811724718,
+      "loss": 0.7384,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813417836653227,
+      "loss": 0.6602,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812868068175057,
+      "loss": 0.4802,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812317506380829,
+      "loss": 0.7205,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811766151361292,
+      "loss": 0.6869,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811214003207327,
+      "loss": 0.4608,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810661062009949,
+      "loss": 0.572,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810107327860298,
+      "loss": 0.5305,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809552800849647,
+      "loss": 0.6355,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00098089974810694,
+      "loss": 0.0495,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808441368611095,
+      "loss": 0.6537,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807884463566392,
+      "loss": 0.7336,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980732676602709,
+      "loss": 0.5905,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806768276085114,
+      "loss": 0.5912,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806208993832523,
+      "loss": 0.4753,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805648919361503,
+      "loss": 0.5264,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805088052764375,
+      "loss": 0.5049,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804526394133586,
+      "loss": 0.4501,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803963943561715,
+      "loss": 0.7543,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803400701141475,
+      "loss": 0.5737,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802836666965701,
+      "loss": 0.6755,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802271841127367,
+      "loss": 0.796,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801706223719578,
+      "loss": 0.5917,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801139814835562,
+      "loss": 0.5474,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800572614568682,
+      "loss": 0.5186,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980000462301243,
+      "loss": 0.5221,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979943584026043,
+      "loss": 0.7611,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798866266406438,
+      "loss": 0.4612,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798295901544338,
+      "loss": 0.6235,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797724745768141,
+      "loss": 0.5256,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797152799171994,
+      "loss": 0.6855,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796580061850173,
+      "loss": 0.5387,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796006533897082,
+      "loss": 0.6053,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979543221540726,
+      "loss": 0.6003,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794857106475367,
+      "loss": 0.8309,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794281207196208,
+      "loss": 0.6488,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793704517664702,
+      "loss": 0.5557,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793127037975911,
+      "loss": 0.6016,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979254876822502,
+      "loss": 0.8694,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791969708507348,
+      "loss": 0.7107,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979138985891834,
+      "loss": 0.5187,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790809219553578,
+      "loss": 0.5726,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790227790508765,
+      "loss": 0.6095,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789645571879747,
+      "loss": 0.5736,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789062563762484,
+      "loss": 0.6115,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788478766253079,
+      "loss": 0.4757,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978789417944776,
+      "loss": 0.7664,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787308803442887,
+      "loss": 0.7067,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786722638334946,
+      "loss": 0.6371,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786135684220559,
+      "loss": 0.5815,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785547941196472,
+      "loss": 0.645,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784959409359566,
+      "loss": 0.4613,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978437008880685,
+      "loss": 0.6266,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783779979635463,
+      "loss": 0.5507,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783189081942673,
+      "loss": 0.4736,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782597395825879,
+      "loss": 0.5681,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782004921382612,
+      "loss": 0.7103,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009781411658710529,
+      "loss": 0.6252,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978081760790742,
+      "loss": 0.6156,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780222769071203,
+      "loss": 0.0504,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779627142299928,
+      "loss": 0.6716,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977903072769177,
+      "loss": 0.6052,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778433525345041,
+      "loss": 0.6484,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777835535358177,
+      "loss": 0.6761,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977723675782975,
+      "loss": 0.4625,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977663719285845,
+      "loss": 0.6151,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776036840543113,
+      "loss": 0.7039,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775435700982694,
+      "loss": 0.5845,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009774833774276277,
+      "loss": 0.4999,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977423106052308,
+      "loss": 0.4675,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773627559822454,
+      "loss": 0.4772,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977302327227387,
+      "loss": 0.5905,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772418197976937,
+      "loss": 0.4164,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771812337031392,
+      "loss": 0.5106,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771205689537096,
+      "loss": 0.672,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770598255594049,
+      "loss": 0.6716,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769990035302372,
+      "loss": 0.0926,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769381028762322,
+      "loss": 0.6617,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976877123607428,
+      "loss": 0.5433,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768160657338762,
+      "loss": 0.6593,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976754929265641,
+      "loss": 0.6871,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766937142127998,
+      "loss": 0.4918,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766324205854426,
+      "loss": 0.6613,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009765710483936728,
+      "loss": 0.4896,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765095976476063,
+      "loss": 0.4904,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764480683573722,
+      "loss": 0.626,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763864605331127,
+      "loss": 0.5262,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763247741849828,
+      "loss": 0.4789,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762630093231499,
+      "loss": 0.6538,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762011659577954,
+      "loss": 0.7125,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761392440991128,
+      "loss": 0.6144,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000976077243757309,
+      "loss": 0.5652,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760151649426034,
+      "loss": 0.7332,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759530076652289,
+      "loss": 0.639,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758907719354308,
+      "loss": 0.7562,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758284577634675,
+      "loss": 0.7245,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757660651596108,
+      "loss": 0.6514,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757035941341445,
+      "loss": 0.6031,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756410446973661,
+      "loss": 0.1797,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755784168595859,
+      "loss": 0.6075,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755157106311266,
+      "loss": 0.5746,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754529260223246,
+      "loss": 0.7286,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753900630435286,
+      "loss": 0.7177,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753271217051005,
+      "loss": 0.5144,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752641020174151,
+      "loss": 0.541,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752010039908601,
+      "loss": 0.5723,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975137827635836,
+      "loss": 0.6801,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750745729627565,
+      "loss": 0.5798,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750112399820477,
+      "loss": 0.5546,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749478287041492,
+      "loss": 0.0645,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748843391395131,
+      "loss": 0.5143,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748207712986046,
+      "loss": 0.58,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747571251919016,
+      "loss": 0.5962,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746934008298951,
+      "loss": 0.7697,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746295982230891,
+      "loss": 0.4849,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974565717382,
+      "loss": 0.6845,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745017583171576,
+      "loss": 0.5757,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744377210391045,
+      "loss": 0.7008,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974373605558396,
+      "loss": 0.6746,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743094118856004,
+      "loss": 0.6873,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974245140031299,
+      "loss": 0.433,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741807900060858,
+      "loss": 0.8115,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741163618205678,
+      "loss": 0.666,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740518554853647,
+      "loss": 0.6205,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739872710111093,
+      "loss": 0.6484,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739226084084473,
+      "loss": 0.463,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738578676880372,
+      "loss": 0.5911,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737930488605502,
+      "loss": 0.5986,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737281519366707,
+      "loss": 0.6166,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736631769270957,
+      "loss": 0.7261,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735981238425352,
+      "loss": 0.7484,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735329926937121,
+      "loss": 0.5377,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973467783491362,
+      "loss": 0.7867,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734024962462335,
+      "loss": 0.6046,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733371309690883,
+      "loss": 0.644,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732716876707004,
+      "loss": 0.7469,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732061663618571,
+      "loss": 0.6846,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731405670533584,
+      "loss": 0.6133,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730748897560172,
+      "loss": 0.7065,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730091344806593,
+      "loss": 0.7437,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972943301238123,
+      "loss": 0.5068,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728773900392603,
+      "loss": 0.5316,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728114008949349,
+      "loss": 0.7004,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727453338160244,
+      "loss": 0.5496,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009726791888134183,
+      "loss": 0.514,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097261296589802,
+      "loss": 0.5695,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725466650807448,
+      "loss": 0.5759,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724802863725213,
+      "loss": 0.5841,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724138297842909,
+      "loss": 0.6938,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723472953270077,
+      "loss": 0.5188,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722806830116388,
+      "loss": 0.6514,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722139928491641,
+      "loss": 0.5887,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721472248505762,
+      "loss": 0.6449,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720803790268807,
+      "loss": 0.5049,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720134553890959,
+      "loss": 0.5784,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971946453948253,
+      "loss": 0.6501,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971879374715396,
+      "loss": 0.4288,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718122177015815,
+      "loss": 0.6654,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717449829178798,
+      "loss": 0.8065,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716776703753725,
+      "loss": 0.837,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716102800851554,
+      "loss": 0.5041,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715428120583366,
+      "loss": 0.6781,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009714752663060369,
+      "loss": 0.7502,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097140764283939,
+      "loss": 0.5538,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713399416695425,
+      "loss": 0.5607,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712721628076536,
+      "loss": 0.7292,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712043062648955,
+      "loss": 0.5121,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711363720524533,
+      "loss": 0.6895,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710683601815246,
+      "loss": 0.631,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00097100027066332,
+      "loss": 0.6848,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709321035090628,
+      "loss": 0.0639,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708638587299893,
+      "loss": 0.6538,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707955363373482,
+      "loss": 0.5105,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707271363424015,
+      "loss": 0.5819,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706586587564235,
+      "loss": 0.5272,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705901035907017,
+      "loss": 0.6122,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705214708565361,
+      "loss": 0.4752,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704527605652396,
+      "loss": 0.594,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703839727281378,
+      "loss": 0.5521,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703151073565695,
+      "loss": 0.6467,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702461644618854,
+      "loss": 0.5217,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701771440554499,
+      "loss": 0.6371,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701080461486398,
+      "loss": 0.6172,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700388707528445,
+      "loss": 0.7501,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699696178794662,
+      "loss": 0.734,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699002875399204,
+      "loss": 0.5795,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698308797456348,
+      "loss": 0.5101,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697613945080499,
+      "loss": 0.5088,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696918318386192,
+      "loss": 0.5908,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696221917488091,
+      "loss": 0.5587,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695524742500982,
+      "loss": 0.6519,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694826793539784,
+      "loss": 0.5193,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694128070719541,
+      "loss": 0.7501,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693428574155425,
+      "loss": 0.5936,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692728303962734,
+      "loss": 0.7341,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692027260256899,
+      "loss": 0.8007,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691325443153472,
+      "loss": 0.541,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690622852768135,
+      "loss": 0.7039,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689919489216699,
+      "loss": 0.6384,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689215352615099,
+      "loss": 0.4988,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688510443079401,
+      "loss": 0.5534,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687804760725796,
+      "loss": 0.5873,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687098305670605,
+      "loss": 0.5414,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686391078030272,
+      "loss": 0.542,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685683077921372,
+      "loss": 0.0617,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684974305460607,
+      "loss": 0.5884,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684264760764806,
+      "loss": 0.7794,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009683554443950924,
+      "loss": 0.6517,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682843355136042,
+      "loss": 0.5323,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682131494437374,
+      "loss": 0.7045,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681418861972257,
+      "loss": 0.787,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680705457858154,
+      "loss": 0.6761,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679991282212658,
+      "loss": 0.8365,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679276335153488,
+      "loss": 0.775,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678560616798491,
+      "loss": 0.51,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967784412726564,
+      "loss": 0.67,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677126866673037,
+      "loss": 0.6517,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676408835138906,
+      "loss": 0.6212,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675690032781604,
+      "loss": 0.6029,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674970459719614,
+      "loss": 0.6145,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674250116071544,
+      "loss": 0.5652,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967352900195613,
+      "loss": 0.4795,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672807117492235,
+      "loss": 0.6215,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672084462798848,
+      "loss": 0.6464,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671361037995087,
+      "loss": 0.4897,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009670636843200194,
+      "loss": 0.6553,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669911878533541,
+      "loss": 0.4761,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669186144114626,
+      "loss": 0.6873,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668459640063072,
+      "loss": 0.6621,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667732366498633,
+      "loss": 0.4723,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667004323541185,
+      "loss": 0.4813,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666275511310731,
+      "loss": 0.5229,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665545929927406,
+      "loss": 0.6783,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664815579511469,
+      "loss": 0.6248,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664084460183303,
+      "loss": 0.5809,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663352572063422,
+      "loss": 0.6166,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662619915272463,
+      "loss": 0.5787,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661886489931193,
+      "loss": 0.6353,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661152296160504,
+      "loss": 0.5317,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660417334081412,
+      "loss": 0.8156,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009659681603815067,
+      "loss": 0.7102,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658945105482739,
+      "loss": 0.6243,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658207839205826,
+      "loss": 0.7863,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009657469805105855,
+      "loss": 0.6943,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656731003304475,
+      "loss": 0.0602,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655991433923468,
+      "loss": 0.6649,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655251097084736,
+      "loss": 0.6975,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654509992910312,
+      "loss": 0.5117,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653768121522355,
+      "loss": 0.5725,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653025483043145,
+      "loss": 0.6849,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652282077595098,
+      "loss": 0.5597,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651537905300747,
+      "loss": 0.4869,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650792966282759,
+      "loss": 0.7346,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650047260663921,
+      "loss": 0.6653,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009649300788567152,
+      "loss": 0.5137,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648553550115493,
+      "loss": 0.4642,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647805545432113,
+      "loss": 0.0589,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647056774640308,
+      "loss": 0.5402,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00096463072378635,
+      "loss": 0.5378,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645556935225237,
+      "loss": 0.5739,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644805866849192,
+      "loss": 0.5166,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644054032859165,
+      "loss": 0.4891,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643301433379084,
+      "loss": 0.7733,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642548068533001,
+      "loss": 0.6162,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641793938445093,
+      "loss": 0.5667,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964103904323967,
+      "loss": 0.5333,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640283383041158,
+      "loss": 0.6667,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639526957974117,
+      "loss": 0.6084,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638769768163228,
+      "loss": 0.5304,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638011813733303,
+      "loss": 0.5125,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637253094809276,
+      "loss": 0.6528,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636493611516208,
+      "loss": 0.6185,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635733363979286,
+      "loss": 0.4907,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634972352323825,
+      "loss": 0.5654,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634210576675265,
+      "loss": 0.5839,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633448037159167,
+      "loss": 0.6686,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632684733901227,
+      "loss": 0.4595,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631920667027259,
+      "loss": 0.6989,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631155836663205,
+      "loss": 0.6013,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009630390242935139,
+      "loss": 0.6136,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000962962388596925,
+      "loss": 0.7688,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628856765891861,
+      "loss": 0.485,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628088882829416,
+      "loss": 0.5585,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000962732023690849,
+      "loss": 0.6169,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626550828255779,
+      "loss": 0.6659,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625780656998107,
+      "loss": 0.6771,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625009723262421,
+      "loss": 0.7874,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624238027175798,
+      "loss": 0.4923,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623465568865439,
+      "loss": 0.5264,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622692348458668,
+      "loss": 0.6283,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621918366082938,
+      "loss": 0.5878,
+      "step": 1204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621143621865826,
+      "loss": 0.682,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009620368115935034,
+      "loss": 0.5931,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619591848418392,
+      "loss": 0.5479,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618814819443853,
+      "loss": 0.6337,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618037029139496,
+      "loss": 0.5181,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617258477633527,
+      "loss": 0.4884,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616479165054276,
+      "loss": 0.6788,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615699091530199,
+      "loss": 0.5308,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614918257189876,
+      "loss": 0.4655,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614136662162016,
+      "loss": 0.5762,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000961335430657545,
+      "loss": 0.5352,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612571190559135,
+      "loss": 0.4886,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611787314242155,
+      "loss": 0.4384,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611002677753716,
+      "loss": 0.635,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610217281223153,
+      "loss": 0.5009,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609431124779927,
+      "loss": 0.5806,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608644208553616,
+      "loss": 0.71,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607856532673934,
+      "loss": 0.6531,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607068097270715,
+      "loss": 0.6431,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009606278902473917,
+      "loss": 0.5284,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605488948413626,
+      "loss": 0.4249,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960469823522005,
+      "loss": 0.6569,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603906763023526,
+      "loss": 0.5908,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603114531954514,
+      "loss": 0.6927,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602321542143599,
+      "loss": 0.5292,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601527793721491,
+      "loss": 0.4578,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600733286819025,
+      "loss": 0.058,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599938021567163,
+      "loss": 0.5813,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599141998096989,
+      "loss": 0.6016,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598345216539714,
+      "loss": 0.692,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009597547677026672,
+      "loss": 0.6377,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009596749379689325,
+      "loss": 0.672,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009595950324659259,
+      "loss": 0.8998,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009595150512068181,
+      "loss": 0.5098,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594349942047928,
+      "loss": 0.5591,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593548614730458,
+      "loss": 0.678,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592746530247859,
+      "loss": 0.6315,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591943688732337,
+      "loss": 0.4865,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591140090316228,
+      "loss": 0.6829,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590335735131991,
+      "loss": 0.499,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589530623312209,
+      "loss": 0.5721,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958872475498959,
+      "loss": 0.625,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587918130296969,
+      "loss": 0.4911,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587110749367301,
+      "loss": 0.4934,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586302612333672,
+      "loss": 0.6321,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585493719329285,
+      "loss": 0.8044,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584684070487475,
+      "loss": 0.4834,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583873665941697,
+      "loss": 0.645,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583062505825532,
+      "loss": 0.5762,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582250590272686,
+      "loss": 0.6,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009581437919416986,
+      "loss": 0.4315,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958062449339239,
+      "loss": 0.5643,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579810312332975,
+      "loss": 0.9166,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578995376372944,
+      "loss": 0.6659,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578179685646626,
+      "loss": 0.5972,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577363240288473,
+      "loss": 0.458,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576546040433062,
+      "loss": 0.468,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575728086215092,
+      "loss": 0.6393,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574909377769391,
+      "loss": 0.6548,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574089915230906,
+      "loss": 0.5863,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573269698734711,
+      "loss": 0.621,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572448728416005,
+      "loss": 0.5775,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571627004410113,
+      "loss": 0.5865,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570804526852477,
+      "loss": 0.6304,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569981295878672,
+      "loss": 0.5728,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569157311624388,
+      "loss": 0.5801,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568332574225449,
+      "loss": 0.6973,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567507083817797,
+      "loss": 0.5078,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009566680840537498,
+      "loss": 0.4688,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565853844520746,
+      "loss": 0.5497,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565026095903855,
+      "loss": 0.644,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564197594823263,
+      "loss": 0.8943,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563368341415537,
+      "loss": 0.0568,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562538335817364,
+      "loss": 0.5256,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561707578165555,
+      "loss": 0.6365,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560876068597046,
+      "loss": 0.6284,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560043807248896,
+      "loss": 0.4534,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955921079425829,
+      "loss": 0.4451,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558377029762532,
+      "loss": 0.6152,
+      "step": 1283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557542513899058,
+      "loss": 0.5315,
+      "step": 1284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556707246805421,
+      "loss": 0.6851,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00095558712286193,
+      "loss": 0.4797,
+      "step": 1286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555034459478496,
+      "loss": 0.5809,
+      "step": 1287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554196939520938,
+      "loss": 0.5868,
+      "step": 1288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553358668884678,
+      "loss": 0.6069,
+      "step": 1289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552519647707885,
+      "loss": 0.4987,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955167987612886,
+      "loss": 0.6744,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550839354286024,
+      "loss": 0.5667,
+      "step": 1292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549998082317921,
+      "loss": 0.0564,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549156060363222,
+      "loss": 0.5596,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548313288560719,
+      "loss": 0.5593,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009547469767049327,
+      "loss": 0.5471,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546625495968085,
+      "loss": 0.6033,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545780475456159,
+      "loss": 0.4569,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954493470565283,
+      "loss": 0.4398,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544088186697515,
+      "loss": 0.7947,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009543240918729743,
+      "loss": 0.6957,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542392901889174,
+      "loss": 0.5514,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541544136315585,
+      "loss": 0.5723,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540694622148884,
+      "loss": 0.531,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539844359529095,
+      "loss": 0.5173,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538993348596372,
+      "loss": 0.599,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538141589490987,
+      "loss": 0.4545,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537289082353336,
+      "loss": 0.6393,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536435827323942,
+      "loss": 0.6761,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535581824543449,
+      "loss": 0.5511,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009534727074152625,
+      "loss": 0.5288,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533871576292357,
+      "loss": 0.5508,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953301533110366,
+      "loss": 0.7521,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532158338727675,
+      "loss": 0.0553,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009531300599305656,
+      "loss": 0.5109,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953044211297899,
+      "loss": 0.6849,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952958287988918,
+      "loss": 0.5917,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528722900177859,
+      "loss": 0.5631,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527862173986779,
+      "loss": 0.4768,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527000701457812,
+      "loss": 0.4818,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526138482732961,
+      "loss": 0.6005,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525275517954343,
+      "loss": 0.554,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524411807264206,
+      "loss": 0.442,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523547350804916,
+      "loss": 0.5742,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522682148718962,
+      "loss": 0.5659,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952181620114896,
+      "loss": 0.6058,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520949508237647,
+      "loss": 0.7731,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520082070127877,
+      "loss": 0.5463,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519213886962635,
+      "loss": 0.6114,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009518344958885026,
+      "loss": 0.6456,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517475286038277,
+      "loss": 0.561,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516604868565737,
+      "loss": 0.5096,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515733706610878,
+      "loss": 0.7144,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514861800317301,
+      "loss": 0.6564,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513989149828718,
+      "loss": 0.4612,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513115755288973,
+      "loss": 0.7357,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951224161684203,
+      "loss": 0.538,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511366734631974,
+      "loss": 0.6615,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510491108803014,
+      "loss": 0.5573,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009509614739499482,
+      "loss": 0.5209,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508737626865832,
+      "loss": 0.5623,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009507859771046641,
+      "loss": 0.4768,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506981172186606,
+      "loss": 0.6641,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950610183043055,
+      "loss": 0.6318,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505221745923415,
+      "loss": 0.6152,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950434091881027,
+      "loss": 0.5286,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503459349236303,
+      "loss": 0.5483,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502577037346824,
+      "loss": 0.5126,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501693983287269,
+      "loss": 0.5923,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950081018720319,
+      "loss": 0.4856,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499925649240268,
+      "loss": 0.557,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499040369544303,
+      "loss": 0.5096,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498154348261216,
+      "loss": 0.7493,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497267585537055,
+      "loss": 0.53,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496380081517985,
+      "loss": 0.467,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495491836350294,
+      "loss": 0.54,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494602850180397,
+      "loss": 0.4635,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493713123154826,
+      "loss": 0.4496,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492822655420235,
+      "loss": 0.7744,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491931447123403,
+      "loss": 0.5144,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491039498411231,
+      "loss": 0.6035,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000949014680943074,
+      "loss": 0.5271,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009489253380329075,
+      "loss": 0.5262,
+      "step": 1363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488359211253501,
+      "loss": 0.6656,
+      "step": 1364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487464302351405,
+      "loss": 0.5035,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486568653770298,
+      "loss": 0.5352,
+      "step": 1366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485672265657812,
+      "loss": 0.6743,
+      "step": 1367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00094847751381617,
+      "loss": 0.5503,
+      "step": 1368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483877271429839,
+      "loss": 0.7668,
+      "step": 1369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482978665610227,
+      "loss": 0.506,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000948207932085098,
+      "loss": 0.4181,
+      "step": 1371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009481179237300339,
+      "loss": 0.6068,
+      "step": 1372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480278415106671,
+      "loss": 0.4315,
+      "step": 1373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009479376854418458,
+      "loss": 0.6323,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478474555384305,
+      "loss": 0.8311,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009477571518152943,
+      "loss": 0.536,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947666774287322,
+      "loss": 0.5979,
+      "step": 1377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475763229694105,
+      "loss": 0.6562,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474857978764697,
+      "loss": 0.4803,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473951990234205,
+      "loss": 0.6525,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473045264251969,
+      "loss": 0.6617,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009472137800967444,
+      "loss": 0.4751,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471229600530209,
+      "loss": 0.0549,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470320663089968,
+      "loss": 0.4799,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469410988796541,
+      "loss": 0.7004,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000946850057779987,
+      "loss": 0.5112,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009467589430250024,
+      "loss": 0.6147,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466677546297186,
+      "loss": 0.4591,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465764926091666,
+      "loss": 0.6921,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464851569783894,
+      "loss": 0.8169,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000946393747752442,
+      "loss": 0.6029,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463022649463914,
+      "loss": 0.5906,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462107085753171,
+      "loss": 0.5653,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461190786543103,
+      "loss": 0.6768,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460273751984749,
+      "loss": 0.536,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009459355982229266,
+      "loss": 0.4558,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945843747742793,
+      "loss": 0.6531,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457518237732139,
+      "loss": 0.5349,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456598263293417,
+      "loss": 0.6594,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009455677554263403,
+      "loss": 0.4923,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454756110793861,
+      "loss": 0.5155,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009453833933036674,
+      "loss": 0.4896,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452911021143847,
+      "loss": 0.7019,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451987375267505,
+      "loss": 0.0541,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451062995559894,
+      "loss": 0.5505,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450137882173384,
+      "loss": 0.4766,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009449212035260462,
+      "loss": 0.5135,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448285454973737,
+      "loss": 0.51,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000944735814146594,
+      "loss": 0.4595,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009446430094889922,
+      "loss": 0.6337,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445501315398656,
+      "loss": 0.5343,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444571803145234,
+      "loss": 0.4307,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443641558282871,
+      "loss": 0.4332,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442710580964897,
+      "loss": 0.5842,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441778871344772,
+      "loss": 0.5557,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009440846429576071,
+      "loss": 0.8113,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439913255812488,
+      "loss": 0.4657,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438979350207844,
+      "loss": 0.6071,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438044712916073,
+      "loss": 0.5343,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009437109344091235,
+      "loss": 0.5481,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436173243887509,
+      "loss": 0.5219,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435236412459194,
+      "loss": 0.4915,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009434298849960712,
+      "loss": 0.0536,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094333605565466,
+      "loss": 0.6768,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009432421532371522,
+      "loss": 0.4509,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000943148177759026,
+      "loss": 0.5349,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430541292357713,
+      "loss": 0.4517,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009429600076828906,
+      "loss": 0.5247,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000942865813115898,
+      "loss": 0.4797,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427715455503199,
+      "loss": 0.5653,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426772050016946,
+      "loss": 0.5389,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425827914855725,
+      "loss": 0.7224,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424883050175159,
+      "loss": 0.5907,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423937456130994,
+      "loss": 0.6455,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009422991132879093,
+      "loss": 0.0531,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000942204408057544,
+      "loss": 0.5776,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421096299376143,
+      "loss": 0.5348,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009420147789437423,
+      "loss": 0.6985,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419198550915628,
+      "loss": 0.4391,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418248583967222,
+      "loss": 0.4604,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009417297888748788,
+      "loss": 0.5219,
+      "step": 1441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416346465417036,
+      "loss": 0.5972,
+      "step": 1442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009415394314128786,
+      "loss": 0.4495,
+      "step": 1443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941444143504099,
+      "loss": 0.5081,
+      "step": 1444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413487828310706,
+      "loss": 0.6532,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009412533494095124,
+      "loss": 0.5077,
+      "step": 1446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411578432551546,
+      "loss": 0.5778,
+      "step": 1447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094106226438374,
+      "loss": 0.6997,
+      "step": 1448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009409666128110231,
+      "loss": 0.718,
+      "step": 1449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408708885527699,
+      "loss": 0.5255,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009407750916247592,
+      "loss": 0.5085,
+      "step": 1451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406792220427814,
+      "loss": 0.5294,
+      "step": 1452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009405832798226391,
+      "loss": 0.6182,
+      "step": 1453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404872649801462,
+      "loss": 0.0527,
+      "step": 1454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009403911775311294,
+      "loss": 0.6698,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009402950174914269,
+      "loss": 0.6017,
+      "step": 1456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401987848768889,
+      "loss": 0.6066,
+      "step": 1457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401024797033778,
+      "loss": 0.7206,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009400061019867678,
+      "loss": 0.6249,
+      "step": 1459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000939909651742945,
+      "loss": 0.6366,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398131289878076,
+      "loss": 0.6927,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397165337372653,
+      "loss": 0.5435,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396198660072406,
+      "loss": 0.7397,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009395231258136673,
+      "loss": 0.5975,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394263131724911,
+      "loss": 0.6458,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393294280996702,
+      "loss": 0.6482,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009392324706111741,
+      "loss": 0.6925,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391354407229847,
+      "loss": 0.6848,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390383384510956,
+      "loss": 0.4257,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389411638115124,
+      "loss": 0.8191,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388439168202525,
+      "loss": 0.5589,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387465974933454,
+      "loss": 0.0519,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009386492058468327,
+      "loss": 0.5511,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385517418967673,
+      "loss": 0.5759,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384542056592147,
+      "loss": 0.5172,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009383565971502519,
+      "loss": 0.6306,
+      "step": 1476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382589163859681,
+      "loss": 0.5765,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009381611633824639,
+      "loss": 0.4484,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380633381558525,
+      "loss": 0.6602,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379654407222584,
+      "loss": 0.5265,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378674710978184,
+      "loss": 0.5802,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377694292986811,
+      "loss": 0.554,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009376713153410068,
+      "loss": 0.5575,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375731292409679,
+      "loss": 0.4939,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374748710147487,
+      "loss": 0.5962,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373765406785453,
+      "loss": 0.4119,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009372781382485657,
+      "loss": 0.7457,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371796637410299,
+      "loss": 0.6031,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009370811171721696,
+      "loss": 0.5575,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369824985582283,
+      "loss": 0.4395,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368838079154618,
+      "loss": 0.4852,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009367850452601372,
+      "loss": 0.5291,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366862106085341,
+      "loss": 0.595,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365873039769435,
+      "loss": 0.4481,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364883253816683,
+      "loss": 0.551,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363892748390235,
+      "loss": 0.6022,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000936290152365336,
+      "loss": 0.6709,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361909579769439,
+      "loss": 0.5883,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360916916901981,
+      "loss": 0.5326,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009359923535214608,
+      "loss": 0.4615,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358929434871057,
+      "loss": 0.5529,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009357934616035193,
+      "loss": 0.6987,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356939078870994,
+      "loss": 0.7266,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355942823542554,
+      "loss": 0.5562,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000935494585021409,
+      "loss": 0.6453,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353948159049933,
+      "loss": 0.5494,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352949750214537,
+      "loss": 0.5287,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351950623872472,
+      "loss": 0.5111,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350950780188425,
+      "loss": 0.4427,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009349950219327205,
+      "loss": 0.5168,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009348948941453734,
+      "loss": 0.6724,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347946946733055,
+      "loss": 0.5542,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000934694423533033,
+      "loss": 0.6335,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000934594080741084,
+      "loss": 0.5789,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009344936663139978,
+      "loss": 0.5348,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343931802683265,
+      "loss": 0.4823,
+      "step": 1516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342926226206329,
+      "loss": 0.6554,
+      "step": 1517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009341919933874924,
+      "loss": 0.5302,
+      "step": 1518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340912925854918,
+      "loss": 0.5988,
+      "step": 1519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339905202312302,
+      "loss": 0.5604,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338896763413179,
+      "loss": 0.6373,
+      "step": 1521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337887609323772,
+      "loss": 0.579,
+      "step": 1522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009336877740210423,
+      "loss": 0.6472,
+      "step": 1523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933586715623959,
+      "loss": 0.5867,
+      "step": 1524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334855857577851,
+      "loss": 0.479,
+      "step": 1525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333843844391898,
+      "loss": 0.9474,
+      "step": 1526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332831116848548,
+      "loss": 0.564,
+      "step": 1527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009331817675114727,
+      "loss": 0.6307,
+      "step": 1528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330803519357486,
+      "loss": 0.545,
+      "step": 1529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932978864974399,
+      "loss": 0.4666,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932877306644152,
+      "loss": 0.5873,
+      "step": 1531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327756769617479,
+      "loss": 0.4772,
+      "step": 1532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326739759439387,
+      "loss": 0.6613,
+      "step": 1533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009325722036074875,
+      "loss": 0.5155,
+      "step": 1534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324703599691702,
+      "loss": 0.5634,
+      "step": 1535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009323684450457737,
+      "loss": 0.562,
+      "step": 1536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322664588540968,
+      "loss": 0.7212,
+      "step": 1537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009321644014109501,
+      "loss": 0.5271,
+      "step": 1538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009320622727331563,
+      "loss": 0.5647,
+      "step": 1539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931960072837549,
+      "loss": 0.5182,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009318578017409744,
+      "loss": 0.6093,
+      "step": 1541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317554594602901,
+      "loss": 0.536,
+      "step": 1542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009316530460123651,
+      "loss": 0.5763,
+      "step": 1543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315505614140808,
+      "loss": 0.5106,
+      "step": 1544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009314480056823298,
+      "loss": 0.6124,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009313453788340164,
+      "loss": 0.5051,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312426808860573,
+      "loss": 0.4912,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311399118553798,
+      "loss": 0.5392,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009310370717589242,
+      "loss": 0.8046,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309341606136414,
+      "loss": 0.6233,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308311784364949,
+      "loss": 0.5648,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000930728125244459,
+      "loss": 0.6122,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306250010545208,
+      "loss": 0.5787,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009305218058836777,
+      "loss": 0.4167,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304185397489403,
+      "loss": 0.4808,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009303152026673298,
+      "loss": 0.4683,
+      "step": 1556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009302117946558796,
+      "loss": 0.4865,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301083157316349,
+      "loss": 0.5798,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000930004765911652,
+      "loss": 0.62,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299011452129995,
+      "loss": 0.67,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297974536527573,
+      "loss": 0.53,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296936912480172,
+      "loss": 0.6058,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009295898580158826,
+      "loss": 0.4821,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294859539734683,
+      "loss": 0.5001,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009293819791379016,
+      "loss": 0.6562,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009292779335263204,
+      "loss": 0.4928,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000929173817155875,
+      "loss": 0.5839,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009290696300437272,
+      "loss": 0.5859,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289653722070501,
+      "loss": 0.6339,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009288610436630289,
+      "loss": 0.5673,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287566444288606,
+      "loss": 0.5939,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009286521745217534,
+      "loss": 0.4991,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928547633958927,
+      "loss": 0.4623,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284430227576135,
+      "loss": 0.5845,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009283383409350559,
+      "loss": 0.6709,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282335885085094,
+      "loss": 0.5274,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009281287654952403,
+      "loss": 0.8302,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280238719125269,
+      "loss": 0.5778,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279189077776593,
+      "loss": 0.5201,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278138731079386,
+      "loss": 0.6333,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277087679206783,
+      "loss": 0.4254,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009276035922332027,
+      "loss": 0.6949,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009274983460628485,
+      "loss": 0.5187,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273930294269635,
+      "loss": 0.7378,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009272876423429074,
+      "loss": 0.5288,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009271821848280513,
+      "loss": 0.5054,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270766568997779,
+      "loss": 0.5936,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269710585754818,
+      "loss": 0.662,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268653898725689,
+      "loss": 0.5411,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009267596508084568,
+      "loss": 0.484,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009266538414005749,
+      "loss": 0.5737,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926547961666364,
+      "loss": 0.6486,
+      "step": 1592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009264420116232761,
+      "loss": 0.6699,
+      "step": 1593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263359912887755,
+      "loss": 0.7415,
+      "step": 1594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009262299006803378,
+      "loss": 0.5831,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009261237398154501,
+      "loss": 0.4941,
+      "step": 1596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009260175087116111,
+      "loss": 0.4736,
+      "step": 1597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259112073863312,
+      "loss": 0.6881,
+      "step": 1598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925804835857132,
+      "loss": 0.5234,
+      "step": 1599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009256983941415475,
+      "loss": 0.4498,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009255918822571223,
+      "loss": 0.5929,
+      "step": 1601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009254853002214132,
+      "loss": 0.5177,
+      "step": 1602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253786480519881,
+      "loss": 0.576,
+      "step": 1603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009252719257664271,
+      "loss": 0.6047,
+      "step": 1604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251651333823213,
+      "loss": 0.054,
+      "step": 1605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009250582709172735,
+      "loss": 0.5662,
+      "step": 1606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009249513383888981,
+      "loss": 0.5204,
+      "step": 1607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000924844335814821,
+      "loss": 0.631,
+      "step": 1608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247372632126798,
+      "loss": 0.7469,
+      "step": 1609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246301206001233,
+      "loss": 0.5701,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009245229079948123,
+      "loss": 0.714,
+      "step": 1611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244156254144186,
+      "loss": 0.5836,
+      "step": 1612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243082728766263,
+      "loss": 0.575,
+      "step": 1613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009242008503991302,
+      "loss": 0.6564,
+      "step": 1614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009240933579996372,
+      "loss": 0.5175,
+      "step": 1615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239857956958653,
+      "loss": 0.4225,
+      "step": 1616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009238781635055443,
+      "loss": 0.7589,
+      "step": 1617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237704614464157,
+      "loss": 0.0519,
+      "step": 1618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236626895362319,
+      "loss": 0.4551,
+      "step": 1619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235548477927575,
+      "loss": 0.5857,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009234469362337681,
+      "loss": 0.4445,
+      "step": 1621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000923338954877051,
+      "loss": 0.5363,
+      "step": 1622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232309037404054,
+      "loss": 0.4723,
+      "step": 1623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009231227828416411,
+      "loss": 0.5148,
+      "step": 1624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009230145921985802,
+      "loss": 0.4255,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009229063318290558,
+      "loss": 0.4758,
+      "step": 1626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922798001750913,
+      "loss": 0.0527,
+      "step": 1627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922689601982008,
+      "loss": 0.5287,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225811325402085,
+      "loss": 0.5614,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224725934433937,
+      "loss": 0.0524,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009223639847094544,
+      "loss": 0.5261,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222553063562931,
+      "loss": 0.4844,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009221465584018231,
+      "loss": 0.662,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220377408639698,
+      "loss": 0.4705,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009219288537606698,
+      "loss": 0.8004,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921819897109871,
+      "loss": 0.4448,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217108709295334,
+      "loss": 0.5111,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009216017752376278,
+      "loss": 0.0519,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009214926100521367,
+      "loss": 0.5522,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921383375391054,
+      "loss": 0.437,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921274071272385,
+      "loss": 0.5017,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921164697714147,
+      "loss": 0.6349,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009210552547343678,
+      "loss": 0.6415,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209457423510874,
+      "loss": 0.6401,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009208361605823567,
+      "loss": 0.0519,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009207265094462387,
+      "loss": 0.4943,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206167889608072,
+      "loss": 0.7776,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009205069991441479,
+      "loss": 0.051,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203971400143574,
+      "loss": 0.5321,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202872115895442,
+      "loss": 0.6797,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201772138878281,
+      "loss": 0.5428,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009200671469273402,
+      "loss": 0.6003,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919957010726223,
+      "loss": 0.5367,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198468053026307,
+      "loss": 0.5887,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009197365306747285,
+      "loss": 0.5653,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009196261868606935,
+      "loss": 0.5519,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009195157738787136,
+      "loss": 0.5601,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009194052917469886,
+      "loss": 0.571,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009192947404837295,
+      "loss": 0.5275,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191841201071588,
+      "loss": 0.5869,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009190734306355102,
+      "loss": 0.6268,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009189626720870288,
+      "loss": 0.5939,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188518444799713,
+      "loss": 0.577,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009187409478326059,
+      "loss": 0.4994,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009186299821632116,
+      "loss": 0.6838,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185189474900793,
+      "loss": 0.4855,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009184078438315111,
+      "loss": 0.5618,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009182966712058203,
+      "loss": 0.5278,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181854296313319,
+      "loss": 0.421,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009180741191263822,
+      "loss": 0.5415,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179627397093185,
+      "loss": 0.6277,
+      "step": 1671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178512913984997,
+      "loss": 0.7073,
+      "step": 1672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009177397742122966,
+      "loss": 0.4784,
+      "step": 1673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009176281881690902,
+      "loss": 0.5017,
+      "step": 1674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175165332872739,
+      "loss": 0.4269,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009174048095852517,
+      "loss": 0.5992,
+      "step": 1676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009172930170814397,
+      "loss": 0.5933,
+      "step": 1677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009171811557942644,
+      "loss": 0.5941,
+      "step": 1678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009170692257421647,
+      "loss": 0.5837,
+      "step": 1679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169572269435897,
+      "loss": 0.4717,
+      "step": 1680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916845159417001,
+      "loss": 0.4526,
+      "step": 1681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009167330231808706,
+      "loss": 0.5096,
+      "step": 1682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009166208182536822,
+      "loss": 0.4994,
+      "step": 1683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009165085446539307,
+      "loss": 0.6576,
+      "step": 1684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009163962024001227,
+      "loss": 0.499,
+      "step": 1685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009162837915107758,
+      "loss": 0.6899,
+      "step": 1686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009161713120044186,
+      "loss": 0.0517,
+      "step": 1687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009160587638995917,
+      "loss": 0.6239,
+      "step": 1688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159461472148465,
+      "loss": 0.5941,
+      "step": 1689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009158334619687459,
+      "loss": 0.5037,
+      "step": 1690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009157207081798642,
+      "loss": 0.5887,
+      "step": 1691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156078858667866,
+      "loss": 0.6207,
+      "step": 1692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009154949950481102,
+      "loss": 0.4558,
+      "step": 1693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153820357424427,
+      "loss": 0.4866,
+      "step": 1694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009152690079684036,
+      "loss": 0.5071,
+      "step": 1695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009151559117446236,
+      "loss": 0.5449,
+      "step": 1696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150427470897445,
+      "loss": 0.582,
+      "step": 1697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009149295140224194,
+      "loss": 0.6422,
+      "step": 1698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148162125613131,
+      "loss": 0.5394,
+      "step": 1699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914702842725101,
+      "loss": 0.725,
+      "step": 1700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009145894045324701,
+      "loss": 0.5503,
+      "step": 1701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914475898002119,
+      "loss": 0.4528,
+      "step": 1702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143623231527571,
+      "loss": 0.6364,
+      "step": 1703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914248680003105,
+      "loss": 0.5347,
+      "step": 1704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141349685718949,
+      "loss": 0.5708,
+      "step": 1705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009140211888778704,
+      "loss": 0.5118,
+      "step": 1706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139073409397856,
+      "loss": 0.4308,
+      "step": 1707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137934247764065,
+      "loss": 0.5687,
+      "step": 1708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136794404065102,
+      "loss": 0.7124,
+      "step": 1709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000913565387848885,
+      "loss": 0.5923,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009134512671223304,
+      "loss": 0.6205,
+      "step": 1711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133370782456572,
+      "loss": 0.6698,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009132228212376874,
+      "loss": 0.5165,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131084961172544,
+      "loss": 0.6536,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009129941029032025,
+      "loss": 0.5057,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009128796416143873,
+      "loss": 0.6499,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009127651122696759,
+      "loss": 0.5008,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009126505148879465,
+      "loss": 0.5076,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125358494880882,
+      "loss": 0.5183,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009124211160890016,
+      "loss": 0.4377,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009123063147095988,
+      "loss": 0.6486,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121914453688024,
+      "loss": 0.5352,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009120765080855468,
+      "loss": 0.5616,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009119615028787771,
+      "loss": 0.6666,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118464297674503,
+      "loss": 0.4502,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009117312887705338,
+      "loss": 0.4817,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116160799070065,
+      "loss": 0.6177,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911500803195859,
+      "loss": 0.498,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113854586560921,
+      "loss": 0.6824,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112700463067187,
+      "loss": 0.5056,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009111545661667623,
+      "loss": 0.4966,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110390182552578,
+      "loss": 0.6578,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009109234025912511,
+      "loss": 0.6534,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009108077191937997,
+      "loss": 0.4249,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009106919680819716,
+      "loss": 0.5173,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105761492748466,
+      "loss": 0.4297,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104602627915151,
+      "loss": 0.5226,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009103443086510792,
+      "loss": 0.6411,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009102282868726517,
+      "loss": 0.5381,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009101121974753568,
+      "loss": 0.5352,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00090999604047833,
+      "loss": 0.5514,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009098798159007174,
+      "loss": 0.5748,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009097635237616767,
+      "loss": 0.5719,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009096471640803765,
+      "loss": 0.5728,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095307368759968,
+      "loss": 0.379,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009094142421677285,
+      "loss": 0.6588,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009092976799747736,
+      "loss": 0.6067,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009091810503163456,
+      "loss": 0.7541,
+      "step": 1748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009090643532116683,
+      "loss": 0.4622,
+      "step": 1749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009089475886799776,
+      "loss": 0.5068,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00090883075674052,
+      "loss": 0.5424,
+      "step": 1751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908713857412553,
+      "loss": 0.0513,
+      "step": 1752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009085968907153455,
+      "loss": 0.4042,
+      "step": 1753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009084798566681775,
+      "loss": 0.4408,
+      "step": 1754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009083627552903397,
+      "loss": 0.5451,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082455866011345,
+      "loss": 0.5773,
+      "step": 1756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009081283506198748,
+      "loss": 0.5063,
+      "step": 1757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009080110473658851,
+      "loss": 0.5839,
+      "step": 1758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009078936768585007,
+      "loss": 0.4617,
+      "step": 1759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077762391170681,
+      "loss": 0.5026,
+      "step": 1760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009076587341609446,
+      "loss": 0.4744,
+      "step": 1761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907541162009499,
+      "loss": 0.5879,
+      "step": 1762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907423522682111,
+      "loss": 0.5271,
+      "step": 1763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009073058161981713,
+      "loss": 0.434,
+      "step": 1764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009071880425770818,
+      "loss": 0.5631,
+      "step": 1765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009070702018382553,
+      "loss": 0.5903,
+      "step": 1766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009069522940011158,
+      "loss": 0.5066,
+      "step": 1767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068343190850983,
+      "loss": 0.7264,
+      "step": 1768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009067162771096489,
+      "loss": 0.3712,
+      "step": 1769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065981680942248,
+      "loss": 0.5786,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000906479992058294,
+      "loss": 0.0522,
+      "step": 1771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009063617490213358,
+      "loss": 0.6545,
+      "step": 1772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009062434390028406,
+      "loss": 0.6016,
+      "step": 1773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061250620223096,
+      "loss": 0.4541,
+      "step": 1774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009060066180992553,
+      "loss": 0.5338,
+      "step": 1775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009058881072532007,
+      "loss": 0.4325,
+      "step": 1776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057695295036805,
+      "loss": 0.4965,
+      "step": 1777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009056508848702402,
+      "loss": 0.5067,
+      "step": 1778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055321733724361,
+      "loss": 0.4049,
+      "step": 1779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009054133950298358,
+      "loss": 0.6248,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052945498620177,
+      "loss": 0.0521,
+      "step": 1781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009051756378885714,
+      "loss": 0.4999,
+      "step": 1782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009050566591290974,
+      "loss": 0.6401,
+      "step": 1783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009049376136032073,
+      "loss": 0.0518,
+      "step": 1784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048185013305236,
+      "loss": 0.5146,
+      "step": 1785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046993223306799,
+      "loss": 0.4925,
+      "step": 1786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009045800766233205,
+      "loss": 0.6403,
+      "step": 1787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009044607642281013,
+      "loss": 0.5576,
+      "step": 1788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009043413851646887,
+      "loss": 0.4806,
+      "step": 1789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00090422193945276,
+      "loss": 0.5482,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009041024271120041,
+      "loss": 0.0515,
+      "step": 1791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009039828481621202,
+      "loss": 0.5978,
+      "step": 1792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009038632026228189,
+      "loss": 0.5646,
+      "step": 1793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009037434905138217,
+      "loss": 0.6462,
+      "step": 1794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009036237118548609,
+      "loss": 0.5165,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009035038666656798,
+      "loss": 0.5118,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000903383954966033,
+      "loss": 0.4723,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009032639767756858,
+      "loss": 0.6158,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009031439321144142,
+      "loss": 0.5724,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009030238210020058,
+      "loss": 0.5025,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029036434582585,
+      "loss": 0.5083,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009027833995029816,
+      "loss": 0.4801,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009026630891559951,
+      "loss": 0.5807,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025427124371301,
+      "loss": 0.5375,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009024222693662286,
+      "loss": 0.4376,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009023017599631433,
+      "loss": 0.5475,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021811842477383,
+      "loss": 0.5409,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009020605422398883,
+      "loss": 0.4633,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009019398339594789,
+      "loss": 0.4664,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018190594264068,
+      "loss": 0.0506,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009016982186605796,
+      "loss": 0.5272,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015773116819155,
+      "loss": 0.4983,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009014563385103441,
+      "loss": 0.5542,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009013352991658057,
+      "loss": 0.4508,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012141936682514,
+      "loss": 0.4967,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009010930220376435,
+      "loss": 0.0509,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009717842939545,
+      "loss": 0.5548,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008504804571689,
+      "loss": 0.5282,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009007291105472811,
+      "loss": 0.051,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009006076745842969,
+      "loss": 0.5782,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009004861725882329,
+      "loss": 0.4257,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003646045791165,
+      "loss": 0.665,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009002429705769859,
+      "loss": 0.5685,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001212706018907,
+      "loss": 0.5051,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008999995046738904,
+      "loss": 0.5714,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008998776728130567,
+      "loss": 0.6807,
+      "step": 1826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008997557750394707,
+      "loss": 0.6306,
+      "step": 1827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008996338113732256,
+      "loss": 0.5573,
+      "step": 1828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008995117818344248,
+      "loss": 0.7041,
+      "step": 1829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993896864431826,
+      "loss": 0.6202,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008992675252196242,
+      "loss": 0.8269,
+      "step": 1831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899145298183886,
+      "loss": 0.6689,
+      "step": 1832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990230053561147,
+      "loss": 0.6364,
+      "step": 1833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008989006467564683,
+      "loss": 0.426,
+      "step": 1834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987782224051153,
+      "loss": 0.5311,
+      "step": 1835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008986557323222353,
+      "loss": 0.6853,
+      "step": 1836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008985331765280185,
+      "loss": 0.4771,
+      "step": 1837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008984105550426662,
+      "loss": 0.5121,
+      "step": 1838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008982878678863903,
+      "loss": 0.5396,
+      "step": 1839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008981651150794136,
+      "loss": 0.4442,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008980422966419697,
+      "loss": 0.4344,
+      "step": 1841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008979194125943031,
+      "loss": 0.4918,
+      "step": 1842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897796462956669,
+      "loss": 0.5122,
+      "step": 1843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008976734477493335,
+      "loss": 0.629,
+      "step": 1844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008975503669925735,
+      "loss": 0.5018,
+      "step": 1845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008974272207066767,
+      "loss": 0.424,
+      "step": 1846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008973040089119415,
+      "loss": 0.5634,
+      "step": 1847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971807316286772,
+      "loss": 0.0505,
+      "step": 1848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897057388877204,
+      "loss": 0.4725,
+      "step": 1849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008969339806778527,
+      "loss": 0.6495,
+      "step": 1850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008968105070509649,
+      "loss": 0.4594,
+      "step": 1851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008966869680168929,
+      "loss": 0.4789,
+      "step": 1852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008965633635960003,
+      "loss": 0.6367,
+      "step": 1853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008964396938086609,
+      "loss": 0.5248,
+      "step": 1854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008963159586752594,
+      "loss": 0.4185,
+      "step": 1855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961921582161914,
+      "loss": 0.5452,
+      "step": 1856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008960682924518633,
+      "loss": 0.639,
+      "step": 1857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959443614026919,
+      "loss": 0.5219,
+      "step": 1858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008958203650891055,
+      "loss": 0.5242,
+      "step": 1859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008956963035315423,
+      "loss": 0.6188,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955721767504519,
+      "loss": 0.588,
+      "step": 1861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008954479847662942,
+      "loss": 0.5305,
+      "step": 1862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008953237275995401,
+      "loss": 0.5801,
+      "step": 1863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008951994052706714,
+      "loss": 0.5749,
+      "step": 1864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008950750178001801,
+      "loss": 0.3922,
+      "step": 1865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949505652085694,
+      "loss": 0.5933,
+      "step": 1866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008948260475163533,
+      "loss": 0.5327,
+      "step": 1867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947014647440561,
+      "loss": 0.614,
+      "step": 1868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008945768169122132,
+      "loss": 0.5104,
+      "step": 1869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008944521040413705,
+      "loss": 0.6284,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008943273261520848,
+      "loss": 0.5751,
+      "step": 1871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008942024832649234,
+      "loss": 0.4937,
+      "step": 1872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008940775754004644,
+      "loss": 0.5573,
+      "step": 1873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939526025792968,
+      "loss": 0.4355,
+      "step": 1874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008938275648220202,
+      "loss": 0.4141,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008937024621492448,
+      "loss": 0.6173,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008935772945815914,
+      "loss": 0.5179,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008934520621396919,
+      "loss": 0.6475,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933267648441885,
+      "loss": 0.5135,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008932014027157343,
+      "loss": 0.5039,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008930759757749929,
+      "loss": 0.5557,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008929504840426388,
+      "loss": 0.5558,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008928249275393571,
+      "loss": 0.4713,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008926993062858437,
+      "loss": 0.4325,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008925736203028049,
+      "loss": 0.6329,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008924478696109577,
+      "loss": 0.4675,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923220542310301,
+      "loss": 0.5529,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008921961741837603,
+      "loss": 0.4991,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008920702294898976,
+      "loss": 0.5214,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008919442201702016,
+      "loss": 0.5637,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008918181462454429,
+      "loss": 0.5676,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008916920077364024,
+      "loss": 0.5408,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008915658046638718,
+      "loss": 0.5483,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008914395370486534,
+      "loss": 0.4774,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913132049115603,
+      "loss": 0.8521,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891186808273416,
+      "loss": 0.5951,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910603471550549,
+      "loss": 0.4761,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008909338215773218,
+      "loss": 0.0518,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008908072315610723,
+      "loss": 0.5693,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008906805771271723,
+      "loss": 0.0518,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008905538582964985,
+      "loss": 0.5476,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008904270750899386,
+      "loss": 0.6503,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008903002275283902,
+      "loss": 0.5498,
+      "step": 1903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008901733156327623,
+      "loss": 0.4637,
+      "step": 1904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008900463394239738,
+      "loss": 0.548,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008899192989229543,
+      "loss": 0.5278,
+      "step": 1906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008897921941506444,
+      "loss": 0.576,
+      "step": 1907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008896650251279952,
+      "loss": 0.5311,
+      "step": 1908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895377918759679,
+      "loss": 0.4908,
+      "step": 1909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889410494415535,
+      "loss": 0.5198,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892831327676792,
+      "loss": 0.45,
+      "step": 1911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008891557069533936,
+      "loss": 0.6166,
+      "step": 1912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008890282169936821,
+      "loss": 0.7622,
+      "step": 1913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008889006629095593,
+      "loss": 0.4478,
+      "step": 1914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887730447220503,
+      "loss": 0.4731,
+      "step": 1915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008886453624521905,
+      "loss": 0.61,
+      "step": 1916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008885176161210262,
+      "loss": 0.5439,
+      "step": 1917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000888389805749614,
+      "loss": 0.5286,
+      "step": 1918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882619313590213,
+      "loss": 0.5873,
+      "step": 1919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008881339929703258,
+      "loss": 0.4283,
+      "step": 1920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880059906046159,
+      "loss": 0.6442,
+      "step": 1921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008878779242829907,
+      "loss": 0.5176,
+      "step": 1922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008877497940265594,
+      "loss": 0.7172,
+      "step": 1923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008876215998564423,
+      "loss": 0.5916,
+      "step": 1924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874933417937697,
+      "loss": 0.5649,
+      "step": 1925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008873650198596828,
+      "loss": 0.3606,
+      "step": 1926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887236634075333,
+      "loss": 0.4585,
+      "step": 1927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008871081844618827,
+      "loss": 0.5835,
+      "step": 1928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008869796710405043,
+      "loss": 0.4242,
+      "step": 1929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008868510938323811,
+      "loss": 0.4971,
+      "step": 1930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867224528587068,
+      "loss": 0.4905,
+      "step": 1931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008865937481406857,
+      "loss": 0.5599,
+      "step": 1932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000886464979699532,
+      "loss": 0.4967,
+      "step": 1933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008863361475564713,
+      "loss": 0.7499,
+      "step": 1934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008862072517327393,
+      "loss": 0.4957,
+      "step": 1935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008860782922495822,
+      "loss": 0.6858,
+      "step": 1936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859492691282565,
+      "loss": 0.5938,
+      "step": 1937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008858201823900295,
+      "loss": 0.5404,
+      "step": 1938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008856910320561788,
+      "loss": 0.6438,
+      "step": 1939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008855618181479925,
+      "loss": 0.5388,
+      "step": 1940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008854325406867695,
+      "loss": 0.4932,
+      "step": 1941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853031996938186,
+      "loss": 0.5417,
+      "step": 1942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851737951904594,
+      "loss": 0.5414,
+      "step": 1943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008850443271980222,
+      "loss": 0.4589,
+      "step": 1944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849147957378472,
+      "loss": 0.5419,
+      "step": 1945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008847852008312856,
+      "loss": 0.5219,
+      "step": 1946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008846555424996987,
+      "loss": 0.6095,
+      "step": 1947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008845258207644584,
+      "loss": 0.4997,
+      "step": 1948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000884396035646947,
+      "loss": 0.5797,
+      "step": 1949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008842661871685573,
+      "loss": 0.4948,
+      "step": 1950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008841362753506925,
+      "loss": 0.4602,
+      "step": 1951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840063002147662,
+      "loss": 0.6107,
+      "step": 1952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008838762617822029,
+      "loss": 0.605,
+      "step": 1953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008837461600744365,
+      "loss": 0.4519,
+      "step": 1954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008836159951129126,
+      "loss": 0.4156,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883485766919086,
+      "loss": 0.5273,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833554755144228,
+      "loss": 0.5308,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008832251209203994,
+      "loss": 0.448,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008830947031585022,
+      "loss": 0.5015,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008829642222502282,
+      "loss": 0.5183,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882833678217085,
+      "loss": 0.5284,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827030710805905,
+      "loss": 0.538,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882572400862273,
+      "loss": 0.3646,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882441667583671,
+      "loss": 0.496,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823108712663338,
+      "loss": 0.5344,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821800119318206,
+      "loss": 0.502,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008820490896017015,
+      "loss": 0.4606,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819181042975565,
+      "loss": 0.4769,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817870560409763,
+      "loss": 0.4505,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008816559448535619,
+      "loss": 0.5635,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008815247707569247,
+      "loss": 0.5099,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008813935337726862,
+      "loss": 0.5518,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008812622339224789,
+      "loss": 0.4662,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008811308712279448,
+      "loss": 0.5894,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809994457107371,
+      "loss": 0.5577,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008808679573925187,
+      "loss": 0.4551,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008807364062949632,
+      "loss": 0.4484,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008806047924397545,
+      "loss": 0.3859,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008804731158485868,
+      "loss": 0.5159,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803413765431647,
+      "loss": 0.4716,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802095745452031,
+      "loss": 0.0506,
+      "step": 1981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800777098764271,
+      "loss": 0.6703,
+      "step": 1982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008799457825585723,
+      "loss": 0.5804,
+      "step": 1983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008798137926133847,
+      "loss": 0.423,
+      "step": 1984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008796817400626202,
+      "loss": 0.494,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008795496249280458,
+      "loss": 0.5912,
+      "step": 1986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000879417447231438,
+      "loss": 0.5642,
+      "step": 1987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008792852069945841,
+      "loss": 0.4414,
+      "step": 1988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008791529042392813,
+      "loss": 0.5382,
+      "step": 1989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790205389873376,
+      "loss": 0.4036,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008788881112605712,
+      "loss": 0.4675,
+      "step": 1991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008787556210808101,
+      "loss": 0.5317,
+      "step": 1992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008786230684698933,
+      "loss": 0.5961,
+      "step": 1993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008784904534496694,
+      "loss": 0.5199,
+      "step": 1994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000878357776041998,
+      "loss": 0.5504,
+      "step": 1995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008782250362687483,
+      "loss": 0.5636,
+      "step": 1996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008780922341518004,
+      "loss": 0.4314,
+      "step": 1997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008779593697130441,
+      "loss": 0.4603,
+      "step": 1998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008778264429743801,
+      "loss": 0.6467,
+      "step": 1999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776934539577186,
+      "loss": 0.6802,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008775604026849808,
+      "loss": 0.5192,
+      "step": 2001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008774272891780977,
+      "loss": 0.5413,
+      "step": 2002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008772941134590108,
+      "loss": 0.5618,
+      "step": 2003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008771608755496716,
+      "loss": 0.5311,
+      "step": 2004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000877027575472042,
+      "loss": 0.5252,
+      "step": 2005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008768942132480946,
+      "loss": 0.4827,
+      "step": 2006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008767607888998111,
+      "loss": 0.5858,
+      "step": 2007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766273024491847,
+      "loss": 0.418,
+      "step": 2008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008764937539182181,
+      "loss": 0.5851,
+      "step": 2009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008763601433289243,
+      "loss": 0.6395,
+      "step": 2010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762264707033267,
+      "loss": 0.4818,
+      "step": 2011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000876092736063459,
+      "loss": 0.5975,
+      "step": 2012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875958939431365,
+      "loss": 0.4856,
+      "step": 2013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008758250808290983,
+      "loss": 0.4823,
+      "step": 2014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008756911602787235,
+      "loss": 0.6399,
+      "step": 2015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875557177802315,
+      "loss": 0.4747,
+      "step": 2016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008754231334219574,
+      "loss": 0.5199,
+      "step": 2017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008752890271597454,
+      "loss": 0.5807,
+      "step": 2018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008751548590377842,
+      "loss": 0.6212,
+      "step": 2019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008750206290781891,
+      "loss": 0.5021,
+      "step": 2020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748863373030853,
+      "loss": 0.4744,
+      "step": 2021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008747519837346085,
+      "loss": 0.6571,
+      "step": 2022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746175683949047,
+      "loss": 0.4071,
+      "step": 2023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744830913061297,
+      "loss": 0.5771,
+      "step": 2024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008743485524904496,
+      "loss": 0.5654,
+      "step": 2025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000874213951970041,
+      "loss": 0.6378,
+      "step": 2026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008740792897670901,
+      "loss": 0.5499,
+      "step": 2027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008739445659037939,
+      "loss": 0.5874,
+      "step": 2028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008738097804023591,
+      "loss": 0.5223,
+      "step": 2029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008736749332850026,
+      "loss": 0.4125,
+      "step": 2030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735400245739516,
+      "loss": 0.5302,
+      "step": 2031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008734050542914436,
+      "loss": 0.4983,
+      "step": 2032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008732700224597259,
+      "loss": 0.5698,
+      "step": 2033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873134929101056,
+      "loss": 0.5531,
+      "step": 2034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729997742377018,
+      "loss": 0.4771,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008728645578919412,
+      "loss": 0.5515,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008727292800860622,
+      "loss": 0.5117,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008725939408423629,
+      "loss": 0.538,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008724585401831516,
+      "loss": 0.6324,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008723230781307467,
+      "loss": 0.4451,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008721875547074766,
+      "loss": 0.4474,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720519699356804,
+      "loss": 0.4866,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008719163238377062,
+      "loss": 0.4707,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717806164359133,
+      "loss": 0.4506,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008716448477526706,
+      "loss": 0.4996,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008715090178103568,
+      "loss": 0.4159,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008713731266313616,
+      "loss": 0.6976,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000871237174238084,
+      "loss": 0.5499,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711011606529333,
+      "loss": 0.5569,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008709650858983289,
+      "loss": 0.475,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008708289499967005,
+      "loss": 0.5586,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008706927529704876,
+      "loss": 0.4252,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008705564948421398,
+      "loss": 0.4301,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000870420175634117,
+      "loss": 0.6125,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702837953688887,
+      "loss": 0.5972,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008701473540689351,
+      "loss": 0.5427,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008700108517567459,
+      "loss": 0.4797,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008698742884548212,
+      "loss": 0.5651,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008697376641856712,
+      "loss": 0.418,
+      "step": 2059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008696009789718156,
+      "loss": 0.6622,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008694642328357849,
+      "loss": 0.5814,
+      "step": 2061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008693274258001192,
+      "loss": 0.4753,
+      "step": 2062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008691905578873684,
+      "loss": 0.5737,
+      "step": 2063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008690536291200932,
+      "loss": 0.5271,
+      "step": 2064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689166395208636,
+      "loss": 0.4048,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00086877958911226,
+      "loss": 0.5292,
+      "step": 2066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008686424779168729,
+      "loss": 0.4491,
+      "step": 2067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008685053059573023,
+      "loss": 0.5074,
+      "step": 2068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008683680732561588,
+      "loss": 0.53,
+      "step": 2069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008682307798360628,
+      "loss": 0.4608,
+      "step": 2070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008680934257196446,
+      "loss": 0.452,
+      "step": 2071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008679560109295446,
+      "loss": 0.3823,
+      "step": 2072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008678185354884132,
+      "loss": 0.5085,
+      "step": 2073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008676809994189108,
+      "loss": 0.4778,
+      "step": 2074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008675434027437077,
+      "loss": 0.4859,
+      "step": 2075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674057454854843,
+      "loss": 0.4596,
+      "step": 2076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008672680276669312,
+      "loss": 0.4423,
+      "step": 2077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008671302493107485,
+      "loss": 0.4856,
+      "step": 2078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008669924104396465,
+      "loss": 0.5154,
+      "step": 2079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008668545110763458,
+      "loss": 0.3793,
+      "step": 2080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008667165512435762,
+      "loss": 0.433,
+      "step": 2081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665785309640784,
+      "loss": 0.3708,
+      "step": 2082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008664404502606022,
+      "loss": 0.6379,
+      "step": 2083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000866302309155908,
+      "loss": 0.4597,
+      "step": 2084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661641076727659,
+      "loss": 0.6405,
+      "step": 2085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008660258458339558,
+      "loss": 0.6937,
+      "step": 2086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008658875236622679,
+      "loss": 0.4093,
+      "step": 2087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008657491411805023,
+      "loss": 0.4944,
+      "step": 2088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008656106984114685,
+      "loss": 0.0522,
+      "step": 2089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008654721953779865,
+      "loss": 0.6057,
+      "step": 2090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008653336321028862,
+      "loss": 0.4926,
+      "step": 2091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008651950086090071,
+      "loss": 0.4908,
+      "step": 2092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865056324919199,
+      "loss": 0.48,
+      "step": 2093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649175810563214,
+      "loss": 0.4515,
+      "step": 2094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008647787770432438,
+      "loss": 0.8413,
+      "step": 2095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008646399129028455,
+      "loss": 0.609,
+      "step": 2096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008645009886580158,
+      "loss": 0.5398,
+      "step": 2097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864362004331654,
+      "loss": 0.6162,
+      "step": 2098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008642229599466691,
+      "loss": 0.5283,
+      "step": 2099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008640838555259801,
+      "loss": 0.7936,
+      "step": 2100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863944691092516,
+      "loss": 0.6002,
+      "step": 2101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008638054666692155,
+      "loss": 0.566,
+      "step": 2102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008636661822790274,
+      "loss": 0.5166,
+      "step": 2103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008635268379449101,
+      "loss": 0.5584,
+      "step": 2104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008633874336898321,
+      "loss": 0.5288,
+      "step": 2105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008632479695367719,
+      "loss": 0.4742,
+      "step": 2106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631084455087174,
+      "loss": 0.7123,
+      "step": 2107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008629688616286668,
+      "loss": 0.4697,
+      "step": 2108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862829217919628,
+      "loss": 0.6072,
+      "step": 2109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862689514404619,
+      "loss": 0.4656,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008625497511066671,
+      "loss": 0.598,
+      "step": 2111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008624099280488103,
+      "loss": 0.6152,
+      "step": 2112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008622700452540952,
+      "loss": 0.5001,
+      "step": 2113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008621301027455797,
+      "loss": 0.6161,
+      "step": 2114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008619901005463307,
+      "loss": 0.581,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008618500386794248,
+      "loss": 0.5658,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617099171679488,
+      "loss": 0.6108,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008615697360349994,
+      "loss": 0.5618,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008614294953036831,
+      "loss": 0.4473,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008612891949971158,
+      "loss": 0.5164,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008611488351384235,
+      "loss": 0.5109,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008610084157507423,
+      "loss": 0.3759,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008608679368572176,
+      "loss": 0.3928,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008607273984810051,
+      "loss": 0.5461,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008605868006452699,
+      "loss": 0.5046,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008604461433731872,
+      "loss": 0.4399,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008603054266879417,
+      "loss": 0.4211,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860164650612728,
+      "loss": 0.5768,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008600238151707509,
+      "loss": 0.5736,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008598829203852242,
+      "loss": 0.5303,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008597419662793721,
+      "loss": 0.5938,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008596009528764283,
+      "loss": 0.5507,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008594598801996367,
+      "loss": 0.5018,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593187482722502,
+      "loss": 0.7045,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008591775571175322,
+      "loss": 0.4169,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008590363067587553,
+      "loss": 0.0531,
+      "step": 2136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588949972192024,
+      "loss": 0.4917,
+      "step": 2137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008587536285221655,
+      "loss": 0.6194,
+      "step": 2138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008586122006909473,
+      "loss": 0.4506,
+      "step": 2139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000858470713748859,
+      "loss": 0.5397,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008583291677192227,
+      "loss": 0.6067,
+      "step": 2141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008581875626253696,
+      "loss": 0.4999,
+      "step": 2142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008580458984906408,
+      "loss": 0.4752,
+      "step": 2143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579041753383872,
+      "loss": 0.4746,
+      "step": 2144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008577623931919691,
+      "loss": 0.5765,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008576205520747571,
+      "loss": 0.5688,
+      "step": 2146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857478652010131,
+      "loss": 0.6179,
+      "step": 2147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008573366930214806,
+      "loss": 0.6282,
+      "step": 2148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008571946751322051,
+      "loss": 0.5896,
+      "step": 2149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857052598365714,
+      "loss": 0.4735,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008569104627454259,
+      "loss": 0.6399,
+      "step": 2151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008567682682947693,
+      "loss": 0.5465,
+      "step": 2152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008566260150371826,
+      "loss": 0.4504,
+      "step": 2153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008564837029961137,
+      "loss": 0.0519,
+      "step": 2154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008563413321950201,
+      "loss": 0.5204,
+      "step": 2155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008561989026573692,
+      "loss": 0.5585,
+      "step": 2156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000856056414406638,
+      "loss": 0.4142,
+      "step": 2157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008559138674663132,
+      "loss": 0.5157,
+      "step": 2158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008557712618598909,
+      "loss": 0.4268,
+      "step": 2159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008556285976108775,
+      "loss": 0.6051,
+      "step": 2160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008554858747427883,
+      "loss": 0.4838,
+      "step": 2161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553430932791488,
+      "loss": 0.5905,
+      "step": 2162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008552002532434942,
+      "loss": 0.578,
+      "step": 2163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008550573546593686,
+      "loss": 0.4613,
+      "step": 2164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549143975503269,
+      "loss": 0.4257,
+      "step": 2165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008547713819399325,
+      "loss": 0.6449,
+      "step": 2166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008546283078517596,
+      "loss": 0.4585,
+      "step": 2167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008544851753093909,
+      "loss": 0.4365,
+      "step": 2168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008543419843364197,
+      "loss": 0.6631,
+      "step": 2169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008541987349564479,
+      "loss": 0.6094,
+      "step": 2170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008540554271930884,
+      "loss": 0.5493,
+      "step": 2171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008539120610699622,
+      "loss": 0.5366,
+      "step": 2172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008537686366107012,
+      "loss": 0.4254,
+      "step": 2173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008536251538389461,
+      "loss": 0.5372,
+      "step": 2174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008534816127783476,
+      "loss": 0.5597,
+      "step": 2175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008533380134525659,
+      "loss": 0.6615,
+      "step": 2176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008531943558852708,
+      "loss": 0.5327,
+      "step": 2177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008530506401001417,
+      "loss": 0.5464,
+      "step": 2178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529068661208676,
+      "loss": 0.562,
+      "step": 2179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000852763033971147,
+      "loss": 0.4805,
+      "step": 2180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008526191436746881,
+      "loss": 0.4507,
+      "step": 2181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000852475195255209,
+      "loss": 0.342,
+      "step": 2182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008523311887364365,
+      "loss": 0.4727,
+      "step": 2183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000852187124142108,
+      "loss": 0.6846,
+      "step": 2184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008520430014959697,
+      "loss": 0.5382,
+      "step": 2185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008518988208217779,
+      "loss": 0.4321,
+      "step": 2186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008517545821432979,
+      "loss": 0.561,
+      "step": 2187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008516102854843053,
+      "loss": 0.5968,
+      "step": 2188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008514659308685846,
+      "loss": 0.5013,
+      "step": 2189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00085132151831993,
+      "loss": 0.489,
+      "step": 2190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008511770478621457,
+      "loss": 0.6342,
+      "step": 2191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510325195190447,
+      "loss": 0.5795,
+      "step": 2192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008508879333144502,
+      "loss": 0.5846,
+      "step": 2193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008507432892721945,
+      "loss": 0.5992,
+      "step": 2194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008505985874161199,
+      "loss": 0.5665,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504538277700777,
+      "loss": 0.4543,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000850309010357929,
+      "loss": 0.0536,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008501641352035443,
+      "loss": 0.5293,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008500192023308038,
+      "loss": 0.4493,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008498742117635972,
+      "loss": 0.4719,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008497291635258235,
+      "loss": 0.4796,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008495840576413916,
+      "loss": 0.7146,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008494388941342191,
+      "loss": 0.4771,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008492936730282342,
+      "loss": 0.4631,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008491483943473734,
+      "loss": 0.5836,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849003058115584,
+      "loss": 0.5134,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008488576643568218,
+      "loss": 0.4362,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008487122130950523,
+      "loss": 0.4142,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008485667043542506,
+      "loss": 0.6069,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484211381584015,
+      "loss": 0.5537,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008482755145314986,
+      "loss": 0.5349,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008481298334975456,
+      "loss": 0.5107,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008479840950805553,
+      "loss": 0.5928,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008478382993045504,
+      "loss": 0.4998,
+      "step": 2214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008476924461935625,
+      "loss": 0.4915,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008475465357716328,
+      "loss": 0.4462,
+      "step": 2216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008474005680628124,
+      "loss": 0.0612,
+      "step": 2217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008472545430911612,
+      "loss": 0.6654,
+      "step": 2218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008471084608807489,
+      "loss": 0.4286,
+      "step": 2219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008469623214556548,
+      "loss": 0.5433,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846816124839967,
+      "loss": 0.4932,
+      "step": 2221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008466698710577836,
+      "loss": 0.5522,
+      "step": 2222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846523560133212,
+      "loss": 0.0768,
+      "step": 2223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846377192090369,
+      "loss": 0.5178,
+      "step": 2224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008462307669533806,
+      "loss": 0.491,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008460842847463826,
+      "loss": 0.5839,
+      "step": 2226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008459377454935197,
+      "loss": 0.4099,
+      "step": 2227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008457911492189467,
+      "loss": 0.5555,
+      "step": 2228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000845644495946827,
+      "loss": 0.5975,
+      "step": 2229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000845497785701334,
+      "loss": 0.4996,
+      "step": 2230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008453510185066505,
+      "loss": 0.4426,
+      "step": 2231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000845204194386968,
+      "loss": 0.5911,
+      "step": 2232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008450573133664883,
+      "loss": 0.541,
+      "step": 2233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008449103754694218,
+      "loss": 0.4574,
+      "step": 2234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008447633807199887,
+      "loss": 0.6087,
+      "step": 2235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008446163291424186,
+      "loss": 0.7024,
+      "step": 2236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008444692207609502,
+      "loss": 0.4406,
+      "step": 2237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008443220555998319,
+      "loss": 0.4194,
+      "step": 2238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000844174833683321,
+      "loss": 0.1084,
+      "step": 2239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008440275550356847,
+      "loss": 0.4192,
+      "step": 2240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008438802196811991,
+      "loss": 0.506,
+      "step": 2241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00084373282764415,
+      "loss": 0.4662,
+      "step": 2242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008435853789488322,
+      "loss": 0.5142,
+      "step": 2243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008434378736195498,
+      "loss": 0.0616,
+      "step": 2244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000843290311680617,
+      "loss": 0.4338,
+      "step": 2245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431426931563564,
+      "loss": 0.6853,
+      "step": 2246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008429950180711002,
+      "loss": 0.4744,
+      "step": 2247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008428472864491903,
+      "loss": 0.5886,
+      "step": 2248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008426994983149775,
+      "loss": 0.4896,
+      "step": 2249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008425516536928222,
+      "loss": 0.5413,
+      "step": 2250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008424037526070936,
+      "loss": 0.5065,
+      "step": 2251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000842255795082171,
+      "loss": 0.5709,
+      "step": 2252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008421077811424424,
+      "loss": 0.6168,
+      "step": 2253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008419597108123054,
+      "loss": 0.4926,
+      "step": 2254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008418115841161664,
+      "loss": 0.5477,
+      "step": 2255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416634010784418,
+      "loss": 0.4263,
+      "step": 2256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841515161723557,
+      "loss": 0.4005,
+      "step": 2257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008413668660759464,
+      "loss": 0.5908,
+      "step": 2258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008412185141600539,
+      "loss": 0.5466,
+      "step": 2259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008410701060003328,
+      "loss": 0.5955,
+      "step": 2260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409216416212457,
+      "loss": 0.5323,
+      "step": 2261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008407731210472641,
+      "loss": 0.5027,
+      "step": 2262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000840624544302869,
+      "loss": 0.6478,
+      "step": 2263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008404759114125508,
+      "loss": 0.574,
+      "step": 2264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008403272224008089,
+      "loss": 0.6724,
+      "step": 2265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000840178477292152,
+      "loss": 0.5762,
+      "step": 2266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008400296761110982,
+      "loss": 0.4835,
+      "step": 2267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008398808188821747,
+      "loss": 0.4821,
+      "step": 2268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008397319056299178,
+      "loss": 0.5348,
+      "step": 2269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008395829363788734,
+      "loss": 0.5374,
+      "step": 2270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008394339111535966,
+      "loss": 0.5122,
+      "step": 2271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392848299786513,
+      "loss": 0.5525,
+      "step": 2272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008391356928786109,
+      "loss": 0.646,
+      "step": 2273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008389864998780582,
+      "loss": 0.4961,
+      "step": 2274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008388372510015849,
+      "loss": 0.4978,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008386879462737919,
+      "loss": 0.3746,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008385385857192898,
+      "loss": 0.5115,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008383891693626977,
+      "loss": 0.5902,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008382396972286446,
+      "loss": 0.6033,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008380901693417677,
+      "loss": 0.4869,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008379405857267148,
+      "loss": 0.5182,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008377909464081415,
+      "loss": 0.4623,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008376412514107135,
+      "loss": 0.4161,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008374915007591053,
+      "loss": 0.5007,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008373416944780006,
+      "loss": 0.5475,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008371918325920924,
+      "loss": 0.4531,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008370419151260827,
+      "loss": 0.5325,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008368919421046828,
+      "loss": 0.4255,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008367419135526131,
+      "loss": 0.3915,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008365918294946031,
+      "loss": 0.4453,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008364416899553915,
+      "loss": 0.4876,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008362914949597261,
+      "loss": 0.543,
+      "step": 2292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000836141244532364,
+      "loss": 0.551,
+      "step": 2293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008359909386980712,
+      "loss": 0.467,
+      "step": 2294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000835840577481623,
+      "loss": 0.4562,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008356901609078039,
+      "loss": 0.4852,
+      "step": 2296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355396890014073,
+      "loss": 0.5186,
+      "step": 2297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008353891617872358,
+      "loss": 0.5238,
+      "step": 2298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008352385792901012,
+      "loss": 0.4932,
+      "step": 2299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008350879415348243,
+      "loss": 0.4842,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008349372485462352,
+      "loss": 0.5408,
+      "step": 2301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008347865003491728,
+      "loss": 0.5522,
+      "step": 2302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008346356969684854,
+      "loss": 0.6768,
+      "step": 2303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083448483842903,
+      "loss": 0.5209,
+      "step": 2304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008343339247556734,
+      "loss": 0.506,
+      "step": 2305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341829559732907,
+      "loss": 0.5347,
+      "step": 2306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008340319321067667,
+      "loss": 0.4176,
+      "step": 2307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000833880853180995,
+      "loss": 0.5299,
+      "step": 2308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008337297192208778,
+      "loss": 0.4883,
+      "step": 2309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008335785302513272,
+      "loss": 0.5262,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008334272862972643,
+      "loss": 0.3978,
+      "step": 2311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008332759873836185,
+      "loss": 0.6,
+      "step": 2312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008331246335353291,
+      "loss": 0.5632,
+      "step": 2313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329732247773437,
+      "loss": 0.5096,
+      "step": 2314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008328217611346198,
+      "loss": 0.5016,
+      "step": 2315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008326702426321234,
+      "loss": 0.5071,
+      "step": 2316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008325186692948294,
+      "loss": 0.5543,
+      "step": 2317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008323670411477223,
+      "loss": 0.6682,
+      "step": 2318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008322153582157949,
+      "loss": 0.5204,
+      "step": 2319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008320636205240498,
+      "loss": 0.4526,
+      "step": 2320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008319118280974982,
+      "loss": 0.4193,
+      "step": 2321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008317599809611602,
+      "loss": 0.3967,
+      "step": 2322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008316080791400656,
+      "loss": 0.6394,
+      "step": 2323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314561226592521,
+      "loss": 0.4543,
+      "step": 2324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008313041115437673,
+      "loss": 0.585,
+      "step": 2325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008311520458186677,
+      "loss": 0.5296,
+      "step": 2326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008309999255090184,
+      "loss": 0.5442,
+      "step": 2327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008308477506398939,
+      "loss": 0.5063,
+      "step": 2328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008306955212363776,
+      "loss": 0.4573,
+      "step": 2329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008305432373235615,
+      "loss": 0.6395,
+      "step": 2330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008303908989265472,
+      "loss": 0.6113,
+      "step": 2331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000830238506070445,
+      "loss": 0.4553,
+      "step": 2332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008300860587803738,
+      "loss": 0.5052,
+      "step": 2333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008299335570814624,
+      "loss": 0.5521,
+      "step": 2334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008297810009988473,
+      "loss": 0.0604,
+      "step": 2335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008296283905576754,
+      "loss": 0.6903,
+      "step": 2336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294757257831012,
+      "loss": 0.5156,
+      "step": 2337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008293230067002894,
+      "loss": 0.4689,
+      "step": 2338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008291702333344124,
+      "loss": 0.4735,
+      "step": 2339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008290174057106527,
+      "loss": 0.0596,
+      "step": 2340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000828864523854201,
+      "loss": 0.5134,
+      "step": 2341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008287115877902572,
+      "loss": 0.532,
+      "step": 2342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00082855859754403,
+      "loss": 0.537,
+      "step": 2343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008284055531407373,
+      "loss": 0.4883,
+      "step": 2344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008282524546056056,
+      "loss": 0.5896,
+      "step": 2345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008280993019638706,
+      "loss": 0.4493,
+      "step": 2346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008279460952407767,
+      "loss": 0.4246,
+      "step": 2347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008277928344615775,
+      "loss": 0.463,
+      "step": 2348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008276395196515351,
+      "loss": 0.0575,
+      "step": 2349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000827486150835921,
+      "loss": 0.4812,
+      "step": 2350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000827332728040015,
+      "loss": 0.5953,
+      "step": 2351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271792512891063,
+      "loss": 0.5563,
+      "step": 2352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008270257206084929,
+      "loss": 0.4224,
+      "step": 2353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008268721360234814,
+      "loss": 0.4607,
+      "step": 2354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008267184975593877,
+      "loss": 0.4933,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008265648052415364,
+      "loss": 0.5808,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008264110590952608,
+      "loss": 0.5676,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008262572591459034,
+      "loss": 0.496,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008261034054188151,
+      "loss": 0.5126,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008259494979393563,
+      "loss": 0.6073,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008257955367328958,
+      "loss": 0.6151,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008256415218248112,
+      "loss": 0.5076,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008254874532404895,
+      "loss": 0.4282,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008253333310053257,
+      "loss": 0.5134,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008251791551447245,
+      "loss": 0.3792,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008250249256840991,
+      "loss": 0.5765,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008248706426488712,
+      "loss": 0.4604,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008247163060644718,
+      "loss": 0.5521,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008245619159563404,
+      "loss": 0.5248,
+      "step": 2369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008244074723499258,
+      "loss": 0.4478,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008242529752706849,
+      "loss": 0.4752,
+      "step": 2371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008240984247440842,
+      "loss": 0.3773,
+      "step": 2372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008239438207955982,
+      "loss": 0.4803,
+      "step": 2373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000823789163450711,
+      "loss": 0.4091,
+      "step": 2374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008236344527349147,
+      "loss": 0.6159,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008234796886737112,
+      "loss": 0.437,
+      "step": 2376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008233248712926101,
+      "loss": 0.6199,
+      "step": 2377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008231700006171305,
+      "loss": 0.4216,
+      "step": 2378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008230150766728001,
+      "loss": 0.4998,
+      "step": 2379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008228600994851554,
+      "loss": 0.7062,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008227050690797416,
+      "loss": 0.5966,
+      "step": 2381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008225499854821127,
+      "loss": 0.462,
+      "step": 2382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008223948487178314,
+      "loss": 0.5498,
+      "step": 2383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008222396588124695,
+      "loss": 0.5386,
+      "step": 2384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008220844157916073,
+      "loss": 0.4363,
+      "step": 2385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219291196808334,
+      "loss": 0.6222,
+      "step": 2386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008217737705057459,
+      "loss": 0.4728,
+      "step": 2387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008216183682919516,
+      "loss": 0.5255,
+      "step": 2388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008214629130650655,
+      "loss": 0.4327,
+      "step": 2389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008213074048507116,
+      "loss": 0.4626,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008211518436745231,
+      "loss": 0.4777,
+      "step": 2391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000820996229562141,
+      "loss": 0.4189,
+      "step": 2392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008208405625392159,
+      "loss": 0.5475,
+      "step": 2393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008206848426314064,
+      "loss": 0.5457,
+      "step": 2394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205290698643806,
+      "loss": 0.5647,
+      "step": 2395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008203732442638146,
+      "loss": 0.5518,
+      "step": 2396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008202173658553933,
+      "loss": 0.4434,
+      "step": 2397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008200614346648107,
+      "loss": 0.3731,
+      "step": 2398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008199054507177696,
+      "loss": 0.678,
+      "step": 2399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008197494140399808,
+      "loss": 0.4044,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195933246571643,
+      "loss": 0.6072,
+      "step": 2401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008194371825950488,
+      "loss": 0.527,
+      "step": 2402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008192809878793712,
+      "loss": 0.5173,
+      "step": 2403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008191247405358778,
+      "loss": 0.4495,
+      "step": 2404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818968440590323,
+      "loss": 0.0555,
+      "step": 2405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008188120880684702,
+      "loss": 0.4755,
+      "step": 2406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008186556829960913,
+      "loss": 0.574,
+      "step": 2407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008184992253989667,
+      "loss": 0.4993,
+      "step": 2408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008183427153028859,
+      "loss": 0.6514,
+      "step": 2409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818186152733647,
+      "loss": 0.4597,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008180295377170562,
+      "loss": 0.5525,
+      "step": 2411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008178728702789289,
+      "loss": 0.4843,
+      "step": 2412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008177161504450887,
+      "loss": 0.5188,
+      "step": 2413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175593782413684,
+      "loss": 0.7062,
+      "step": 2414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008174025536936091,
+      "loss": 0.6038,
+      "step": 2415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008172456768276604,
+      "loss": 0.3566,
+      "step": 2416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008170887476693807,
+      "loss": 0.575,
+      "step": 2417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008169317662446372,
+      "loss": 0.4243,
+      "step": 2418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167747325793051,
+      "loss": 0.4192,
+      "step": 2419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000816617646699269,
+      "loss": 0.6047,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008164605086304214,
+      "loss": 0.5173,
+      "step": 2421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008163033183986641,
+      "loss": 0.4399,
+      "step": 2422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008161460760299069,
+      "loss": 0.519,
+      "step": 2423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008159887815500684,
+      "loss": 0.4253,
+      "step": 2424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008158314349850759,
+      "loss": 0.5585,
+      "step": 2425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000815674036360865,
+      "loss": 0.5519,
+      "step": 2426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008155165857033804,
+      "loss": 0.4684,
+      "step": 2427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008153590830385749,
+      "loss": 0.5406,
+      "step": 2428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152015283924099,
+      "loss": 0.5732,
+      "step": 2429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008150439217908557,
+      "loss": 0.5547,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008148862632598906,
+      "loss": 0.4866,
+      "step": 2431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008147285528255022,
+      "loss": 0.5814,
+      "step": 2432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008145707905136863,
+      "loss": 0.5713,
+      "step": 2433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008144129763504468,
+      "loss": 0.5028,
+      "step": 2434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008142551103617972,
+      "loss": 0.4448,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008140971925737583,
+      "loss": 0.4202,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008139392230123604,
+      "loss": 0.5012,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008137812017036418,
+      "loss": 0.564,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008136231286736495,
+      "loss": 0.4843,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134650039484393,
+      "loss": 0.4758,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008133068275540751,
+      "loss": 0.5228,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008131485995166295,
+      "loss": 0.5485,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008129903198621833,
+      "loss": 0.5411,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008128319886168267,
+      "loss": 0.5593,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126736058066573,
+      "loss": 0.5997,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008125151714577819,
+      "loss": 0.5801,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008123566855963156,
+      "loss": 0.4133,
+      "step": 2447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008121981482483818,
+      "loss": 0.4172,
+      "step": 2448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008120395594401129,
+      "loss": 0.4838,
+      "step": 2449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000811880919197649,
+      "loss": 0.4869,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008117222275471396,
+      "loss": 0.5077,
+      "step": 2451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008115634845147422,
+      "loss": 0.5422,
+      "step": 2452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008114046901266224,
+      "loss": 0.457,
+      "step": 2453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008112458444089547,
+      "loss": 0.0573,
+      "step": 2454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008110869473879224,
+      "loss": 0.5612,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008109279990897164,
+      "loss": 0.6722,
+      "step": 2456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008107689995405367,
+      "loss": 0.663,
+      "step": 2457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008106099487665919,
+      "loss": 0.5728,
+      "step": 2458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000810450846794098,
+      "loss": 0.5077,
+      "step": 2459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008102916936492808,
+      "loss": 0.537,
+      "step": 2460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008101324893583734,
+      "loss": 0.4862,
+      "step": 2461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809973233947618,
+      "loss": 0.4615,
+      "step": 2462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008098139274432652,
+      "loss": 0.5238,
+      "step": 2463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008096545698715736,
+      "loss": 0.5258,
+      "step": 2464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008094951612588105,
+      "loss": 0.5466,
+      "step": 2465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008093357016312517,
+      "loss": 0.5239,
+      "step": 2466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008091761910151813,
+      "loss": 0.6801,
+      "step": 2467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008090166294368917,
+      "loss": 0.4364,
+      "step": 2468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008088570169226839,
+      "loss": 0.4801,
+      "step": 2469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808697353498867,
+      "loss": 0.4258,
+      "step": 2470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008085376391917589,
+      "loss": 0.5564,
+      "step": 2471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008083778740276857,
+      "loss": 0.4422,
+      "step": 2472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008082180580329818,
+      "loss": 0.4377,
+      "step": 2473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00080805819123399,
+      "loss": 0.5668,
+      "step": 2474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008078982736570612,
+      "loss": 0.5214,
+      "step": 2475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077383053285556,
+      "loss": 0.4863,
+      "step": 2476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008075782862748407,
+      "loss": 0.5736,
+      "step": 2477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807418216522293,
+      "loss": 0.4707,
+      "step": 2478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807258096097297,
+      "loss": 0.5496,
+      "step": 2479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807097925026246,
+      "loss": 0.5028,
+      "step": 2480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069377033355412,
+      "loss": 0.5699,
+      "step": 2481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000806777431051592,
+      "loss": 0.4734,
+      "step": 2482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000806617108200817,
+      "loss": 0.4661,
+      "step": 2483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064567348096423,
+      "loss": 0.4733,
+      "step": 2484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008062963109045026,
+      "loss": 0.5001,
+      "step": 2485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000806135836511841,
+      "loss": 0.5052,
+      "step": 2486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008059753116581087,
+      "loss": 0.5905,
+      "step": 2487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008058147363697657,
+      "loss": 0.7382,
+      "step": 2488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056541106732797,
+      "loss": 0.5077,
+      "step": 2489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008054934345951273,
+      "loss": 0.0562,
+      "step": 2490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008053327081617927,
+      "loss": 0.6393,
+      "step": 2491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008051719313997692,
+      "loss": 0.5464,
+      "step": 2492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008050111043355576,
+      "loss": 0.4681,
+      "step": 2493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008048502269956677,
+      "loss": 0.498,
+      "step": 2494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046892994066173,
+      "loss": 0.5463,
+      "step": 2495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008045283215949323,
+      "loss": 0.4554,
+      "step": 2496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008043672935871471,
+      "loss": 0.0552,
+      "step": 2497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008042062154098042,
+      "loss": 0.5687,
+      "step": 2498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008040450870894548,
+      "loss": 0.53,
+      "step": 2499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008038839086526577,
+      "loss": 0.4896,
+      "step": 2500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008037226801259805,
+      "loss": 0.4269,
+      "step": 2501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035614015359988,
+      "loss": 0.4502,
+      "step": 2502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008034000729092968,
+      "loss": 0.4318,
+      "step": 2503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008032386942724661,
+      "loss": 0.4427,
+      "step": 2504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008030772656521074,
+      "loss": 0.4438,
+      "step": 2505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008029157870748298,
+      "loss": 0.4951,
+      "step": 2506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008027542585672493,
+      "loss": 0.4806,
+      "step": 2507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008025926801559916,
+      "loss": 0.6456,
+      "step": 2508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008024310518676901,
+      "loss": 0.479,
+      "step": 2509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008022693737289859,
+      "loss": 0.4229,
+      "step": 2510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008021076457665293,
+      "loss": 0.5908,
+      "step": 2511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008019458680069778,
+      "loss": 0.4666,
+      "step": 2512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000801784040476998,
+      "loss": 0.5198,
+      "step": 2513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000801622163203264,
+      "loss": 0.4735,
+      "step": 2514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014602362124587,
+      "loss": 0.502,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008012982595312725,
+      "loss": 0.5723,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008011362331864049,
+      "loss": 0.5814,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008009741572045627,
+      "loss": 0.6503,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008008120316124612,
+      "loss": 0.4187,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008006498564368243,
+      "loss": 0.4948,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008004876317043835,
+      "loss": 0.394,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008003253574418785,
+      "loss": 0.3556,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008001630336760575,
+      "loss": 0.6241,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000800000660433677,
+      "loss": 0.7538,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007998382377415006,
+      "loss": 0.5703,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007996757656263015,
+      "loss": 0.5706,
+      "step": 2526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00079951324411486,
+      "loss": 0.5217,
+      "step": 2527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007993506732339653,
+      "loss": 0.4305,
+      "step": 2528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007991880530104136,
+      "loss": 0.5365,
+      "step": 2529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007990253834710108,
+      "loss": 0.533,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007988626646425694,
+      "loss": 0.5645,
+      "step": 2531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007986998965519111,
+      "loss": 0.5302,
+      "step": 2532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007985370792258652,
+      "loss": 0.4375,
+      "step": 2533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007983742126912694,
+      "loss": 0.4724,
+      "step": 2534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007982112969749692,
+      "loss": 0.5704,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007980483321038183,
+      "loss": 0.421,
+      "step": 2536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007978853181046788,
+      "loss": 0.5046,
+      "step": 2537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007977222550044205,
+      "loss": 0.4749,
+      "step": 2538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007975591428299218,
+      "loss": 0.4727,
+      "step": 2539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007973959816080684,
+      "loss": 0.0539,
+      "step": 2540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007972327713657547,
+      "loss": 0.0535,
+      "step": 2541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000797069512129883,
+      "loss": 0.5779,
+      "step": 2542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007969062039273639,
+      "loss": 0.5625,
+      "step": 2543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007967428467851157,
+      "loss": 0.5757,
+      "step": 2544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007965794407300647,
+      "loss": 0.4172,
+      "step": 2545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796415985789146,
+      "loss": 0.5045,
+      "step": 2546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007962524819893018,
+      "loss": 0.6232,
+      "step": 2547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007960889293574829,
+      "loss": 0.053,
+      "step": 2548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007959253279206482,
+      "loss": 0.6752,
+      "step": 2549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007957616777057645,
+      "loss": 0.7879,
+      "step": 2550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007955979787398066,
+      "loss": 0.4264,
+      "step": 2551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007954342310497575,
+      "loss": 0.485,
+      "step": 2552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952704346626077,
+      "loss": 0.3948,
+      "step": 2553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007951065896053566,
+      "loss": 0.5206,
+      "step": 2554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007949426959050108,
+      "loss": 0.5156,
+      "step": 2555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007947787535885853,
+      "loss": 0.504,
+      "step": 2556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007946147626831035,
+      "loss": 0.64,
+      "step": 2557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000794450723215596,
+      "loss": 0.5381,
+      "step": 2558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007942866352131017,
+      "loss": 0.4974,
+      "step": 2559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007941224987026681,
+      "loss": 0.5892,
+      "step": 2560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007939583137113497,
+      "loss": 0.5201,
+      "step": 2561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007937940802662098,
+      "loss": 0.3823,
+      "step": 2562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007936297983943192,
+      "loss": 0.4298,
+      "step": 2563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007934654681227571,
+      "loss": 0.5048,
+      "step": 2564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007933010894786101,
+      "loss": 0.5109,
+      "step": 2565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007931366624889734,
+      "loss": 0.4905,
+      "step": 2566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007929721871809494,
+      "loss": 0.5962,
+      "step": 2567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007928076635816497,
+      "loss": 0.6519,
+      "step": 2568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007926430917181924,
+      "loss": 0.3978,
+      "step": 2569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007924784716177047,
+      "loss": 0.5818,
+      "step": 2570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792313803307321,
+      "loss": 0.4465,
+      "step": 2571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007921490868141843,
+      "loss": 0.6274,
+      "step": 2572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007919843221654448,
+      "loss": 0.4226,
+      "step": 2573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007918195093882614,
+      "loss": 0.5309,
+      "step": 2574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007916546485098004,
+      "loss": 0.4289,
+      "step": 2575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791489739557236,
+      "loss": 0.6088,
+      "step": 2576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791324782557751,
+      "loss": 0.4698,
+      "step": 2577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007911597775385351,
+      "loss": 0.4379,
+      "step": 2578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007909947245267869,
+      "loss": 0.5105,
+      "step": 2579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007908296235497121,
+      "loss": 0.0516,
+      "step": 2580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007906644746345251,
+      "loss": 0.3851,
+      "step": 2581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007904992778084473,
+      "loss": 0.5133,
+      "step": 2582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903340330987087,
+      "loss": 0.3489,
+      "step": 2583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007901687405325471,
+      "loss": 0.0522,
+      "step": 2584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007900034001372078,
+      "loss": 0.4834,
+      "step": 2585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007898380119399446,
+      "loss": 0.4464,
+      "step": 2586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007896725759680183,
+      "loss": 0.5988,
+      "step": 2587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007895070922486984,
+      "loss": 0.5232,
+      "step": 2588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007893415608092619,
+      "loss": 0.4587,
+      "step": 2589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891759816769939,
+      "loss": 0.5047,
+      "step": 2590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007890103548791867,
+      "loss": 0.502,
+      "step": 2591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007888446804431415,
+      "loss": 0.4733,
+      "step": 2592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007886789583961666,
+      "loss": 0.4573,
+      "step": 2593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007885131887655782,
+      "loss": 0.5711,
+      "step": 2594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007883473715787006,
+      "loss": 0.4524,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007881815068628659,
+      "loss": 0.4608,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000788015594645414,
+      "loss": 0.5959,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878496349536921,
+      "loss": 0.4568,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007876836278150563,
+      "loss": 0.0517,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007875175732568698,
+      "loss": 0.4968,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007873514713065036,
+      "loss": 0.3925,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007871853219913367,
+      "loss": 0.4681,
+      "step": 2602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007870191253387561,
+      "loss": 0.5753,
+      "step": 2603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007868528813761561,
+      "loss": 0.4949,
+      "step": 2604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007866865901309391,
+      "loss": 0.4137,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007865202516305154,
+      "loss": 0.5878,
+      "step": 2606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007863538659023028,
+      "loss": 0.4668,
+      "step": 2607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007861874329737273,
+      "loss": 0.4237,
+      "step": 2608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007860209528722221,
+      "loss": 0.4385,
+      "step": 2609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007858544256252286,
+      "loss": 0.4526,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000785687851260196,
+      "loss": 0.525,
+      "step": 2611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000785521229804581,
+      "loss": 0.5216,
+      "step": 2612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007853545612858484,
+      "loss": 0.6278,
+      "step": 2613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007851878457314703,
+      "loss": 0.5673,
+      "step": 2614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007850210831689268,
+      "loss": 0.5493,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000784854273625706,
+      "loss": 0.0515,
+      "step": 2616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007846874171293033,
+      "loss": 0.5822,
+      "step": 2617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000784520513707222,
+      "loss": 0.5771,
+      "step": 2618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007843535633869734,
+      "loss": 0.4482,
+      "step": 2619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007841865661960761,
+      "loss": 0.4249,
+      "step": 2620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007840195221620567,
+      "loss": 0.4686,
+      "step": 2621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007838524313124493,
+      "loss": 0.3947,
+      "step": 2622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007836852936747964,
+      "loss": 0.5211,
+      "step": 2623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835181092766471,
+      "loss": 0.4879,
+      "step": 2624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007833508781455588,
+      "loss": 0.4796,
+      "step": 2625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007831836003090969,
+      "loss": 0.5449,
+      "step": 2626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007830162757948341,
+      "loss": 0.4384,
+      "step": 2627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007828489046303508,
+      "loss": 0.5198,
+      "step": 2628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007826814868432352,
+      "loss": 0.4042,
+      "step": 2629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007825140224610833,
+      "loss": 0.4429,
+      "step": 2630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007823465115114985,
+      "loss": 0.5037,
+      "step": 2631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007821789540220919,
+      "loss": 0.467,
+      "step": 2632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007820113500204825,
+      "loss": 0.5023,
+      "step": 2633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000781843699534297,
+      "loss": 0.5271,
+      "step": 2634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007816760025911694,
+      "loss": 0.4877,
+      "step": 2635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007815082592187417,
+      "loss": 0.5199,
+      "step": 2636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007813404694446632,
+      "loss": 0.4419,
+      "step": 2637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007811726332965914,
+      "loss": 0.5286,
+      "step": 2638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007810047508021909,
+      "loss": 0.4969,
+      "step": 2639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007808368219891342,
+      "loss": 0.4191,
+      "step": 2640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007806688468851014,
+      "loss": 0.4823,
+      "step": 2641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007805008255177803,
+      "loss": 0.4908,
+      "step": 2642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000780332757914866,
+      "loss": 0.4061,
+      "step": 2643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007801646441040617,
+      "loss": 0.495,
+      "step": 2644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779996484113078,
+      "loss": 0.4301,
+      "step": 2645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779828277969633,
+      "loss": 0.474,
+      "step": 2646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007796600257014525,
+      "loss": 0.443,
+      "step": 2647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007794917273362699,
+      "loss": 0.4761,
+      "step": 2648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007793233829018263,
+      "loss": 0.5444,
+      "step": 2649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007791549924258702,
+      "loss": 0.5118,
+      "step": 2650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007789865559361577,
+      "loss": 0.4725,
+      "step": 2651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007788180734604528,
+      "loss": 0.5913,
+      "step": 2652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007786495450265267,
+      "loss": 0.3995,
+      "step": 2653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007784809706621585,
+      "loss": 0.473,
+      "step": 2654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007783123503951343,
+      "loss": 0.4873,
+      "step": 2655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007781436842532488,
+      "loss": 0.5065,
+      "step": 2656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000777974972264303,
+      "loss": 0.4725,
+      "step": 2657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007778062144561064,
+      "loss": 0.4919,
+      "step": 2658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007776374108564757,
+      "loss": 0.4467,
+      "step": 2659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007774685614932353,
+      "loss": 0.4579,
+      "step": 2660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007772996663942167,
+      "loss": 0.5046,
+      "step": 2661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007771307255872596,
+      "loss": 0.5212,
+      "step": 2662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007769617391002105,
+      "loss": 0.4539,
+      "step": 2663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007767927069609242,
+      "loss": 0.0763,
+      "step": 2664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007766236291972626,
+      "loss": 0.5975,
+      "step": 2665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007764545058370951,
+      "loss": 0.4894,
+      "step": 2666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007762853369082987,
+      "loss": 0.5121,
+      "step": 2667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007761161224387579,
+      "loss": 0.4886,
+      "step": 2668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007759468624563646,
+      "loss": 0.4639,
+      "step": 2669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007757775569890186,
+      "loss": 0.4026,
+      "step": 2670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007756082060646266,
+      "loss": 0.6777,
+      "step": 2671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007754388097111033,
+      "loss": 0.6287,
+      "step": 2672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007752693679563706,
+      "loss": 0.4606,
+      "step": 2673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007750998808283581,
+      "loss": 0.459,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007749303483550026,
+      "loss": 0.5326,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007747607705642484,
+      "loss": 0.4954,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007745911474840476,
+      "loss": 0.5303,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007744214791423596,
+      "loss": 0.4493,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007742517655671511,
+      "loss": 0.5842,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007740820067863965,
+      "loss": 0.4434,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007739122028280775,
+      "loss": 0.5983,
+      "step": 2681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007737423537201831,
+      "loss": 0.437,
+      "step": 2682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007735724594907101,
+      "loss": 0.5559,
+      "step": 2683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007734025201676625,
+      "loss": 0.6497,
+      "step": 2684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007732325357790519,
+      "loss": 0.5898,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007730625063528969,
+      "loss": 0.4852,
+      "step": 2686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007728924319172242,
+      "loss": 0.5779,
+      "step": 2687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000772722312500067,
+      "loss": 0.5115,
+      "step": 2688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007725521481294674,
+      "loss": 0.5387,
+      "step": 2689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007723819388334731,
+      "loss": 0.5464,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007722116846401408,
+      "loss": 0.5114,
+      "step": 2691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007720413855775332,
+      "loss": 0.5173,
+      "step": 2692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007718710416737216,
+      "loss": 0.5759,
+      "step": 2693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771700652956784,
+      "loss": 0.4204,
+      "step": 2694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007715302194548058,
+      "loss": 0.6014,
+      "step": 2695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007713597411958802,
+      "loss": 0.4708,
+      "step": 2696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007711892182081073,
+      "loss": 0.4802,
+      "step": 2697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007710186505195951,
+      "loss": 0.5026,
+      "step": 2698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007708480381584584,
+      "loss": 0.5521,
+      "step": 2699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706773811528196,
+      "loss": 0.5037,
+      "step": 2700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007705066795308087,
+      "loss": 0.5488,
+      "step": 2701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007703359333205626,
+      "loss": 0.4211,
+      "step": 2702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007701651425502261,
+      "loss": 0.4714,
+      "step": 2703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007699943072479507,
+      "loss": 0.5088,
+      "step": 2704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007698234274418958,
+      "loss": 0.5231,
+      "step": 2705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007696525031602278,
+      "loss": 0.5543,
+      "step": 2706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007694815344311207,
+      "loss": 0.3997,
+      "step": 2707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007693105212827554,
+      "loss": 0.3887,
+      "step": 2708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007691394637433204,
+      "loss": 0.4105,
+      "step": 2709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007689683618410118,
+      "loss": 0.5132,
+      "step": 2710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007687972156040326,
+      "loss": 0.4303,
+      "step": 2711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768626025060593,
+      "loss": 0.4083,
+      "step": 2712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007684547902389111,
+      "loss": 0.5214,
+      "step": 2713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682835111672118,
+      "loss": 0.2211,
+      "step": 2714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768112187873727,
+      "loss": 0.4454,
+      "step": 2715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007679408203866969,
+      "loss": 0.4355,
+      "step": 2716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007677694087343681,
+      "loss": 0.4945,
+      "step": 2717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007675979529449949,
+      "loss": 0.4447,
+      "step": 2718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007674264530468386,
+      "loss": 0.4825,
+      "step": 2719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007672549090681682,
+      "loss": 0.5132,
+      "step": 2720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007670833210372592,
+      "loss": 0.5149,
+      "step": 2721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007669116889823954,
+      "loss": 0.5074,
+      "step": 2722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007667400129318669,
+      "loss": 0.5372,
+      "step": 2723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007665682929139716,
+      "loss": 0.7104,
+      "step": 2724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007663965289570146,
+      "loss": 0.4399,
+      "step": 2725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007662247210893081,
+      "loss": 0.4884,
+      "step": 2726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007660528693391713,
+      "loss": 0.3879,
+      "step": 2727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007658809737349313,
+      "loss": 0.5277,
+      "step": 2728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007657090343049219,
+      "loss": 0.5225,
+      "step": 2729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007655370510774842,
+      "loss": 0.5134,
+      "step": 2730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007653650240809667,
+      "loss": 0.5712,
+      "step": 2731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000765192953343725,
+      "loss": 0.5524,
+      "step": 2732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007650208388941217,
+      "loss": 0.5518,
+      "step": 2733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007648486807605273,
+      "loss": 0.4575,
+      "step": 2734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007646764789713184,
+      "loss": 0.1033,
+      "step": 2735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007645042335548799,
+      "loss": 0.5341,
+      "step": 2736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007643319445396031,
+      "loss": 0.52,
+      "step": 2737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007641596119538871,
+      "loss": 0.3882,
+      "step": 2738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007639872358261377,
+      "loss": 0.5607,
+      "step": 2739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000763814816184768,
+      "loss": 0.4977,
+      "step": 2740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007636423530581984,
+      "loss": 0.4729,
+      "step": 2741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007634698464748562,
+      "loss": 0.488,
+      "step": 2742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632972964631764,
+      "loss": 0.4553,
+      "step": 2743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007631247030516007,
+      "loss": 0.4561,
+      "step": 2744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007629520662685779,
+      "loss": 0.6111,
+      "step": 2745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762779386142564,
+      "loss": 0.519,
+      "step": 2746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007626066627020227,
+      "loss": 0.4108,
+      "step": 2747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007624338959754242,
+      "loss": 0.3987,
+      "step": 2748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762261085991246,
+      "loss": 0.535,
+      "step": 2749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007620882327779726,
+      "loss": 0.4378,
+      "step": 2750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007619153363640962,
+      "loss": 0.4913,
+      "step": 2751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007617423967781154,
+      "loss": 0.4666,
+      "step": 2752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007615694140485363,
+      "loss": 0.0816,
+      "step": 2753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007613963882038722,
+      "loss": 0.578,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007612233192726429,
+      "loss": 0.5151,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007610502072833762,
+      "loss": 0.4843,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007608770522646065,
+      "loss": 0.5995,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007607038542448751,
+      "loss": 0.5098,
+      "step": 2758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007605306132527308,
+      "loss": 0.4897,
+      "step": 2759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007603573293167292,
+      "loss": 0.4845,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601840024654331,
+      "loss": 0.453,
+      "step": 2761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007600106327274125,
+      "loss": 0.4082,
+      "step": 2762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007598372201312444,
+      "loss": 0.5648,
+      "step": 2763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007596637647055124,
+      "loss": 0.6927,
+      "step": 2764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007594902664788078,
+      "loss": 0.4373,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593167254797288,
+      "loss": 0.5385,
+      "step": 2766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007591431417368804,
+      "loss": 0.5651,
+      "step": 2767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007589695152788749,
+      "loss": 0.5865,
+      "step": 2768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007587958461343316,
+      "loss": 0.4749,
+      "step": 2769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007586221343318766,
+      "loss": 0.4495,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007584483799001431,
+      "loss": 0.4664,
+      "step": 2771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007582745828677719,
+      "loss": 0.4938,
+      "step": 2772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007581007432634102,
+      "loss": 0.5123,
+      "step": 2773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007579268611157122,
+      "loss": 0.3876,
+      "step": 2774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007577529364533392,
+      "loss": 0.673,
+      "step": 2775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007575789693049599,
+      "loss": 0.5016,
+      "step": 2776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007574049596992497,
+      "loss": 0.4426,
+      "step": 2777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007572309076648907,
+      "loss": 0.5127,
+      "step": 2778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570568132305728,
+      "loss": 0.4331,
+      "step": 2779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007568826764249918,
+      "loss": 0.4799,
+      "step": 2780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007567084972768515,
+      "loss": 0.4598,
+      "step": 2781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000756534275814862,
+      "loss": 0.4108,
+      "step": 2782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007563600120677407,
+      "loss": 0.4569,
+      "step": 2783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007561857060642119,
+      "loss": 0.4644,
+      "step": 2784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007560113578330068,
+      "loss": 0.5192,
+      "step": 2785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007558369674028638,
+      "loss": 0.553,
+      "step": 2786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007556625348025278,
+      "loss": 0.4017,
+      "step": 2787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007554880600607512,
+      "loss": 0.4115,
+      "step": 2788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007553135432062929,
+      "loss": 0.4254,
+      "step": 2789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007551389842679188,
+      "loss": 0.4005,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007549643832744023,
+      "loss": 0.3864,
+      "step": 2791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007547897402545228,
+      "loss": 0.3952,
+      "step": 2792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007546150552370673,
+      "loss": 0.0809,
+      "step": 2793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007544403282508296,
+      "loss": 0.5128,
+      "step": 2794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542655593246102,
+      "loss": 0.5897,
+      "step": 2795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007540907484872168,
+      "loss": 0.4517,
+      "step": 2796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007539158957674639,
+      "loss": 0.5511,
+      "step": 2797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007537410011941725,
+      "loss": 0.4601,
+      "step": 2798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007535660647961715,
+      "loss": 0.6289,
+      "step": 2799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007533910866022955,
+      "loss": 0.5299,
+      "step": 2800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000753216066641387,
+      "loss": 0.5027,
+      "step": 2801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007530410049422945,
+      "loss": 0.4994,
+      "step": 2802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007528659015338742,
+      "loss": 0.5105,
+      "step": 2803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007526907564449884,
+      "loss": 0.4374,
+      "step": 2804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007525155697045069,
+      "loss": 0.4302,
+      "step": 2805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007523403413413065,
+      "loss": 0.5536,
+      "step": 2806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007521650713842699,
+      "loss": 0.4978,
+      "step": 2807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519897598622873,
+      "loss": 0.3864,
+      "step": 2808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007518144068042561,
+      "loss": 0.5276,
+      "step": 2809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007516390122390799,
+      "loss": 0.4382,
+      "step": 2810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007514635761956693,
+      "loss": 0.5941,
+      "step": 2811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751288098702942,
+      "loss": 0.4053,
+      "step": 2812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511125797898223,
+      "loss": 0.5109,
+      "step": 2813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007509370194852414,
+      "loss": 0.5822,
+      "step": 2814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007507614178181373,
+      "loss": 0.5634,
+      "step": 2815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007505857748174545,
+      "loss": 0.5206,
+      "step": 2816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007504100905121454,
+      "loss": 0.4001,
+      "step": 2817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007502343649311675,
+      "loss": 0.4679,
+      "step": 2818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007500585981034868,
+      "loss": 0.5105,
+      "step": 2819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007498827900580753,
+      "loss": 0.449,
+      "step": 2820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007497069408239113,
+      "loss": 0.4176,
+      "step": 2821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007495310504299808,
+      "loss": 0.5623,
+      "step": 2822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007493551189052764,
+      "loss": 0.45,
+      "step": 2823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007491791462787969,
+      "loss": 0.4873,
+      "step": 2824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007490031325795485,
+      "loss": 0.554,
+      "step": 2825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007488270778365438,
+      "loss": 0.4518,
+      "step": 2826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007486509820788024,
+      "loss": 0.4592,
+      "step": 2827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007484748453353507,
+      "loss": 0.5831,
+      "step": 2828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007482986676352215,
+      "loss": 0.5353,
+      "step": 2829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007481224490074545,
+      "loss": 0.4073,
+      "step": 2830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007479461894810966,
+      "loss": 0.5204,
+      "step": 2831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007477698890852008,
+      "loss": 0.5001,
+      "step": 2832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007475935478488269,
+      "loss": 0.4669,
+      "step": 2833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007474171658010421,
+      "loss": 0.5253,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007472407429709195,
+      "loss": 0.5751,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007470642793875393,
+      "loss": 0.5044,
+      "step": 2836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007468877750799886,
+      "loss": 0.5022,
+      "step": 2837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007467112300773609,
+      "loss": 0.4946,
+      "step": 2838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465346444087566,
+      "loss": 0.4497,
+      "step": 2839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007463580181032826,
+      "loss": 0.4319,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007461813511900526,
+      "loss": 0.5382,
+      "step": 2841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007460046436981871,
+      "loss": 0.5046,
+      "step": 2842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007458278956568132,
+      "loss": 0.4276,
+      "step": 2843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007456511070950646,
+      "loss": 0.4403,
+      "step": 2844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007454742780420817,
+      "loss": 0.4799,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007452974085270121,
+      "loss": 0.4033,
+      "step": 2846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000745120498579009,
+      "loss": 0.6257,
+      "step": 2847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007449435482272331,
+      "loss": 0.463,
+      "step": 2848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007447665575008517,
+      "loss": 0.4954,
+      "step": 2849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007445895264290384,
+      "loss": 0.3979,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007444124550409736,
+      "loss": 0.4237,
+      "step": 2851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007442353433658445,
+      "loss": 0.5103,
+      "step": 2852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000744058191432845,
+      "loss": 0.5352,
+      "step": 2853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007438809992711751,
+      "loss": 0.4786,
+      "step": 2854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007437037669100417,
+      "loss": 0.4601,
+      "step": 2855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007435264943786589,
+      "loss": 0.452,
+      "step": 2856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007433491817062468,
+      "loss": 0.4911,
+      "step": 2857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000743171828922032,
+      "loss": 0.6045,
+      "step": 2858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007429944360552481,
+      "loss": 0.6162,
+      "step": 2859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007428170031351352,
+      "loss": 0.5068,
+      "step": 2860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007426395301909399,
+      "loss": 0.5214,
+      "step": 2861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007424620172519155,
+      "loss": 0.5074,
+      "step": 2862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742284464347322,
+      "loss": 0.5603,
+      "step": 2863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007421068715064258,
+      "loss": 0.4951,
+      "step": 2864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007419292387584997,
+      "loss": 0.4337,
+      "step": 2865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007417515661328238,
+      "loss": 0.4512,
+      "step": 2866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007415738536586837,
+      "loss": 0.5216,
+      "step": 2867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007413961013653725,
+      "loss": 0.5402,
+      "step": 2868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007412183092821895,
+      "loss": 0.4608,
+      "step": 2869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007410404774384403,
+      "loss": 0.4171,
+      "step": 2870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007408626058634378,
+      "loss": 0.5446,
+      "step": 2871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007406846945865006,
+      "loss": 0.6511,
+      "step": 2872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007405067436369545,
+      "loss": 0.0849,
+      "step": 2873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007403287530441312,
+      "loss": 0.4929,
+      "step": 2874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007401507228373698,
+      "loss": 0.4625,
+      "step": 2875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007399726530460149,
+      "loss": 0.4999,
+      "step": 2876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007397945436994185,
+      "loss": 0.437,
+      "step": 2877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007396163948269385,
+      "loss": 0.5173,
+      "step": 2878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00073943820645794,
+      "loss": 0.3619,
+      "step": 2879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007392599786217939,
+      "loss": 0.5621,
+      "step": 2880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007390817113478778,
+      "loss": 0.6031,
+      "step": 2881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007389034046655762,
+      "loss": 0.3981,
+      "step": 2882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007387250586042795,
+      "loss": 0.571,
+      "step": 2883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007385466731933852,
+      "loss": 0.5939,
+      "step": 2884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007383682484622966,
+      "loss": 0.5627,
+      "step": 2885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007381897844404243,
+      "loss": 0.4655,
+      "step": 2886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007380112811571847,
+      "loss": 0.0732,
+      "step": 2887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007378327386420008,
+      "loss": 0.4805,
+      "step": 2888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007376541569243022,
+      "loss": 0.537,
+      "step": 2889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374755360335252,
+      "loss": 0.4448,
+      "step": 2890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000737296875999112,
+      "loss": 0.4012,
+      "step": 2891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007371181768505116,
+      "loss": 0.5732,
+      "step": 2892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007369394386171795,
+      "loss": 0.489,
+      "step": 2893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007367606613285776,
+      "loss": 0.5165,
+      "step": 2894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365818450141738,
+      "loss": 0.4194,
+      "step": 2895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007364029897034432,
+      "loss": 0.4547,
+      "step": 2896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007362240954258668,
+      "loss": 0.4836,
+      "step": 2897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007360451622109319,
+      "loss": 0.4456,
+      "step": 2898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007358661900881328,
+      "loss": 0.0672,
+      "step": 2899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356871790869698,
+      "loss": 0.4415,
+      "step": 2900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007355081292369498,
+      "loss": 0.5256,
+      "step": 2901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007353290405675857,
+      "loss": 0.5392,
+      "step": 2902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007351499131083973,
+      "loss": 0.4067,
+      "step": 2903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007349707468889106,
+      "loss": 0.5995,
+      "step": 2904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000734791541938658,
+      "loss": 0.4517,
+      "step": 2905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007346122982871779,
+      "loss": 0.4254,
+      "step": 2906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000734433015964016,
+      "loss": 0.5259,
+      "step": 2907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007342536949987235,
+      "loss": 0.4595,
+      "step": 2908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007340743354208583,
+      "loss": 0.5422,
+      "step": 2909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007338949372599848,
+      "loss": 0.5293,
+      "step": 2910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007337155005456733,
+      "loss": 0.5601,
+      "step": 2911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007335360253075012,
+      "loss": 0.4821,
+      "step": 2912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333565115750513,
+      "loss": 0.0628,
+      "step": 2913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007331769593779136,
+      "loss": 0.0626,
+      "step": 2914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007329973687456841,
+      "loss": 0.457,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007328177397079652,
+      "loss": 0.4697,
+      "step": 2916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007326380722943652,
+      "loss": 0.5239,
+      "step": 2917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007324583665344993,
+      "loss": 0.5619,
+      "step": 2918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007322786224579891,
+      "loss": 0.4473,
+      "step": 2919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007320988400944617,
+      "loss": 0.4619,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007319190194735514,
+      "loss": 0.4725,
+      "step": 2921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007317391606248983,
+      "loss": 0.5031,
+      "step": 2922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007315592635781492,
+      "loss": 0.4727,
+      "step": 2923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007313793283629565,
+      "loss": 0.4062,
+      "step": 2924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007311993550089797,
+      "loss": 0.4526,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007310193435458841,
+      "loss": 0.4973,
+      "step": 2926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007308392940033413,
+      "loss": 0.4717,
+      "step": 2927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007306592064110293,
+      "loss": 0.467,
+      "step": 2928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007304790807986326,
+      "loss": 0.5643,
+      "step": 2929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007302989171958413,
+      "loss": 0.4864,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007301187156323526,
+      "loss": 0.5438,
+      "step": 2931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007299384761378693,
+      "loss": 0.6433,
+      "step": 2932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007297581987421007,
+      "loss": 0.5122,
+      "step": 2933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295778834747624,
+      "loss": 0.566,
+      "step": 2934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293975303655762,
+      "loss": 0.5272,
+      "step": 2935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007292171394442699,
+      "loss": 0.3783,
+      "step": 2936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007290367107405782,
+      "loss": 0.5135,
+      "step": 2937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000728856244284241,
+      "loss": 0.496,
+      "step": 2938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007286757401050054,
+      "loss": 0.5284,
+      "step": 2939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007284951982326242,
+      "loss": 0.4779,
+      "step": 2940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007283146186968565,
+      "loss": 0.4834,
+      "step": 2941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007281340015274678,
+      "loss": 0.4332,
+      "step": 2942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007279533467542294,
+      "loss": 0.0589,
+      "step": 2943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007277726544069192,
+      "loss": 0.3602,
+      "step": 2944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000727591924515321,
+      "loss": 0.4589,
+      "step": 2945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007274111571092252,
+      "loss": 0.5734,
+      "step": 2946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007272303522184279,
+      "loss": 0.4366,
+      "step": 2947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007270495098727314,
+      "loss": 0.4349,
+      "step": 2948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007268686301019447,
+      "loss": 0.4092,
+      "step": 2949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007266877129358825,
+      "loss": 0.0586,
+      "step": 2950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007265067584043656,
+      "loss": 0.4062,
+      "step": 2951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007263257665372215,
+      "loss": 0.3936,
+      "step": 2952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007261447373642833,
+      "loss": 0.4655,
+      "step": 2953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007259636709153905,
+      "loss": 0.4271,
+      "step": 2954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007257825672203886,
+      "loss": 0.5604,
+      "step": 2955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007256014263091295,
+      "loss": 0.4402,
+      "step": 2956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007254202482114709,
+      "loss": 0.6085,
+      "step": 2957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007252390329572769,
+      "loss": 0.4485,
+      "step": 2958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007250577805764177,
+      "loss": 0.5083,
+      "step": 2959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007248764910987693,
+      "loss": 0.4546,
+      "step": 2960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007246951645542142,
+      "loss": 0.4487,
+      "step": 2961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007245138009726411,
+      "loss": 0.5927,
+      "step": 2962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007243324003839443,
+      "loss": 0.4102,
+      "step": 2963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007241509628180244,
+      "loss": 0.5868,
+      "step": 2964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007239694883047883,
+      "loss": 0.4366,
+      "step": 2965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007237879768741489,
+      "loss": 0.509,
+      "step": 2966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007236064285560252,
+      "loss": 0.4434,
+      "step": 2967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007234248433803421,
+      "loss": 0.3858,
+      "step": 2968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007232432213770307,
+      "loss": 0.5615,
+      "step": 2969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230615625760282,
+      "loss": 0.4257,
+      "step": 2970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007228798670072778,
+      "loss": 0.5331,
+      "step": 2971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000722698134700729,
+      "loss": 0.424,
+      "step": 2972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007225163656863369,
+      "loss": 0.457,
+      "step": 2973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000722334559994063,
+      "loss": 0.4912,
+      "step": 2974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007221527176538746,
+      "loss": 0.5337,
+      "step": 2975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007219708386957453,
+      "loss": 0.3722,
+      "step": 2976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007217889231496548,
+      "loss": 0.5018,
+      "step": 2977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216069710455886,
+      "loss": 0.4869,
+      "step": 2978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007214249824135379,
+      "loss": 0.4805,
+      "step": 2979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007212429572835007,
+      "loss": 0.5593,
+      "step": 2980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007210608956854807,
+      "loss": 0.5233,
+      "step": 2981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007208787976494871,
+      "loss": 0.5228,
+      "step": 2982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007206966632055358,
+      "loss": 0.6562,
+      "step": 2983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007205144923836485,
+      "loss": 0.4844,
+      "step": 2984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007203322852138528,
+      "loss": 0.6224,
+      "step": 2985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007201500417261822,
+      "loss": 0.5081,
+      "step": 2986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007199677619506763,
+      "loss": 0.4249,
+      "step": 2987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000719785445917381,
+      "loss": 0.4216,
+      "step": 2988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007196030936563474,
+      "loss": 0.4623,
+      "step": 2989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007194207051976334,
+      "loss": 0.4866,
+      "step": 2990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007192382805713024,
+      "loss": 0.4211,
+      "step": 2991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007190558198074239,
+      "loss": 0.4901,
+      "step": 2992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007188733229360731,
+      "loss": 0.4969,
+      "step": 2993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007186907899873319,
+      "loss": 0.451,
+      "step": 2994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000718508220991287,
+      "loss": 0.4938,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007183256159780321,
+      "loss": 0.57,
+      "step": 2996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007181429749776662,
+      "loss": 0.5403,
+      "step": 2997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007179602980202942,
+      "loss": 0.4166,
+      "step": 2998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007177775851360278,
+      "loss": 0.4082,
+      "step": 2999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007175948363549835,
+      "loss": 0.452,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007174120517072845,
+      "loss": 0.4797,
+      "step": 3001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007172292312230593,
+      "loss": 0.517,
+      "step": 3002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007170463749324428,
+      "loss": 0.389,
+      "step": 3003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007168634828655755,
+      "loss": 0.4304,
+      "step": 3004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007166805550526041,
+      "loss": 0.3803,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007164975915236808,
+      "loss": 0.4308,
+      "step": 3006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007163145923089642,
+      "loss": 0.4821,
+      "step": 3007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000716131557438618,
+      "loss": 0.4645,
+      "step": 3008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159484869428128,
+      "loss": 0.5171,
+      "step": 3009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007157653808517241,
+      "loss": 0.4208,
+      "step": 3010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000715582239195534,
+      "loss": 0.4418,
+      "step": 3011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007153990620044299,
+      "loss": 0.4708,
+      "step": 3012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007152158493086056,
+      "loss": 0.4355,
+      "step": 3013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007150326011382603,
+      "loss": 0.4524,
+      "step": 3014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007148493175235991,
+      "loss": 0.3883,
+      "step": 3015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007146659984948332,
+      "loss": 0.4634,
+      "step": 3016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007144826440821795,
+      "loss": 0.3933,
+      "step": 3017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007142992543158609,
+      "loss": 0.4238,
+      "step": 3018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007141158292261055,
+      "loss": 0.3866,
+      "step": 3019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007139323688431481,
+      "loss": 0.6895,
+      "step": 3020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007137488731972286,
+      "loss": 0.4613,
+      "step": 3021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007135653423185933,
+      "loss": 0.4418,
+      "step": 3022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007133817762374937,
+      "loss": 0.0647,
+      "step": 3023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007131981749841876,
+      "loss": 0.563,
+      "step": 3024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007130145385889385,
+      "loss": 0.4338,
+      "step": 3025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007128308670820154,
+      "loss": 0.5199,
+      "step": 3026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007126471604936933,
+      "loss": 0.4158,
+      "step": 3027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007124634188542531,
+      "loss": 0.491,
+      "step": 3028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007122796421939815,
+      "loss": 0.4371,
+      "step": 3029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007120958305431705,
+      "loss": 0.5585,
+      "step": 3030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007119119839321185,
+      "loss": 0.3568,
+      "step": 3031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007117281023911289,
+      "loss": 0.4248,
+      "step": 3032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007115441859505119,
+      "loss": 0.4264,
+      "step": 3033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007113602346405824,
+      "loss": 0.4564,
+      "step": 3034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007111762484916619,
+      "loss": 0.4249,
+      "step": 3035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007109922275340769,
+      "loss": 0.5458,
+      "step": 3036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007108081717981603,
+      "loss": 0.0664,
+      "step": 3037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007106240813142501,
+      "loss": 0.484,
+      "step": 3038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007104399561126906,
+      "loss": 0.3655,
+      "step": 3039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007102557962238317,
+      "loss": 0.5452,
+      "step": 3040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007100716016780284,
+      "loss": 0.4309,
+      "step": 3041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007098873725056424,
+      "loss": 0.4221,
+      "step": 3042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007097031087370403,
+      "loss": 0.6619,
+      "step": 3043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007095188104025951,
+      "loss": 0.4347,
+      "step": 3044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007093344775326847,
+      "loss": 0.4963,
+      "step": 3045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007091501101576934,
+      "loss": 0.6273,
+      "step": 3046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007089657083080106,
+      "loss": 0.4067,
+      "step": 3047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000708781272014032,
+      "loss": 0.4749,
+      "step": 3048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007085968013061584,
+      "loss": 0.4548,
+      "step": 3049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007084122962147968,
+      "loss": 0.4039,
+      "step": 3050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007082277567703593,
+      "loss": 0.4038,
+      "step": 3051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007080431830032641,
+      "loss": 0.4956,
+      "step": 3052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007078585749439349,
+      "loss": 0.0614,
+      "step": 3053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007076739326228012,
+      "loss": 0.4692,
+      "step": 3054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007074892560702979,
+      "loss": 0.5027,
+      "step": 3055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007073045453168655,
+      "loss": 0.4273,
+      "step": 3056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007071198003929507,
+      "loss": 0.496,
+      "step": 3057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007069350213290049,
+      "loss": 0.558,
+      "step": 3058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007067502081554861,
+      "loss": 0.0607,
+      "step": 3059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007065653609028573,
+      "loss": 0.6002,
+      "step": 3060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007063804796015872,
+      "loss": 0.4146,
+      "step": 3061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007061955642821505,
+      "loss": 0.4324,
+      "step": 3062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007060106149750269,
+      "loss": 0.4561,
+      "step": 3063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007058256317107023,
+      "loss": 0.4623,
+      "step": 3064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007056406145196677,
+      "loss": 0.4579,
+      "step": 3065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007054555634324199,
+      "loss": 0.4478,
+      "step": 3066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007052704784794614,
+      "loss": 0.3896,
+      "step": 3067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007050853596913,
+      "loss": 0.5198,
+      "step": 3068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007049002070984497,
+      "loss": 0.4922,
+      "step": 3069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007047150207314292,
+      "loss": 0.5226,
+      "step": 3070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007045298006207632,
+      "loss": 0.5743,
+      "step": 3071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007043445467969822,
+      "loss": 0.0588,
+      "step": 3072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000704159259290622,
+      "loss": 0.5768,
+      "step": 3073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007039739381322237,
+      "loss": 0.5278,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007037885833523343,
+      "loss": 0.457,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007036031949815063,
+      "loss": 0.475,
+      "step": 3076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034177730502978,
+      "loss": 0.4316,
+      "step": 3077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007032323175892723,
+      "loss": 0.4653,
+      "step": 3078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007030468286289986,
+      "loss": 0.5875,
+      "step": 3079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007028613062000516,
+      "loss": 0.4089,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007026757503330112,
+      "loss": 0.3584,
+      "step": 3081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007024901610584628,
+      "loss": 0.4741,
+      "step": 3082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000702304538406998,
+      "loss": 0.512,
+      "step": 3083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000702118882409213,
+      "loss": 0.4185,
+      "step": 3084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007019331930957101,
+      "loss": 0.5021,
+      "step": 3085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007017474704970968,
+      "loss": 0.4263,
+      "step": 3086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007015617146439862,
+      "loss": 0.4667,
+      "step": 3087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007013759255669969,
+      "loss": 0.4016,
+      "step": 3088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007011901032967529,
+      "loss": 0.4949,
+      "step": 3089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007010042478638836,
+      "loss": 0.4998,
+      "step": 3090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007008183592990243,
+      "loss": 0.3802,
+      "step": 3091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007006324376328152,
+      "loss": 0.4523,
+      "step": 3092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000700446482895902,
+      "loss": 0.4484,
+      "step": 3093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007002604951189366,
+      "loss": 0.5211,
+      "step": 3094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007000744743325755,
+      "loss": 0.5142,
+      "step": 3095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006998884205674805,
+      "loss": 0.5273,
+      "step": 3096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00069970233385432,
+      "loss": 0.462,
+      "step": 3097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006995162142237667,
+      "loss": 0.4072,
+      "step": 3098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006993300617064992,
+      "loss": 0.4434,
+      "step": 3099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006991438763332016,
+      "loss": 0.4266,
+      "step": 3100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006989576581345628,
+      "loss": 0.4938,
+      "step": 3101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006987714071412781,
+      "loss": 0.5288,
+      "step": 3102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006985851233840473,
+      "loss": 0.5229,
+      "step": 3103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006983988068935761,
+      "loss": 0.5262,
+      "step": 3104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006982124577005755,
+      "loss": 0.0571,
+      "step": 3105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698026075835762,
+      "loss": 0.4374,
+      "step": 3106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006978396613298571,
+      "loss": 0.4222,
+      "step": 3107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006976532142135881,
+      "loss": 0.4713,
+      "step": 3108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006974667345176874,
+      "loss": 0.5461,
+      "step": 3109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000697280222272893,
+      "loss": 0.4624,
+      "step": 3110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000697093677509948,
+      "loss": 0.552,
+      "step": 3111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000696907100259601,
+      "loss": 0.5293,
+      "step": 3112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006967204905526062,
+      "loss": 0.4618,
+      "step": 3113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965338484197225,
+      "loss": 0.5011,
+      "step": 3114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006963471738917149,
+      "loss": 0.4402,
+      "step": 3115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006961604669993533,
+      "loss": 0.5005,
+      "step": 3116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006959737277734129,
+      "loss": 0.5334,
+      "step": 3117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006957869562446746,
+      "loss": 0.4117,
+      "step": 3118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006956001524439239,
+      "loss": 0.3373,
+      "step": 3119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006954133164019528,
+      "loss": 0.4508,
+      "step": 3120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006952264481495574,
+      "loss": 0.4144,
+      "step": 3121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006950395477175398,
+      "loss": 0.5114,
+      "step": 3122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006948526151367069,
+      "loss": 0.5612,
+      "step": 3123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006946656504378718,
+      "loss": 0.3828,
+      "step": 3124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006944786536518519,
+      "loss": 0.3892,
+      "step": 3125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006942916248094704,
+      "loss": 0.4706,
+      "step": 3126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006941045639415557,
+      "loss": 0.5703,
+      "step": 3127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006939174710789416,
+      "loss": 0.6202,
+      "step": 3128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006937303462524668,
+      "loss": 0.0562,
+      "step": 3129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006935431894929758,
+      "loss": 0.5375,
+      "step": 3130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000693356000831318,
+      "loss": 0.3851,
+      "step": 3131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006931687802983478,
+      "loss": 0.4719,
+      "step": 3132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006929815279249256,
+      "loss": 0.4789,
+      "step": 3133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006927942437419165,
+      "loss": 0.5212,
+      "step": 3134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000692606927780191,
+      "loss": 0.4987,
+      "step": 3135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006924195800706247,
+      "loss": 0.5281,
+      "step": 3136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006922322006440986,
+      "loss": 0.4845,
+      "step": 3137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006920447895314989,
+      "loss": 0.4381,
+      "step": 3138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006918573467637173,
+      "loss": 0.0556,
+      "step": 3139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006916698723716499,
+      "loss": 0.5198,
+      "step": 3140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000691482366386199,
+      "loss": 0.4727,
+      "step": 3141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006912948288382712,
+      "loss": 0.4723,
+      "step": 3142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006911072597587791,
+      "loss": 0.4554,
+      "step": 3143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00069091965917864,
+      "loss": 0.4575,
+      "step": 3144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006907320271287763,
+      "loss": 0.4442,
+      "step": 3145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006905443636401162,
+      "loss": 0.4043,
+      "step": 3146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006903566687435927,
+      "loss": 0.5183,
+      "step": 3147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006901689424701436,
+      "loss": 0.4562,
+      "step": 3148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006899811848507126,
+      "loss": 0.4678,
+      "step": 3149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006897933959162482,
+      "loss": 0.4432,
+      "step": 3150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006896055756977038,
+      "loss": 0.4412,
+      "step": 3151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006894177242260385,
+      "loss": 0.3993,
+      "step": 3152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006892298415322164,
+      "loss": 0.4595,
+      "step": 3153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006890419276472062,
+      "loss": 0.391,
+      "step": 3154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006888539826019824,
+      "loss": 0.4368,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006886660064275245,
+      "loss": 0.4822,
+      "step": 3156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006884779991548171,
+      "loss": 0.4257,
+      "step": 3157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006882899608148496,
+      "loss": 0.462,
+      "step": 3158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000688101891438617,
+      "loss": 0.3861,
+      "step": 3159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006879137910571191,
+      "loss": 0.4694,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006877256597013611,
+      "loss": 0.4146,
+      "step": 3161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006875374974023527,
+      "loss": 0.588,
+      "step": 3162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006873493041911097,
+      "loss": 0.5205,
+      "step": 3163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006871610800986518,
+      "loss": 0.5575,
+      "step": 3164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000686972825156005,
+      "loss": 0.4141,
+      "step": 3165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006867845393941994,
+      "loss": 0.5618,
+      "step": 3166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006865962228442707,
+      "loss": 0.4912,
+      "step": 3167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006864078755372595,
+      "loss": 0.5521,
+      "step": 3168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006862194975042116,
+      "loss": 0.4097,
+      "step": 3169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006860310887761778,
+      "loss": 0.4249,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006858426493842138,
+      "loss": 0.6145,
+      "step": 3171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856541793593807,
+      "loss": 0.4894,
+      "step": 3172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006854656787327443,
+      "loss": 0.4871,
+      "step": 3173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852771475353754,
+      "loss": 0.4843,
+      "step": 3174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006850885857983504,
+      "loss": 0.4005,
+      "step": 3175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006848999935527503,
+      "loss": 0.574,
+      "step": 3176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006847113708296609,
+      "loss": 0.5104,
+      "step": 3177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006845227176601736,
+      "loss": 0.4404,
+      "step": 3178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006843340340753844,
+      "loss": 0.4087,
+      "step": 3179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006841453201063946,
+      "loss": 0.4865,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006839565757843101,
+      "loss": 0.465,
+      "step": 3181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006837678011402424,
+      "loss": 0.4873,
+      "step": 3182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006835789962053073,
+      "loss": 0.506,
+      "step": 3183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006833901610106261,
+      "loss": 0.483,
+      "step": 3184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006832012955873252,
+      "loss": 0.4149,
+      "step": 3185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006830123999665351,
+      "loss": 0.6279,
+      "step": 3186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006828234741793926,
+      "loss": 0.4915,
+      "step": 3187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006826345182570383,
+      "loss": 0.6515,
+      "step": 3188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006824455322306184,
+      "loss": 0.4299,
+      "step": 3189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006822565161312838,
+      "loss": 0.5299,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006820674699901905,
+      "loss": 0.4338,
+      "step": 3191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006818783938384996,
+      "loss": 0.3982,
+      "step": 3192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006816892877073767,
+      "loss": 0.511,
+      "step": 3193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006815001516279926,
+      "loss": 0.4132,
+      "step": 3194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006813109856315233,
+      "loss": 0.0558,
+      "step": 3195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000681121789749149,
+      "loss": 0.5714,
+      "step": 3196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006809325640120559,
+      "loss": 0.4336,
+      "step": 3197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000680743308451434,
+      "loss": 0.5911,
+      "step": 3198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006805540230984791,
+      "loss": 0.3863,
+      "step": 3199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006803647079843911,
+      "loss": 0.5356,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006801753631403757,
+      "loss": 0.5092,
+      "step": 3201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006799859885976427,
+      "loss": 0.5062,
+      "step": 3202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006797965843874075,
+      "loss": 0.4625,
+      "step": 3203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006796071505408896,
+      "loss": 0.0556,
+      "step": 3204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000679417687089314,
+      "loss": 0.4316,
+      "step": 3205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006792281940639107,
+      "loss": 0.5145,
+      "step": 3206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006790386714959136,
+      "loss": 0.4667,
+      "step": 3207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788491194165629,
+      "loss": 0.428,
+      "step": 3208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006786595378571022,
+      "loss": 0.5833,
+      "step": 3209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006784699268487812,
+      "loss": 0.4137,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006782802864228536,
+      "loss": 0.4517,
+      "step": 3211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006780906166105785,
+      "loss": 0.3928,
+      "step": 3212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006779009174432193,
+      "loss": 0.5813,
+      "step": 3213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006777111889520448,
+      "loss": 0.4353,
+      "step": 3214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006775214311683284,
+      "loss": 0.6316,
+      "step": 3215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773316441233482,
+      "loss": 0.5619,
+      "step": 3216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006771418278483871,
+      "loss": 0.5918,
+      "step": 3217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006769519823747332,
+      "loss": 0.5134,
+      "step": 3218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006767621077336791,
+      "loss": 0.5082,
+      "step": 3219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006765722039565222,
+      "loss": 0.504,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006763822710745648,
+      "loss": 0.4537,
+      "step": 3221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676192309119114,
+      "loss": 0.4485,
+      "step": 3222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006760023181214817,
+      "loss": 0.4778,
+      "step": 3223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006758122981129845,
+      "loss": 0.4108,
+      "step": 3224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006756222491249437,
+      "loss": 0.408,
+      "step": 3225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000675432171188686,
+      "loss": 0.4897,
+      "step": 3226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006752420643355417,
+      "loss": 0.4708,
+      "step": 3227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000675051928596847,
+      "loss": 0.4313,
+      "step": 3228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006748617640039422,
+      "loss": 0.5137,
+      "step": 3229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006746715705881728,
+      "loss": 0.4984,
+      "step": 3230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006744813483808884,
+      "loss": 0.4425,
+      "step": 3231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742910974134442,
+      "loss": 0.3873,
+      "step": 3232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006741008177171995,
+      "loss": 0.5165,
+      "step": 3233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006739105093235185,
+      "loss": 0.4704,
+      "step": 3234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006737201722637702,
+      "loss": 0.4056,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006735298065693285,
+      "loss": 0.4908,
+      "step": 3236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006733394122715716,
+      "loss": 0.3599,
+      "step": 3237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006731489894018826,
+      "loss": 0.4841,
+      "step": 3238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006729585379916494,
+      "loss": 0.4446,
+      "step": 3239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006727680580722645,
+      "loss": 0.5751,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006725775496751255,
+      "loss": 0.4424,
+      "step": 3241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006723870128316336,
+      "loss": 0.4658,
+      "step": 3242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000672196447573196,
+      "loss": 0.3973,
+      "step": 3243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006720058539312238,
+      "loss": 0.3688,
+      "step": 3244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006718152319371332,
+      "loss": 0.4988,
+      "step": 3245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006716245816223446,
+      "loss": 0.3846,
+      "step": 3246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006714339030182836,
+      "loss": 0.5234,
+      "step": 3247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006712431961563798,
+      "loss": 0.51,
+      "step": 3248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006710524610680682,
+      "loss": 0.4524,
+      "step": 3249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006708616977847879,
+      "loss": 0.4095,
+      "step": 3250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006706709063379829,
+      "loss": 0.3937,
+      "step": 3251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006704800867591018,
+      "loss": 0.3577,
+      "step": 3252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006702892390795978,
+      "loss": 0.551,
+      "step": 3253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006700983633309289,
+      "loss": 0.3654,
+      "step": 3254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699074595445574,
+      "loss": 0.4204,
+      "step": 3255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006697165277519505,
+      "loss": 0.4668,
+      "step": 3256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00066952556798458,
+      "loss": 0.4847,
+      "step": 3257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006693345802739222,
+      "loss": 0.4257,
+      "step": 3258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006691435646514578,
+      "loss": 0.5436,
+      "step": 3259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006689525211486726,
+      "loss": 0.4122,
+      "step": 3260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006687614497970566,
+      "loss": 0.5002,
+      "step": 3261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006685703506281046,
+      "loss": 0.4246,
+      "step": 3262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006683792236733162,
+      "loss": 0.4182,
+      "step": 3263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006681880689641947,
+      "loss": 0.432,
+      "step": 3264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000667996886532249,
+      "loss": 0.4385,
+      "step": 3265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006678056764089919,
+      "loss": 0.5302,
+      "step": 3266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006676144386259413,
+      "loss": 0.5336,
+      "step": 3267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000667423173214619,
+      "loss": 0.3923,
+      "step": 3268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000667231880206552,
+      "loss": 0.4578,
+      "step": 3269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006670405596332715,
+      "loss": 0.4417,
+      "step": 3270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006668492115263131,
+      "loss": 0.457,
+      "step": 3271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006666578359172175,
+      "loss": 0.4561,
+      "step": 3272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006664664328375292,
+      "loss": 0.4543,
+      "step": 3273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666275002318798,
+      "loss": 0.5151,
+      "step": 3274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006660835443925775,
+      "loss": 0.5303,
+      "step": 3275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006658920590904262,
+      "loss": 0.4299,
+      "step": 3276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006657005464439072,
+      "loss": 0.4343,
+      "step": 3277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000665509006484588,
+      "loss": 0.5367,
+      "step": 3278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006653174392440404,
+      "loss": 0.055,
+      "step": 3279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006651258447538409,
+      "loss": 0.3594,
+      "step": 3280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006649342230455706,
+      "loss": 0.3958,
+      "step": 3281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006647425741508146,
+      "loss": 0.5575,
+      "step": 3282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006645508981011633,
+      "loss": 0.473,
+      "step": 3283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006643591949282108,
+      "loss": 0.4332,
+      "step": 3284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000664167464663556,
+      "loss": 0.3605,
+      "step": 3285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006639757073388025,
+      "loss": 0.5111,
+      "step": 3286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006637839229855576,
+      "loss": 0.5255,
+      "step": 3287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006635921116354339,
+      "loss": 0.413,
+      "step": 3288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006634002733200482,
+      "loss": 0.467,
+      "step": 3289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006632084080710213,
+      "loss": 0.493,
+      "step": 3290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006630165159199793,
+      "loss": 0.3702,
+      "step": 3291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006628245968985515,
+      "loss": 0.5427,
+      "step": 3292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000662632651038373,
+      "loss": 0.6132,
+      "step": 3293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006624406783710823,
+      "loss": 0.5447,
+      "step": 3294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006622486789283229,
+      "loss": 0.45,
+      "step": 3295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006620566527417423,
+      "loss": 0.4304,
+      "step": 3296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006618645998429928,
+      "loss": 0.5883,
+      "step": 3297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006616725202637308,
+      "loss": 0.4528,
+      "step": 3298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006614804140356175,
+      "loss": 0.4027,
+      "step": 3299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006612882811903178,
+      "loss": 0.3896,
+      "step": 3300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006610961217595016,
+      "loss": 0.4103,
+      "step": 3301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006609039357748429,
+      "loss": 0.4773,
+      "step": 3302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006607117232680205,
+      "loss": 0.5546,
+      "step": 3303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006605194842707168,
+      "loss": 0.5343,
+      "step": 3304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000660327218814619,
+      "loss": 0.4438,
+      "step": 3305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006601349269314187,
+      "loss": 0.4945,
+      "step": 3306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006599426086528122,
+      "loss": 0.438,
+      "step": 3307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006597502640104993,
+      "loss": 0.4628,
+      "step": 3308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006595578930361849,
+      "loss": 0.4451,
+      "step": 3309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006593654957615778,
+      "loss": 0.516,
+      "step": 3310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006591730722183913,
+      "loss": 0.4183,
+      "step": 3311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000658980622438343,
+      "loss": 0.479,
+      "step": 3312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006587881464531547,
+      "loss": 0.5314,
+      "step": 3313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006585956442945531,
+      "loss": 0.5042,
+      "step": 3314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584031159942682,
+      "loss": 0.0543,
+      "step": 3315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006582105615840355,
+      "loss": 0.4598,
+      "step": 3316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006580179810955935,
+      "loss": 0.5111,
+      "step": 3317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006578253745606863,
+      "loss": 0.4779,
+      "step": 3318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006576327420110612,
+      "loss": 0.4559,
+      "step": 3319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006574400834784706,
+      "loss": 0.4376,
+      "step": 3320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006572473989946708,
+      "loss": 0.4218,
+      "step": 3321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006570546885914222,
+      "loss": 0.4483,
+      "step": 3322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006568619523004898,
+      "loss": 0.3916,
+      "step": 3323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006566691901536427,
+      "loss": 0.4207,
+      "step": 3324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006564764021826548,
+      "loss": 0.5319,
+      "step": 3325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006562835884193032,
+      "loss": 0.4363,
+      "step": 3326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00065609074889537,
+      "loss": 0.4553,
+      "step": 3327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006558978836426415,
+      "loss": 0.4523,
+      "step": 3328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006557049926929081,
+      "loss": 0.5221,
+      "step": 3329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006555120760779644,
+      "loss": 0.3722,
+      "step": 3330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006553191338296092,
+      "loss": 0.4218,
+      "step": 3331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006551261659796457,
+      "loss": 0.3973,
+      "step": 3332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006549331725598813,
+      "loss": 0.5036,
+      "step": 3333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006547401536021276,
+      "loss": 0.4535,
+      "step": 3334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006545471091382,
+      "loss": 0.5526,
+      "step": 3335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006543540391999189,
+      "loss": 0.4808,
+      "step": 3336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000654160943819108,
+      "loss": 0.4221,
+      "step": 3337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006539678230275961,
+      "loss": 0.4384,
+      "step": 3338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006537746768572155,
+      "loss": 0.5581,
+      "step": 3339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006535815053398031,
+      "loss": 0.5282,
+      "step": 3340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006533883085071997,
+      "loss": 0.0534,
+      "step": 3341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006531950863912501,
+      "loss": 0.4745,
+      "step": 3342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000653001839023804,
+      "loss": 0.5499,
+      "step": 3343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006528085664367146,
+      "loss": 0.0529,
+      "step": 3344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006526152686618394,
+      "loss": 0.4709,
+      "step": 3345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006524219457310403,
+      "loss": 0.4543,
+      "step": 3346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000652228597676183,
+      "loss": 0.4661,
+      "step": 3347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006520352245291375,
+      "loss": 0.0523,
+      "step": 3348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000651841826321778,
+      "loss": 0.4612,
+      "step": 3349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006516484030859828,
+      "loss": 0.4093,
+      "step": 3350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006514549548536343,
+      "loss": 0.4061,
+      "step": 3351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006512614816566189,
+      "loss": 0.481,
+      "step": 3352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006510679835268273,
+      "loss": 0.5049,
+      "step": 3353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006508744604961543,
+      "loss": 0.4059,
+      "step": 3354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006506809125964985,
+      "loss": 0.4357,
+      "step": 3355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006504873398597633,
+      "loss": 0.4073,
+      "step": 3356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006502937423178552,
+      "loss": 0.4473,
+      "step": 3357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006501001200026858,
+      "loss": 0.4058,
+      "step": 3358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00064990647294617,
+      "loss": 0.4495,
+      "step": 3359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006497128011802272,
+      "loss": 0.4862,
+      "step": 3360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006495191047367806,
+      "loss": 0.5468,
+      "step": 3361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006493253836477581,
+      "loss": 0.4084,
+      "step": 3362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006491316379450906,
+      "loss": 0.4973,
+      "step": 3363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006489378676607139,
+      "loss": 0.446,
+      "step": 3364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006487440728265677,
+      "loss": 0.4978,
+      "step": 3365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006485502534745956,
+      "loss": 0.4411,
+      "step": 3366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006483564096367451,
+      "loss": 0.5183,
+      "step": 3367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006481625413449681,
+      "loss": 0.4491,
+      "step": 3368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006479686486312205,
+      "loss": 0.4376,
+      "step": 3369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006477747315274618,
+      "loss": 0.4346,
+      "step": 3370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006475807900656558,
+      "loss": 0.4584,
+      "step": 3371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006473868242777705,
+      "loss": 0.5624,
+      "step": 3372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006471928341957777,
+      "loss": 0.4204,
+      "step": 3373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006469988198516531,
+      "loss": 0.4426,
+      "step": 3374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006468047812773767,
+      "loss": 0.4198,
+      "step": 3375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006466107185049321,
+      "loss": 0.5123,
+      "step": 3376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006464166315663074,
+      "loss": 0.4422,
+      "step": 3377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006462225204934939,
+      "loss": 0.4185,
+      "step": 3378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006460283853184879,
+      "loss": 0.53,
+      "step": 3379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000645834226073289,
+      "loss": 0.5784,
+      "step": 3380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006456400427899008,
+      "loss": 0.5787,
+      "step": 3381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000645445835500331,
+      "loss": 0.4372,
+      "step": 3382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006452516042365912,
+      "loss": 0.4686,
+      "step": 3383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006450573490306972,
+      "loss": 0.4843,
+      "step": 3384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006448630699146683,
+      "loss": 0.4857,
+      "step": 3385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006446687669205281,
+      "loss": 0.4459,
+      "step": 3386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006444744400803039,
+      "loss": 0.0521,
+      "step": 3387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006442800894260272,
+      "loss": 0.4358,
+      "step": 3388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000644085714989733,
+      "loss": 0.4929,
+      "step": 3389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006438913168034607,
+      "loss": 0.4601,
+      "step": 3390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006436968948992535,
+      "loss": 0.3993,
+      "step": 3391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006435024493091579,
+      "loss": 0.49,
+      "step": 3392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006433079800652253,
+      "loss": 0.4543,
+      "step": 3393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006431134871995105,
+      "loss": 0.0515,
+      "step": 3394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006429189707440721,
+      "loss": 0.4369,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006427244307309725,
+      "loss": 0.3679,
+      "step": 3396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425298671922785,
+      "loss": 0.4846,
+      "step": 3397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006423352801600601,
+      "loss": 0.5635,
+      "step": 3398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006421406696663918,
+      "loss": 0.4667,
+      "step": 3399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006419460357433518,
+      "loss": 0.5291,
+      "step": 3400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006417513784230215,
+      "loss": 0.4637,
+      "step": 3401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006415566977374874,
+      "loss": 0.5769,
+      "step": 3402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006413619937188387,
+      "loss": 0.5442,
+      "step": 3403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641167266399169,
+      "loss": 0.4349,
+      "step": 3404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006409725158105757,
+      "loss": 0.4397,
+      "step": 3405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006407777419851602,
+      "loss": 0.3615,
+      "step": 3406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006405829449550271,
+      "loss": 0.5513,
+      "step": 3407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006403881247522855,
+      "loss": 0.4985,
+      "step": 3408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006401932814090481,
+      "loss": 0.4934,
+      "step": 3409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006399984149574311,
+      "loss": 0.4221,
+      "step": 3410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000639803525429555,
+      "loss": 0.6077,
+      "step": 3411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006396086128575437,
+      "loss": 0.515,
+      "step": 3412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006394136772735256,
+      "loss": 0.4962,
+      "step": 3413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006392187187096317,
+      "loss": 0.4609,
+      "step": 3414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000639023737197998,
+      "loss": 0.4026,
+      "step": 3415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006388287327707635,
+      "loss": 0.4151,
+      "step": 3416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000638633705460071,
+      "loss": 0.5134,
+      "step": 3417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006384386552980678,
+      "loss": 0.5349,
+      "step": 3418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006382435823169041,
+      "loss": 0.4504,
+      "step": 3419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380484865487346,
+      "loss": 0.478,
+      "step": 3420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006378533680257169,
+      "loss": 0.4412,
+      "step": 3421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000637658226780013,
+      "loss": 0.4875,
+      "step": 3422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006374630628437886,
+      "loss": 0.4348,
+      "step": 3423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006372678762492129,
+      "loss": 0.4484,
+      "step": 3424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000637072667028459,
+      "loss": 0.3596,
+      "step": 3425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006368774352137037,
+      "loss": 0.4462,
+      "step": 3426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006366821808371276,
+      "loss": 0.541,
+      "step": 3427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006364869039309147,
+      "loss": 0.4268,
+      "step": 3428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000636291604527253,
+      "loss": 0.4573,
+      "step": 3429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006360962826583343,
+      "loss": 0.5615,
+      "step": 3430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006359009383563539,
+      "loss": 0.4586,
+      "step": 3431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006357055716535107,
+      "loss": 0.0517,
+      "step": 3432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006355101825820075,
+      "loss": 0.5156,
+      "step": 3433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000635314771174051,
+      "loss": 0.4406,
+      "step": 3434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006351193374618511,
+      "loss": 0.4038,
+      "step": 3435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006349238814776213,
+      "loss": 0.476,
+      "step": 3436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006347284032535794,
+      "loss": 0.4114,
+      "step": 3437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006345329028219467,
+      "loss": 0.3976,
+      "step": 3438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006343373802149474,
+      "loss": 0.4379,
+      "step": 3439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006341418354648106,
+      "loss": 0.4293,
+      "step": 3440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006339462686037677,
+      "loss": 0.5557,
+      "step": 3441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000633750679664055,
+      "loss": 0.4261,
+      "step": 3442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006335550686779117,
+      "loss": 0.4159,
+      "step": 3443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006333594356775807,
+      "loss": 0.3489,
+      "step": 3444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006331637806953086,
+      "loss": 0.4963,
+      "step": 3445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006329681037633459,
+      "loss": 0.4792,
+      "step": 3446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006327724049139464,
+      "loss": 0.4897,
+      "step": 3447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006325766841793674,
+      "loss": 0.4657,
+      "step": 3448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006323809415918702,
+      "loss": 0.494,
+      "step": 3449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006321851771837194,
+      "loss": 0.4066,
+      "step": 3450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006319893909871833,
+      "loss": 0.5048,
+      "step": 3451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006317935830345338,
+      "loss": 0.4167,
+      "step": 3452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006315977533580464,
+      "loss": 0.4957,
+      "step": 3453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006314019019900002,
+      "loss": 0.5038,
+      "step": 3454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006312060289626778,
+      "loss": 0.3842,
+      "step": 3455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006310101343083652,
+      "loss": 0.5164,
+      "step": 3456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006308142180593524,
+      "loss": 0.4259,
+      "step": 3457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006306182802479328,
+      "loss": 0.0515,
+      "step": 3458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006304223209064029,
+      "loss": 0.5018,
+      "step": 3459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006302263400670636,
+      "loss": 0.5249,
+      "step": 3460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006300303377622184,
+      "loss": 0.3705,
+      "step": 3461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006298343140241753,
+      "loss": 0.5259,
+      "step": 3462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006296382688852448,
+      "loss": 0.5206,
+      "step": 3463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000629442202377742,
+      "loss": 0.3622,
+      "step": 3464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006292461145339845,
+      "loss": 0.4548,
+      "step": 3465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006290500053862942,
+      "loss": 0.3765,
+      "step": 3466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006288538749669961,
+      "loss": 0.052,
+      "step": 3467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000628657723308419,
+      "loss": 0.3809,
+      "step": 3468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000628461550442895,
+      "loss": 0.4133,
+      "step": 3469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006282653564027595,
+      "loss": 0.4406,
+      "step": 3470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006280691412203519,
+      "loss": 0.3955,
+      "step": 3471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006278729049280145,
+      "loss": 0.4633,
+      "step": 3472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006276766475580935,
+      "loss": 0.4294,
+      "step": 3473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006274803691429386,
+      "loss": 0.4392,
+      "step": 3474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006272840697149027,
+      "loss": 0.4158,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006270877493063423,
+      "loss": 0.0523,
+      "step": 3476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006268914079496172,
+      "loss": 0.4792,
+      "step": 3477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006266950456770909,
+      "loss": 0.4751,
+      "step": 3478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264986625211304,
+      "loss": 0.3517,
+      "step": 3479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006263022585141059,
+      "loss": 0.4158,
+      "step": 3480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000626105833688391,
+      "loss": 0.5149,
+      "step": 3481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006259093880763628,
+      "loss": 0.45,
+      "step": 3482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006257129217104023,
+      "loss": 0.5421,
+      "step": 3483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006255164346228929,
+      "loss": 0.0522,
+      "step": 3484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006253199268462224,
+      "loss": 0.4565,
+      "step": 3485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006251233984127816,
+      "loss": 0.4192,
+      "step": 3486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006249268493549647,
+      "loss": 0.605,
+      "step": 3487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006247302797051693,
+      "loss": 0.0522,
+      "step": 3488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006245336894957962,
+      "loss": 0.384,
+      "step": 3489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006243370787592504,
+      "loss": 0.4036,
+      "step": 3490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006241404475279392,
+      "loss": 0.5714,
+      "step": 3491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006239437958342739,
+      "loss": 0.4623,
+      "step": 3492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006237471237106691,
+      "loss": 0.4395,
+      "step": 3493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006235504311895427,
+      "loss": 0.5295,
+      "step": 3494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623353718303316,
+      "loss": 0.3857,
+      "step": 3495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006231569850844134,
+      "loss": 0.4679,
+      "step": 3496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006229602315652634,
+      "loss": 0.4023,
+      "step": 3497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006227634577782967,
+      "loss": 0.3988,
+      "step": 3498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006225666637559485,
+      "loss": 0.3679,
+      "step": 3499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006223698495306565,
+      "loss": 0.4902,
+      "step": 3500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006221730151348624,
+      "loss": 0.4203,
+      "step": 3501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006219761606010104,
+      "loss": 0.5994,
+      "step": 3502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006217792859615487,
+      "loss": 0.5059,
+      "step": 3503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006215823912489285,
+      "loss": 0.4446,
+      "step": 3504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006213854764956047,
+      "loss": 0.3754,
+      "step": 3505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000621188541734035,
+      "loss": 0.4005,
+      "step": 3506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006209915869966804,
+      "loss": 0.4612,
+      "step": 3507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006207946123160058,
+      "loss": 0.5798,
+      "step": 3508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006205976177244786,
+      "loss": 0.4308,
+      "step": 3509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006204006032545702,
+      "loss": 0.5651,
+      "step": 3510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006202035689387549,
+      "loss": 0.4595,
+      "step": 3511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006200065148095101,
+      "loss": 0.4971,
+      "step": 3512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006198094408993168,
+      "loss": 0.4939,
+      "step": 3513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006196123472406591,
+      "loss": 0.3944,
+      "step": 3514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006194152338660247,
+      "loss": 0.3603,
+      "step": 3515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006192181008079038,
+      "loss": 0.4784,
+      "step": 3516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006190209480987906,
+      "loss": 0.6024,
+      "step": 3517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000618823775771182,
+      "loss": 0.4645,
+      "step": 3518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006186265838575787,
+      "loss": 0.3779,
+      "step": 3519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006184293723904839,
+      "loss": 0.4036,
+      "step": 3520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006182321414024047,
+      "loss": 0.3548,
+      "step": 3521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006180348909258509,
+      "loss": 0.4364,
+      "step": 3522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006178376209933362,
+      "loss": 0.3812,
+      "step": 3523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006176403316373766,
+      "loss": 0.4648,
+      "step": 3524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006174430228904919,
+      "loss": 0.4833,
+      "step": 3525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006172456947852048,
+      "loss": 0.4031,
+      "step": 3526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006170483473540418,
+      "loss": 0.4552,
+      "step": 3527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006168509806295316,
+      "loss": 0.4247,
+      "step": 3528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006166535946442069,
+      "loss": 0.4506,
+      "step": 3529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006164561894306032,
+      "loss": 0.3901,
+      "step": 3530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006162587650212593,
+      "loss": 0.5596,
+      "step": 3531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000616061321448717,
+      "loss": 0.4987,
+      "step": 3532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006158638587455215,
+      "loss": 0.4979,
+      "step": 3533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156663769442211,
+      "loss": 0.4398,
+      "step": 3534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000615468876077367,
+      "loss": 0.4548,
+      "step": 3535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006152713561775137,
+      "loss": 0.5337,
+      "step": 3536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006150738172772193,
+      "loss": 0.5013,
+      "step": 3537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006148762594090438,
+      "loss": 0.4681,
+      "step": 3538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006146786826055517,
+      "loss": 0.4546,
+      "step": 3539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00061448108689931,
+      "loss": 0.5042,
+      "step": 3540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006142834723228886,
+      "loss": 0.5051,
+      "step": 3541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006140858389088609,
+      "loss": 0.5253,
+      "step": 3542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006138881866898034,
+      "loss": 0.3909,
+      "step": 3543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006136905156982954,
+      "loss": 0.4989,
+      "step": 3544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006134928259669194,
+      "loss": 0.3978,
+      "step": 3545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006132951175282613,
+      "loss": 0.4673,
+      "step": 3546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006130973904149098,
+      "loss": 0.4227,
+      "step": 3547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006128996446594563,
+      "loss": 0.3995,
+      "step": 3548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006127018802944962,
+      "loss": 0.5176,
+      "step": 3549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006125040973526271,
+      "loss": 0.5421,
+      "step": 3550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006123062958664501,
+      "loss": 0.4076,
+      "step": 3551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121084758685696,
+      "loss": 0.346,
+      "step": 3552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006119106373915922,
+      "loss": 0.3738,
+      "step": 3553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006117127804681284,
+      "loss": 0.3701,
+      "step": 3554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006115149051307913,
+      "loss": 0.427,
+      "step": 3555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006113170114121974,
+      "loss": 0.4146,
+      "step": 3556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006111190993449653,
+      "loss": 0.3938,
+      "step": 3557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000610921168961718,
+      "loss": 0.4789,
+      "step": 3558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006107232202950807,
+      "loss": 0.3907,
+      "step": 3559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006105252533776814,
+      "loss": 0.3909,
+      "step": 3560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006103272682421516,
+      "loss": 0.5017,
+      "step": 3561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006101292649211258,
+      "loss": 0.4868,
+      "step": 3562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099312434472412,
+      "loss": 0.4462,
+      "step": 3563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006097332038531381,
+      "loss": 0.4572,
+      "step": 3564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006095351461714598,
+      "loss": 0.5438,
+      "step": 3565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006093370704348529,
+      "loss": 0.4484,
+      "step": 3566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006091389766759663,
+      "loss": 0.5358,
+      "step": 3567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006089408649274524,
+      "loss": 0.5078,
+      "step": 3568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006087427352219665,
+      "loss": 0.4672,
+      "step": 3569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006085445875921667,
+      "loss": 0.5255,
+      "step": 3570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000608346422070714,
+      "loss": 0.4213,
+      "step": 3571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006081482386902726,
+      "loss": 0.492,
+      "step": 3572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006079500374835094,
+      "loss": 0.0514,
+      "step": 3573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006077518184830947,
+      "loss": 0.5443,
+      "step": 3574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006075535817217009,
+      "loss": 0.4426,
+      "step": 3575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006073553272320043,
+      "loss": 0.5104,
+      "step": 3576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006071570550466833,
+      "loss": 0.42,
+      "step": 3577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006069587651984197,
+      "loss": 0.4501,
+      "step": 3578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000606760457719898,
+      "loss": 0.0514,
+      "step": 3579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006065621326438058,
+      "loss": 0.4767,
+      "step": 3580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006063637900028334,
+      "loss": 0.4106,
+      "step": 3581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000606165429829674,
+      "loss": 0.4612,
+      "step": 3582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006059670521570238,
+      "loss": 0.4183,
+      "step": 3583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000605768657017582,
+      "loss": 0.3916,
+      "step": 3584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006055702444440503,
+      "loss": 0.4816,
+      "step": 3585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006053718144691336,
+      "loss": 0.4684,
+      "step": 3586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006051733671255396,
+      "loss": 0.4429,
+      "step": 3587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006049749024459788,
+      "loss": 0.45,
+      "step": 3588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006047764204631646,
+      "loss": 0.4641,
+      "step": 3589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006045779212098132,
+      "loss": 0.0508,
+      "step": 3590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006043794047186437,
+      "loss": 0.4674,
+      "step": 3591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006041808710223782,
+      "loss": 0.5078,
+      "step": 3592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006039823201537411,
+      "loss": 0.3979,
+      "step": 3593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006037837521454602,
+      "loss": 0.4793,
+      "step": 3594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006035851670302659,
+      "loss": 0.4398,
+      "step": 3595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006033865648408915,
+      "loss": 0.4634,
+      "step": 3596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006031879456100729,
+      "loss": 0.4617,
+      "step": 3597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006029893093705492,
+      "loss": 0.4891,
+      "step": 3598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000602790656155062,
+      "loss": 0.5173,
+      "step": 3599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006025919859963556,
+      "loss": 0.4294,
+      "step": 3600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006023932989271772,
+      "loss": 0.4783,
+      "step": 3601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006021945949802772,
+      "loss": 0.4617,
+      "step": 3602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006019958741884082,
+      "loss": 0.3911,
+      "step": 3603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006017971365843258,
+      "loss": 0.4591,
+      "step": 3604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006015983822007883,
+      "loss": 0.0503,
+      "step": 3605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006013996110705571,
+      "loss": 0.4471,
+      "step": 3606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006012008232263959,
+      "loss": 0.4021,
+      "step": 3607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006010020187010713,
+      "loss": 0.4592,
+      "step": 3608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000600803197527353,
+      "loss": 0.5123,
+      "step": 3609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006006043597380127,
+      "loss": 0.4523,
+      "step": 3610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006004055053658258,
+      "loss": 0.408,
+      "step": 3611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006002066344435695,
+      "loss": 0.4464,
+      "step": 3612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006000077470040243,
+      "loss": 0.465,
+      "step": 3613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005998088430799734,
+      "loss": 0.4557,
+      "step": 3614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005996099227042024,
+      "loss": 0.4939,
+      "step": 3615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005994109859094999,
+      "loss": 0.4608,
+      "step": 3616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000599212032728657,
+      "loss": 0.4707,
+      "step": 3617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005990130631944677,
+      "loss": 0.4723,
+      "step": 3618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005988140773397286,
+      "loss": 0.5471,
+      "step": 3619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000598615075197239,
+      "loss": 0.0511,
+      "step": 3620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005984160567998009,
+      "loss": 0.4795,
+      "step": 3621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005982170221802189,
+      "loss": 0.4675,
+      "step": 3622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005980179713713003,
+      "loss": 0.5118,
+      "step": 3623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005978189044058552,
+      "loss": 0.4874,
+      "step": 3624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005976198213166963,
+      "loss": 0.5117,
+      "step": 3625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005974207221366387,
+      "loss": 0.5604,
+      "step": 3626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005972216068985004,
+      "loss": 0.4486,
+      "step": 3627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005970224756351023,
+      "loss": 0.0515,
+      "step": 3628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005968233283792676,
+      "loss": 0.528,
+      "step": 3629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005966241651638219,
+      "loss": 0.3485,
+      "step": 3630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000596424986021594,
+      "loss": 0.4216,
+      "step": 3631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000596225790985415,
+      "loss": 0.5024,
+      "step": 3632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005960265800881188,
+      "loss": 0.4877,
+      "step": 3633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005958273533625413,
+      "loss": 0.4324,
+      "step": 3634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005956281108415219,
+      "loss": 0.4727,
+      "step": 3635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005954288525579024,
+      "loss": 0.4089,
+      "step": 3636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005952295785445264,
+      "loss": 0.4807,
+      "step": 3637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005950302888342411,
+      "loss": 0.5919,
+      "step": 3638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000594830983459896,
+      "loss": 0.3964,
+      "step": 3639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005946316624543428,
+      "loss": 0.4207,
+      "step": 3640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005944323258504359,
+      "loss": 0.4889,
+      "step": 3641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005942329736810328,
+      "loss": 0.4288,
+      "step": 3642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000594033605978993,
+      "loss": 0.4674,
+      "step": 3643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005938342227771785,
+      "loss": 0.4484,
+      "step": 3644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005936348241084544,
+      "loss": 0.4313,
+      "step": 3645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005934354100056882,
+      "loss": 0.4083,
+      "step": 3646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005932359805017494,
+      "loss": 0.4481,
+      "step": 3647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005930365356295103,
+      "loss": 0.3972,
+      "step": 3648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005928370754218463,
+      "loss": 0.4562,
+      "step": 3649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926375999116348,
+      "loss": 0.4537,
+      "step": 3650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005924381091317555,
+      "loss": 0.3794,
+      "step": 3651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005922386031150912,
+      "loss": 0.4315,
+      "step": 3652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005920390818945269,
+      "loss": 0.4564,
+      "step": 3653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005918395455029501,
+      "loss": 0.4467,
+      "step": 3654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005916399939732507,
+      "loss": 0.4464,
+      "step": 3655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005914404273383215,
+      "loss": 0.5427,
+      "step": 3656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005912408456310573,
+      "loss": 0.3644,
+      "step": 3657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910412488843558,
+      "loss": 0.4294,
+      "step": 3658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005908416371311167,
+      "loss": 0.4689,
+      "step": 3659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005906420104042427,
+      "loss": 0.4868,
+      "step": 3660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005904423687366387,
+      "loss": 0.5216,
+      "step": 3661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005902427121612122,
+      "loss": 0.452,
+      "step": 3662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005900430407108728,
+      "loss": 0.4596,
+      "step": 3663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005898433544185329,
+      "loss": 0.4369,
+      "step": 3664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005896436533171075,
+      "loss": 0.4203,
+      "step": 3665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005894439374395133,
+      "loss": 0.4876,
+      "step": 3666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005892442068186702,
+      "loss": 0.464,
+      "step": 3667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890444614875004,
+      "loss": 0.0515,
+      "step": 3668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005888447014789282,
+      "loss": 0.4662,
+      "step": 3669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005886449268258803,
+      "loss": 0.4216,
+      "step": 3670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005884451375612865,
+      "loss": 0.5159,
+      "step": 3671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005882453337180782,
+      "loss": 0.051,
+      "step": 3672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005880455153291895,
+      "loss": 0.4333,
+      "step": 3673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005878456824275571,
+      "loss": 0.5631,
+      "step": 3674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005876458350461198,
+      "loss": 0.4799,
+      "step": 3675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587445973217819,
+      "loss": 0.4008,
+      "step": 3676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005872460969755982,
+      "loss": 0.4434,
+      "step": 3677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005870462063524035,
+      "loss": 0.3988,
+      "step": 3678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005868463013811834,
+      "loss": 0.4633,
+      "step": 3679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005866463820948887,
+      "loss": 0.3757,
+      "step": 3680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005864464485264727,
+      "loss": 0.3871,
+      "step": 3681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005862465007088905,
+      "loss": 0.0502,
+      "step": 3682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005860465386751004,
+      "loss": 0.4672,
+      "step": 3683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005858465624580623,
+      "loss": 0.4103,
+      "step": 3684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005856465720907388,
+      "loss": 0.4334,
+      "step": 3685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005854465676060948,
+      "loss": 0.424,
+      "step": 3686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005852465490370978,
+      "loss": 0.4795,
+      "step": 3687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005850465164167167,
+      "loss": 0.0499,
+      "step": 3688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005848464697779238,
+      "loss": 0.4666,
+      "step": 3689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005846464091536929,
+      "loss": 0.4233,
+      "step": 3690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000584446334577001,
+      "loss": 0.463,
+      "step": 3691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005842462460808263,
+      "loss": 0.4562,
+      "step": 3692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00058404614369815,
+      "loss": 0.4438,
+      "step": 3693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005838460274619557,
+      "loss": 0.4418,
+      "step": 3694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005836458974052285,
+      "loss": 0.3728,
+      "step": 3695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005834457535609569,
+      "loss": 0.4492,
+      "step": 3696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005832455959621307,
+      "loss": 0.4243,
+      "step": 3697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005830454246417424,
+      "loss": 0.3816,
+      "step": 3698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005828452396327866,
+      "loss": 0.3615,
+      "step": 3699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005826450409682603,
+      "loss": 0.4357,
+      "step": 3700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005824448286811627,
+      "loss": 0.4347,
+      "step": 3701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005822446028044957,
+      "loss": 0.3944,
+      "step": 3702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005820443633712622,
+      "loss": 0.4084,
+      "step": 3703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005818441104144686,
+      "loss": 0.396,
+      "step": 3704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005816438439671231,
+      "loss": 0.4061,
+      "step": 3705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005814435640622359,
+      "loss": 0.4638,
+      "step": 3706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005812432707328195,
+      "loss": 0.4405,
+      "step": 3707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005810429640118889,
+      "loss": 0.4379,
+      "step": 3708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005808426439324612,
+      "loss": 0.3826,
+      "step": 3709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005806423105275553,
+      "loss": 0.3771,
+      "step": 3710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000580441963830193,
+      "loss": 0.4566,
+      "step": 3711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005802416038733976,
+      "loss": 0.5356,
+      "step": 3712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000580041230690195,
+      "loss": 0.4625,
+      "step": 3713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000579840844313613,
+      "loss": 0.5446,
+      "step": 3714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005796404447766823,
+      "loss": 0.4772,
+      "step": 3715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005794400321124344,
+      "loss": 0.3842,
+      "step": 3716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005792396063539045,
+      "loss": 0.4675,
+      "step": 3717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005790391675341288,
+      "loss": 0.4105,
+      "step": 3718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005788387156861462,
+      "loss": 0.3683,
+      "step": 3719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005786382508429978,
+      "loss": 0.0499,
+      "step": 3720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005784377730377264,
+      "loss": 0.5044,
+      "step": 3721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005782372823033774,
+      "loss": 0.4636,
+      "step": 3722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000578036778672998,
+      "loss": 0.5092,
+      "step": 3723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005778362621796379,
+      "loss": 0.4972,
+      "step": 3724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005776357328563485,
+      "loss": 0.4639,
+      "step": 3725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005774351907361836,
+      "loss": 0.4365,
+      "step": 3726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005772346358521992,
+      "loss": 0.3549,
+      "step": 3727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005770340682374526,
+      "loss": 0.416,
+      "step": 3728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005768334879250046,
+      "loss": 0.5254,
+      "step": 3729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005766328949479167,
+      "loss": 0.4351,
+      "step": 3730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005764322893392535,
+      "loss": 0.415,
+      "step": 3731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000576231671132081,
+      "loss": 0.4569,
+      "step": 3732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005760310403594678,
+      "loss": 0.5178,
+      "step": 3733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005758303970544841,
+      "loss": 0.4244,
+      "step": 3734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005756297412502025,
+      "loss": 0.5043,
+      "step": 3735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005754290729796976,
+      "loss": 0.4819,
+      "step": 3736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005752283922760461,
+      "loss": 0.491,
+      "step": 3737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005750276991723264,
+      "loss": 0.4943,
+      "step": 3738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005748269937016194,
+      "loss": 0.0471,
+      "step": 3739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005746262758970077,
+      "loss": 0.4097,
+      "step": 3740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005744255457915761,
+      "loss": 0.4397,
+      "step": 3741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005742248034184117,
+      "loss": 0.4563,
+      "step": 3742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005740240488106029,
+      "loss": 0.4814,
+      "step": 3743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005738232820012407,
+      "loss": 0.5345,
+      "step": 3744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005736225030234179,
+      "loss": 0.0511,
+      "step": 3745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005734217119102296,
+      "loss": 0.4817,
+      "step": 3746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005732209086947725,
+      "loss": 0.0517,
+      "step": 3747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005730200934101452,
+      "loss": 0.5179,
+      "step": 3748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005728192660894491,
+      "loss": 0.4929,
+      "step": 3749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005726184267657866,
+      "loss": 0.484,
+      "step": 3750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005724175754722626,
+      "loss": 0.3831,
+      "step": 3751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005722167122419839,
+      "loss": 0.5156,
+      "step": 3752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005720158371080592,
+      "loss": 0.5174,
+      "step": 3753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005718149501035994,
+      "loss": 0.5947,
+      "step": 3754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005716140512617168,
+      "loss": 0.459,
+      "step": 3755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005714131406155264,
+      "loss": 0.3953,
+      "step": 3756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005712122181981446,
+      "loss": 0.432,
+      "step": 3757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005710112840426897,
+      "loss": 0.3881,
+      "step": 3758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005708103381822827,
+      "loss": 0.4147,
+      "step": 3759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005706093806500453,
+      "loss": 0.4082,
+      "step": 3760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005704084114791023,
+      "loss": 0.4329,
+      "step": 3761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005702074307025794,
+      "loss": 0.5056,
+      "step": 3762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005700064383536052,
+      "loss": 0.4132,
+      "step": 3763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005698054344653096,
+      "loss": 0.4363,
+      "step": 3764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005696044190708244,
+      "loss": 0.4618,
+      "step": 3765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005694033922032835,
+      "loss": 0.5989,
+      "step": 3766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005692023538958226,
+      "loss": 0.4408,
+      "step": 3767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005690013041815793,
+      "loss": 0.5242,
+      "step": 3768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005688002430936932,
+      "loss": 0.6178,
+      "step": 3769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005685991706653055,
+      "loss": 0.5143,
+      "step": 3770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683980869295597,
+      "loss": 0.4586,
+      "step": 3771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005681969919196005,
+      "loss": 0.4258,
+      "step": 3772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005679958856685751,
+      "loss": 0.3669,
+      "step": 3773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005677947682096322,
+      "loss": 0.4421,
+      "step": 3774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005675936395759227,
+      "loss": 0.0517,
+      "step": 3775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005673924998005986,
+      "loss": 0.0518,
+      "step": 3776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005671913489168147,
+      "loss": 0.4532,
+      "step": 3777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005669901869577269,
+      "loss": 0.4504,
+      "step": 3778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005667890139564935,
+      "loss": 0.4708,
+      "step": 3779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005665878299462738,
+      "loss": 0.4432,
+      "step": 3780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005663866349602298,
+      "loss": 0.4609,
+      "step": 3781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005661854290315248,
+      "loss": 0.4225,
+      "step": 3782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005659842121933241,
+      "loss": 0.3964,
+      "step": 3783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005657829844787946,
+      "loss": 0.4099,
+      "step": 3784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005655817459211053,
+      "loss": 0.4019,
+      "step": 3785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005653804965534267,
+      "loss": 0.478,
+      "step": 3786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005651792364089312,
+      "loss": 0.3732,
+      "step": 3787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005649779655207929,
+      "loss": 0.4238,
+      "step": 3788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000564776683922188,
+      "loss": 0.4162,
+      "step": 3789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005645753916462942,
+      "loss": 0.4778,
+      "step": 3790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005643740887262905,
+      "loss": 0.4871,
+      "step": 3791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005641727751953584,
+      "loss": 0.4658,
+      "step": 3792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005639714510866812,
+      "loss": 0.358,
+      "step": 3793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000563770116433443,
+      "loss": 0.4321,
+      "step": 3794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005635687712688307,
+      "loss": 0.4073,
+      "step": 3795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005633674156260322,
+      "loss": 0.3809,
+      "step": 3796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005631660495382378,
+      "loss": 0.4349,
+      "step": 3797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005629646730386388,
+      "loss": 0.5223,
+      "step": 3798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005627632861604286,
+      "loss": 0.4431,
+      "step": 3799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005625618889368024,
+      "loss": 0.4281,
+      "step": 3800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005623604814009568,
+      "loss": 0.5143,
+      "step": 3801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005621590635860904,
+      "loss": 0.4979,
+      "step": 3802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005619576355254031,
+      "loss": 0.4072,
+      "step": 3803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005617561972520971,
+      "loss": 0.483,
+      "step": 3804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005615547487993757,
+      "loss": 0.4292,
+      "step": 3805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005613532902004442,
+      "loss": 0.6125,
+      "step": 3806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005611518214885096,
+      "loss": 0.4177,
+      "step": 3807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005609503426967802,
+      "loss": 0.4568,
+      "step": 3808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005607488538584662,
+      "loss": 0.5057,
+      "step": 3809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005605473550067797,
+      "loss": 0.4717,
+      "step": 3810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005603458461749341,
+      "loss": 0.4915,
+      "step": 3811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005601443273961445,
+      "loss": 0.4663,
+      "step": 3812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005599427987036279,
+      "loss": 0.3921,
+      "step": 3813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005597412601306025,
+      "loss": 0.4044,
+      "step": 3814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005595397117102885,
+      "loss": 0.4961,
+      "step": 3815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005593381534759076,
+      "loss": 0.4904,
+      "step": 3816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005591365854606829,
+      "loss": 0.4965,
+      "step": 3817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005589350076978395,
+      "loss": 0.5131,
+      "step": 3818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005587334202206041,
+      "loss": 0.0502,
+      "step": 3819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005585318230622045,
+      "loss": 0.5228,
+      "step": 3820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005583302162558706,
+      "loss": 0.4066,
+      "step": 3821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005581285998348337,
+      "loss": 0.5219,
+      "step": 3822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005579269738323266,
+      "loss": 0.3889,
+      "step": 3823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005577253382815838,
+      "loss": 0.4127,
+      "step": 3824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005575236932158413,
+      "loss": 0.4491,
+      "step": 3825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005573220386683369,
+      "loss": 0.4841,
+      "step": 3826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005571203746723097,
+      "loss": 0.3951,
+      "step": 3827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005569187012610003,
+      "loss": 0.4459,
+      "step": 3828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005567170184676512,
+      "loss": 0.3948,
+      "step": 3829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005565153263255062,
+      "loss": 0.4193,
+      "step": 3830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005563136248678105,
+      "loss": 0.4281,
+      "step": 3831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005561119141278112,
+      "loss": 0.3798,
+      "step": 3832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005559101941387567,
+      "loss": 0.4446,
+      "step": 3833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005557084649338969,
+      "loss": 0.4894,
+      "step": 3834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005555067265464832,
+      "loss": 0.4219,
+      "step": 3835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005553049790097688,
+      "loss": 0.5103,
+      "step": 3836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005551032223570083,
+      "loss": 0.4285,
+      "step": 3837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005549014566214574,
+      "loss": 0.5586,
+      "step": 3838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005546996818363738,
+      "loss": 0.3989,
+      "step": 3839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005544978980350164,
+      "loss": 0.4395,
+      "step": 3840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005542961052506459,
+      "loss": 0.3981,
+      "step": 3841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005540943035165239,
+      "loss": 0.3448,
+      "step": 3842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005538924928659143,
+      "loss": 0.6603,
+      "step": 3843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536906733320815,
+      "loss": 0.4077,
+      "step": 3844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005534888449482924,
+      "loss": 0.3854,
+      "step": 3845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005532870077478144,
+      "loss": 0.3625,
+      "step": 3846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005530851617639169,
+      "loss": 0.4436,
+      "step": 3847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005528833070298709,
+      "loss": 0.3604,
+      "step": 3848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005526814435789483,
+      "loss": 0.4685,
+      "step": 3849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005524795714444226,
+      "loss": 0.521,
+      "step": 3850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005522776906595691,
+      "loss": 0.3623,
+      "step": 3851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005520758012576643,
+      "loss": 0.3524,
+      "step": 3852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005518739032719861,
+      "loss": 0.3729,
+      "step": 3853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005516719967358135,
+      "loss": 0.3706,
+      "step": 3854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005514700816824275,
+      "loss": 0.5282,
+      "step": 3855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005512681581451101,
+      "loss": 0.3986,
+      "step": 3856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005510662261571449,
+      "loss": 0.4268,
+      "step": 3857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005508642857518167,
+      "loss": 0.5026,
+      "step": 3858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005506623369624121,
+      "loss": 0.4003,
+      "step": 3859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005504603798222183,
+      "loss": 0.4253,
+      "step": 3860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005502584143645247,
+      "loss": 0.433,
+      "step": 3861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005500564406226216,
+      "loss": 0.4174,
+      "step": 3862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005498544586298009,
+      "loss": 0.4664,
+      "step": 3863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005496524684193556,
+      "loss": 0.4557,
+      "step": 3864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005494504700245805,
+      "loss": 0.3713,
+      "step": 3865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000549248463478771,
+      "loss": 0.4734,
+      "step": 3866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005490464488152247,
+      "loss": 0.4019,
+      "step": 3867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005488444260672399,
+      "loss": 0.4216,
+      "step": 3868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005486423952681166,
+      "loss": 0.4136,
+      "step": 3869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005484403564511561,
+      "loss": 0.0515,
+      "step": 3870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005482383096496606,
+      "loss": 0.3903,
+      "step": 3871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005480362548969342,
+      "loss": 0.4994,
+      "step": 3872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005478341922262818,
+      "loss": 0.4194,
+      "step": 3873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005476321216710103,
+      "loss": 0.4191,
+      "step": 3874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000547430043264427,
+      "loss": 0.4153,
+      "step": 3875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005472279570398412,
+      "loss": 0.4624,
+      "step": 3876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005470258630305632,
+      "loss": 0.5234,
+      "step": 3877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005468237612699044,
+      "loss": 0.4362,
+      "step": 3878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000546621651791178,
+      "loss": 0.4294,
+      "step": 3879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005464195346276979,
+      "loss": 0.4369,
+      "step": 3880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005462174098127798,
+      "loss": 0.4412,
+      "step": 3881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005460152773797402,
+      "loss": 0.0526,
+      "step": 3882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005458131373618971,
+      "loss": 0.5406,
+      "step": 3883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005456109897925698,
+      "loss": 0.3691,
+      "step": 3884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005454088347050786,
+      "loss": 0.4148,
+      "step": 3885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005452066721327454,
+      "loss": 0.4899,
+      "step": 3886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005450045021088928,
+      "loss": 0.4926,
+      "step": 3887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005448023246668453,
+      "loss": 0.559,
+      "step": 3888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000544600139839928,
+      "loss": 0.5234,
+      "step": 3889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005443979476614675,
+      "loss": 0.4067,
+      "step": 3890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005441957481647917,
+      "loss": 0.3936,
+      "step": 3891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005439935413832297,
+      "loss": 0.3749,
+      "step": 3892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005437913273501114,
+      "loss": 0.4586,
+      "step": 3893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005435891060987685,
+      "loss": 0.3865,
+      "step": 3894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005433868776625333,
+      "loss": 0.4224,
+      "step": 3895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005431846420747401,
+      "loss": 0.4282,
+      "step": 3896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005429823993687233,
+      "loss": 0.5294,
+      "step": 3897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005427801495778192,
+      "loss": 0.5658,
+      "step": 3898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005425778927353653,
+      "loss": 0.5162,
+      "step": 3899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423756288746998,
+      "loss": 0.4381,
+      "step": 3900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005421733580291624,
+      "loss": 0.5271,
+      "step": 3901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005419710802320939,
+      "loss": 0.3964,
+      "step": 3902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005417687955168362,
+      "loss": 0.4943,
+      "step": 3903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005415665039167324,
+      "loss": 0.3782,
+      "step": 3904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005413642054651265,
+      "loss": 0.4861,
+      "step": 3905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005411619001953641,
+      "loss": 0.4384,
+      "step": 3906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005409595881407913,
+      "loss": 0.4149,
+      "step": 3907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000540757269334756,
+      "loss": 0.4147,
+      "step": 3908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005405549438106067,
+      "loss": 0.4489,
+      "step": 3909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005403526116016931,
+      "loss": 0.4313,
+      "step": 3910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005401502727413663,
+      "loss": 0.4236,
+      "step": 3911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000539947927262978,
+      "loss": 0.462,
+      "step": 3912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005397455751998814,
+      "loss": 0.7457,
+      "step": 3913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000539543216585431,
+      "loss": 0.4836,
+      "step": 3914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005393408514529814,
+      "loss": 0.4102,
+      "step": 3915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005391384798358892,
+      "loss": 0.5558,
+      "step": 3916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005389361017675119,
+      "loss": 0.4579,
+      "step": 3917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000538733717281208,
+      "loss": 0.399,
+      "step": 3918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005385313264103367,
+      "loss": 0.4697,
+      "step": 3919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005383289291882586,
+      "loss": 0.502,
+      "step": 3920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005381265256483355,
+      "loss": 0.5326,
+      "step": 3921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005379241158239298,
+      "loss": 0.3917,
+      "step": 3922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005377216997484054,
+      "loss": 0.4308,
+      "step": 3923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005375192774551269,
+      "loss": 0.4582,
+      "step": 3924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00053731684897746,
+      "loss": 0.4652,
+      "step": 3925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005371144143487714,
+      "loss": 0.4736,
+      "step": 3926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005369119736024291,
+      "loss": 0.4958,
+      "step": 3927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005367095267718016,
+      "loss": 0.3579,
+      "step": 3928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005365070738902588,
+      "loss": 0.0529,
+      "step": 3929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005363046149911715,
+      "loss": 0.4393,
+      "step": 3930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005361021501079115,
+      "loss": 0.4658,
+      "step": 3931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005358996792738513,
+      "loss": 0.4185,
+      "step": 3932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005356972025223649,
+      "loss": 0.4994,
+      "step": 3933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005354947198868269,
+      "loss": 0.4889,
+      "step": 3934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005352922314006129,
+      "loss": 0.4897,
+      "step": 3935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005350897370970997,
+      "loss": 0.4958,
+      "step": 3936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005348872370096649,
+      "loss": 0.3193,
+      "step": 3937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005346847311716868,
+      "loss": 0.4015,
+      "step": 3938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005344822196165453,
+      "loss": 0.524,
+      "step": 3939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005342797023776205,
+      "loss": 0.5438,
+      "step": 3940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005340771794882938,
+      "loss": 0.4863,
+      "step": 3941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005338746509819478,
+      "loss": 0.4536,
+      "step": 3942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005336721168919655,
+      "loss": 0.4779,
+      "step": 3943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005334695772517311,
+      "loss": 0.5254,
+      "step": 3944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005332670320946297,
+      "loss": 0.4935,
+      "step": 3945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005330644814540472,
+      "loss": 0.4825,
+      "step": 3946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005328619253633706,
+      "loss": 0.4505,
+      "step": 3947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005326593638559875,
+      "loss": 0.0528,
+      "step": 3948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000532456796965287,
+      "loss": 0.5862,
+      "step": 3949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005322542247246583,
+      "loss": 0.4617,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005320516471674918,
+      "loss": 0.394,
+      "step": 3951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005318490643271789,
+      "loss": 0.4371,
+      "step": 3952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005316464762371119,
+      "loss": 0.4353,
+      "step": 3953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005314438829306839,
+      "loss": 0.4012,
+      "step": 3954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005312412844412886,
+      "loss": 0.0518,
+      "step": 3955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005310386808023208,
+      "loss": 0.4137,
+      "step": 3956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005308360720471763,
+      "loss": 0.3972,
+      "step": 3957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005306334582092515,
+      "loss": 0.472,
+      "step": 3958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005304308393219437,
+      "loss": 0.4453,
+      "step": 3959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000530228215418651,
+      "loss": 0.431,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005300255865327724,
+      "loss": 0.4646,
+      "step": 3961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005298229526977075,
+      "loss": 0.051,
+      "step": 3962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005296203139468572,
+      "loss": 0.3944,
+      "step": 3963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005294176703136227,
+      "loss": 0.4352,
+      "step": 3964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005292150218314063,
+      "loss": 0.4083,
+      "step": 3965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005290123685336109,
+      "loss": 0.3812,
+      "step": 3966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005288097104536405,
+      "loss": 0.4475,
+      "step": 3967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005286070476248995,
+      "loss": 0.4337,
+      "step": 3968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005284043800807934,
+      "loss": 0.5179,
+      "step": 3969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005282017078547285,
+      "loss": 0.4163,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005279990309801115,
+      "loss": 0.4421,
+      "step": 3971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005277963494903499,
+      "loss": 0.4906,
+      "step": 3972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005275936634188528,
+      "loss": 0.5369,
+      "step": 3973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005273909727990286,
+      "loss": 0.42,
+      "step": 3974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005271882776642878,
+      "loss": 0.4744,
+      "step": 3975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005269855780480411,
+      "loss": 0.4116,
+      "step": 3976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005267828739836996,
+      "loss": 0.4115,
+      "step": 3977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005265801655046758,
+      "loss": 0.4845,
+      "step": 3978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005263774526443824,
+      "loss": 0.5592,
+      "step": 3979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005261747354362333,
+      "loss": 0.4384,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005259720139136426,
+      "loss": 0.4518,
+      "step": 3981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005257692881100252,
+      "loss": 0.5272,
+      "step": 3982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005255665580587974,
+      "loss": 0.0465,
+      "step": 3983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005253638237933751,
+      "loss": 0.4292,
+      "step": 3984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005251610853471756,
+      "loss": 0.4833,
+      "step": 3985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000524958342753617,
+      "loss": 0.4806,
+      "step": 3986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005247555960461175,
+      "loss": 0.4098,
+      "step": 3987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005245528452580965,
+      "loss": 0.4711,
+      "step": 3988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005243500904229738,
+      "loss": 0.4893,
+      "step": 3989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005241473315741701,
+      "loss": 0.4076,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005239445687451064,
+      "loss": 0.4297,
+      "step": 3991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005237418019692046,
+      "loss": 0.5012,
+      "step": 3992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005235390312798875,
+      "loss": 0.4906,
+      "step": 3993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005233362567105776,
+      "loss": 0.4346,
+      "step": 3994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005231334782946994,
+      "loss": 0.4948,
+      "step": 3995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005229306960656769,
+      "loss": 0.3791,
+      "step": 3996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005227279100569353,
+      "loss": 0.4117,
+      "step": 3997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005225251203019002,
+      "loss": 0.4833,
+      "step": 3998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005223223268339981,
+      "loss": 0.5604,
+      "step": 3999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005221195296866556,
+      "loss": 0.4828,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005219167288933004,
+      "loss": 0.4879,
+      "step": 4001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005217139244873608,
+      "loss": 0.3912,
+      "step": 4002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005215111165022652,
+      "loss": 0.3881,
+      "step": 4003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005213083049714431,
+      "loss": 0.488,
+      "step": 4004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005211054899283244,
+      "loss": 0.4621,
+      "step": 4005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005209026714063394,
+      "loss": 0.4598,
+      "step": 4006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005206998494389193,
+      "loss": 0.4955,
+      "step": 4007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005204970240594955,
+      "loss": 0.4545,
+      "step": 4008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005202941953015005,
+      "loss": 0.4033,
+      "step": 4009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005200913631983665,
+      "loss": 0.5524,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005198885277835274,
+      "loss": 0.4233,
+      "step": 4011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005196856890904168,
+      "loss": 0.4451,
+      "step": 4012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000519482847152469,
+      "loss": 0.4039,
+      "step": 4013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005192800020031189,
+      "loss": 0.475,
+      "step": 4014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005190771536758017,
+      "loss": 0.417,
+      "step": 4015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005188743022039539,
+      "loss": 0.4343,
+      "step": 4016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005186714476210116,
+      "loss": 0.4614,
+      "step": 4017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005184685899604118,
+      "loss": 0.3961,
+      "step": 4018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000518265729255592,
+      "loss": 0.4524,
+      "step": 4019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005180628655399904,
+      "loss": 0.4197,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005178599988470452,
+      "loss": 0.5393,
+      "step": 4021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005176571292101955,
+      "loss": 0.3965,
+      "step": 4022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005174542566628808,
+      "loss": 0.4346,
+      "step": 4023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005172513812385411,
+      "loss": 0.4825,
+      "step": 4024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005170485029706166,
+      "loss": 0.5662,
+      "step": 4025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005168456218925486,
+      "loss": 0.3767,
+      "step": 4026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000516642738037778,
+      "loss": 0.4025,
+      "step": 4027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005164398514397469,
+      "loss": 0.4438,
+      "step": 4028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005162369621318973,
+      "loss": 0.3912,
+      "step": 4029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005160340701476722,
+      "loss": 0.3855,
+      "step": 4030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005158311755205146,
+      "loss": 0.4254,
+      "step": 4031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005156282782838681,
+      "loss": 0.3784,
+      "step": 4032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005154253784711767,
+      "loss": 0.0562,
+      "step": 4033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000515222476115885,
+      "loss": 0.5126,
+      "step": 4034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005150195712514376,
+      "loss": 0.4609,
+      "step": 4035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005148166639112799,
+      "loss": 0.3926,
+      "step": 4036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005146137541288576,
+      "loss": 0.4421,
+      "step": 4037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005144108419376169,
+      "loss": 0.4537,
+      "step": 4038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142079273710038,
+      "loss": 0.4729,
+      "step": 4039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005140050104624657,
+      "loss": 0.4287,
+      "step": 4040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005138020912454497,
+      "loss": 0.4674,
+      "step": 4041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005135991697534035,
+      "loss": 0.5226,
+      "step": 4042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005133962460197748,
+      "loss": 0.474,
+      "step": 4043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005131933200780124,
+      "loss": 0.4995,
+      "step": 4044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005129903919615647,
+      "loss": 0.3882,
+      "step": 4045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005127874617038811,
+      "loss": 0.4308,
+      "step": 4046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000512584529338411,
+      "loss": 0.4415,
+      "step": 4047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005123815948986038,
+      "loss": 0.1376,
+      "step": 4048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005121786584179103,
+      "loss": 0.5435,
+      "step": 4049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005119757199297805,
+      "loss": 0.4608,
+      "step": 4050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005117727794676654,
+      "loss": 0.426,
+      "step": 4051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005115698370650161,
+      "loss": 0.4452,
+      "step": 4052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005113668927552842,
+      "loss": 0.4082,
+      "step": 4053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005111639465719212,
+      "loss": 0.3682,
+      "step": 4054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005109609985483794,
+      "loss": 0.4659,
+      "step": 4055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005107580487181112,
+      "loss": 0.438,
+      "step": 4056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005105550971145692,
+      "loss": 0.3344,
+      "step": 4057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005103521437712063,
+      "loss": 0.4208,
+      "step": 4058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005101491887214757,
+      "loss": 0.4462,
+      "step": 4059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005099462319988314,
+      "loss": 0.4689,
+      "step": 4060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005097432736367267,
+      "loss": 0.3884,
+      "step": 4061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000509540313668616,
+      "loss": 0.3844,
+      "step": 4062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005093373521279535,
+      "loss": 0.4196,
+      "step": 4063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000509134389048194,
+      "loss": 0.3566,
+      "step": 4064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005089314244627922,
+      "loss": 0.4264,
+      "step": 4065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005087284584052031,
+      "loss": 0.5184,
+      "step": 4066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005085254909088825,
+      "loss": 0.5076,
+      "step": 4067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005083225220072856,
+      "loss": 0.3798,
+      "step": 4068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005081195517338682,
+      "loss": 0.4346,
+      "step": 4069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005079165801220866,
+      "loss": 0.4177,
+      "step": 4070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005077136072053972,
+      "loss": 0.399,
+      "step": 4071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000507510633017256,
+      "loss": 0.1244,
+      "step": 4072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00050730765759112,
+      "loss": 0.4642,
+      "step": 4073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005071046809604461,
+      "loss": 0.3757,
+      "step": 4074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005069017031586916,
+      "loss": 0.4828,
+      "step": 4075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005066987242193133,
+      "loss": 0.4362,
+      "step": 4076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005064957441757691,
+      "loss": 0.5209,
+      "step": 4077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005062927630615164,
+      "loss": 0.4093,
+      "step": 4078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005060897809100133,
+      "loss": 0.5393,
+      "step": 4079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005058867977547175,
+      "loss": 0.3849,
+      "step": 4080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005056838136290875,
+      "loss": 0.4402,
+      "step": 4081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005054808285665815,
+      "loss": 0.4464,
+      "step": 4082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000505277842600658,
+      "loss": 0.3782,
+      "step": 4083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005050748557647756,
+      "loss": 0.3745,
+      "step": 4084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005048718680923929,
+      "loss": 0.4191,
+      "step": 4085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005046688796169693,
+      "loss": 0.4647,
+      "step": 4086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005044658903719634,
+      "loss": 0.3677,
+      "step": 4087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005042629003908347,
+      "loss": 0.0636,
+      "step": 4088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005040599097070424,
+      "loss": 0.4594,
+      "step": 4089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005038569183540458,
+      "loss": 0.5007,
+      "step": 4090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005036539263653045,
+      "loss": 0.4731,
+      "step": 4091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005034509337742781,
+      "loss": 0.4425,
+      "step": 4092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005032479406144266,
+      "loss": 0.5176,
+      "step": 4093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005030449469192094,
+      "loss": 0.4938,
+      "step": 4094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005028419527220867,
+      "loss": 0.4753,
+      "step": 4095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005026389580565185,
+      "loss": 0.3767,
+      "step": 4096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005024359629559648,
+      "loss": 0.442,
+      "step": 4097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005022329674538857,
+      "loss": 0.4686,
+      "step": 4098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005020299715837415,
+      "loss": 0.5953,
+      "step": 4099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005018269753789924,
+      "loss": 0.4113,
+      "step": 4100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005016239788730989,
+      "loss": 0.4001,
+      "step": 4101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005014209820995211,
+      "loss": 0.4559,
+      "step": 4102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005012179850917195,
+      "loss": 0.4512,
+      "step": 4103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005010149878831549,
+      "loss": 0.575,
+      "step": 4104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005008119905072873,
+      "loss": 0.0629,
+      "step": 4105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005006089929975774,
+      "loss": 0.4357,
+      "step": 4106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005004059953874859,
+      "loss": 0.4291,
+      "step": 4107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005002029977104733,
+      "loss": 0.4253,
+      "step": 4108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0005,
+      "loss": 0.43,
+      "step": 4109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004997970022895268,
+      "loss": 0.4836,
+      "step": 4110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499594004612514,
+      "loss": 0.4484,
+      "step": 4111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004993910070024225,
+      "loss": 0.5653,
+      "step": 4112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499188009492713,
+      "loss": 0.4813,
+      "step": 4113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004989850121168454,
+      "loss": 0.4565,
+      "step": 4114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004987820149082805,
+      "loss": 0.3934,
+      "step": 4115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000498579017900479,
+      "loss": 0.4779,
+      "step": 4116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004983760211269011,
+      "loss": 0.7081,
+      "step": 4117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004981730246210077,
+      "loss": 0.421,
+      "step": 4118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004979700284162585,
+      "loss": 0.5106,
+      "step": 4119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004977670325461144,
+      "loss": 0.5315,
+      "step": 4120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004975640370440352,
+      "loss": 0.426,
+      "step": 4121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004973610419434815,
+      "loss": 0.5387,
+      "step": 4122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004971580472779133,
+      "loss": 0.3843,
+      "step": 4123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004969550530807906,
+      "loss": 0.43,
+      "step": 4124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004967520593855735,
+      "loss": 0.4098,
+      "step": 4125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004965490662257218,
+      "loss": 0.4395,
+      "step": 4126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004963460736346955,
+      "loss": 0.4166,
+      "step": 4127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004961430816459544,
+      "loss": 0.5063,
+      "step": 4128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004959400902929578,
+      "loss": 0.3494,
+      "step": 4129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004957370996091654,
+      "loss": 0.4507,
+      "step": 4130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004955341096280366,
+      "loss": 0.4931,
+      "step": 4131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004953311203830308,
+      "loss": 0.426,
+      "step": 4132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004951281319076071,
+      "loss": 0.5085,
+      "step": 4133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004949251442352245,
+      "loss": 0.4355,
+      "step": 4134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004947221573993422,
+      "loss": 0.473,
+      "step": 4135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004945191714334186,
+      "loss": 0.4451,
+      "step": 4136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004943161863709125,
+      "loss": 0.4255,
+      "step": 4137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004941132022452825,
+      "loss": 0.4323,
+      "step": 4138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004939102190899868,
+      "loss": 0.5042,
+      "step": 4139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004937072369384836,
+      "loss": 0.4824,
+      "step": 4140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000493504255824231,
+      "loss": 0.386,
+      "step": 4141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004933012757806868,
+      "loss": 0.0614,
+      "step": 4142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004930982968413086,
+      "loss": 0.3856,
+      "step": 4143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004928953190395539,
+      "loss": 0.5101,
+      "step": 4144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004926923424088801,
+      "loss": 0.4641,
+      "step": 4145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004924893669827441,
+      "loss": 0.405,
+      "step": 4146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004922863927946029,
+      "loss": 0.427,
+      "step": 4147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004920834198779133,
+      "loss": 0.3818,
+      "step": 4148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004918804482661318,
+      "loss": 0.447,
+      "step": 4149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004916774779927147,
+      "loss": 0.4653,
+      "step": 4150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004914745090911178,
+      "loss": 0.4424,
+      "step": 4151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004912715415947969,
+      "loss": 0.4268,
+      "step": 4152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000491068575537208,
+      "loss": 0.5728,
+      "step": 4153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004908656109518061,
+      "loss": 0.3988,
+      "step": 4154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004906626478720465,
+      "loss": 0.4531,
+      "step": 4155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000490459686331384,
+      "loss": 0.4191,
+      "step": 4156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004902567263632735,
+      "loss": 0.393,
+      "step": 4157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004900537680011688,
+      "loss": 0.4824,
+      "step": 4158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004898508112785243,
+      "loss": 0.5117,
+      "step": 4159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004896478562287939,
+      "loss": 0.3829,
+      "step": 4160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004894449028854309,
+      "loss": 0.4129,
+      "step": 4161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004892419512818889,
+      "loss": 0.4032,
+      "step": 4162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004890390014516206,
+      "loss": 0.3975,
+      "step": 4163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004888360534280789,
+      "loss": 0.5112,
+      "step": 4164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000488633107244716,
+      "loss": 0.4127,
+      "step": 4165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000488430162934984,
+      "loss": 0.486,
+      "step": 4166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048822722053233473,
+      "loss": 0.0602,
+      "step": 4167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004880242800702196,
+      "loss": 0.46,
+      "step": 4168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004878213415820898,
+      "loss": 0.4218,
+      "step": 4169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004876184051013961,
+      "loss": 0.4229,
+      "step": 4170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048741547066158907,
+      "loss": 0.4431,
+      "step": 4171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048721253829611905,
+      "loss": 0.4408,
+      "step": 4172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048700960803843536,
+      "loss": 0.4437,
+      "step": 4173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048680667992198774,
+      "loss": 0.4731,
+      "step": 4174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004866037539802252,
+      "loss": 0.451,
+      "step": 4175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004864008302465966,
+      "loss": 0.4579,
+      "step": 4176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004861979087545503,
+      "loss": 0.4463,
+      "step": 4177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048599498953753417,
+      "loss": 0.4214,
+      "step": 4178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048579207262899627,
+      "loss": 0.5203,
+      "step": 4179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048558915806238336,
+      "loss": 0.4652,
+      "step": 4180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004853862458711425,
+      "loss": 0.5236,
+      "step": 4181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004851833360887201,
+      "loss": 0.5182,
+      "step": 4182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004849804287485625,
+      "loss": 0.5,
+      "step": 4183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004847775238841151,
+      "loss": 0.4269,
+      "step": 4184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048457462152882324,
+      "loss": 0.4177,
+      "step": 4185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000484371721716132,
+      "loss": 0.3507,
+      "step": 4186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004841688244794855,
+      "loss": 0.4015,
+      "step": 4187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004839659298523279,
+      "loss": 0.0592,
+      "step": 4188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004837630378681027,
+      "loss": 0.4003,
+      "step": 4189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004835601485602532,
+      "loss": 0.4019,
+      "step": 4190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000483357261962222,
+      "loss": 0.4042,
+      "step": 4191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048315437810745145,
+      "loss": 0.4517,
+      "step": 4192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004829514970293832,
+      "loss": 0.402,
+      "step": 4193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048274861876145903,
+      "loss": 0.3972,
+      "step": 4194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048254574333711926,
+      "loss": 0.4048,
+      "step": 4195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004823428707898046,
+      "loss": 0.4573,
+      "step": 4196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048214000115295494,
+      "loss": 0.427,
+      "step": 4197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004819371344600097,
+      "loss": 0.4093,
+      "step": 4198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048173427074440795,
+      "loss": 0.4012,
+      "step": 4199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004815314100395882,
+      "loss": 0.5428,
+      "step": 4200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004813285523789886,
+      "loss": 0.3875,
+      "step": 4201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004811256977960462,
+      "loss": 0.414,
+      "step": 4202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004809228463241982,
+      "loss": 0.4452,
+      "step": 4203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004807199979968812,
+      "loss": 0.4832,
+      "step": 4204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048051715284753115,
+      "loss": 0.5115,
+      "step": 4205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004803143109095832,
+      "loss": 0.4053,
+      "step": 4206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004801114722164725,
+      "loss": 0.4991,
+      "step": 4207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004799086368016334,
+      "loss": 0.4323,
+      "step": 4208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047970580469849976,
+      "loss": 0.5128,
+      "step": 4209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004795029759405046,
+      "loss": 0.4578,
+      "step": 4210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004793001505610809,
+      "loss": 0.5317,
+      "step": 4211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004790973285936607,
+      "loss": 0.4406,
+      "step": 4212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047889451007167565,
+      "loss": 0.4745,
+      "step": 4213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047869169502855684,
+      "loss": 0.5582,
+      "step": 4214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047848888349773466,
+      "loss": 0.4041,
+      "step": 4215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047828607551263935,
+      "loss": 0.4038,
+      "step": 4216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004780832711066996,
+      "loss": 0.4327,
+      "step": 4217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047788047031334445,
+      "loss": 0.4653,
+      "step": 4218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004776776731660021,
+      "loss": 0.3336,
+      "step": 4219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004774748796980999,
+      "loss": 0.4573,
+      "step": 4220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047727208994306476,
+      "loss": 0.415,
+      "step": 4221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047706930393432323,
+      "loss": 0.4564,
+      "step": 4222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004768665217053008,
+      "loss": 0.4236,
+      "step": 4223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047666374328942247,
+      "loss": 0.4587,
+      "step": 4224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004764609687201127,
+      "loss": 0.381,
+      "step": 4225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047625819803079545,
+      "loss": 0.4601,
+      "step": 4226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047605543125489363,
+      "loss": 0.4432,
+      "step": 4227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047585266842582985,
+      "loss": 0.4456,
+      "step": 4228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047564990957702603,
+      "loss": 0.4482,
+      "step": 4229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047544715474190347,
+      "loss": 0.4614,
+      "step": 4230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004752444039538826,
+      "loss": 0.3567,
+      "step": 4231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004750416572463831,
+      "loss": 0.4164,
+      "step": 4232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004748389146528244,
+      "loss": 0.3737,
+      "step": 4233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047463617620662504,
+      "loss": 0.4373,
+      "step": 4234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004744334419412027,
+      "loss": 0.4075,
+      "step": 4235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047423071188997467,
+      "loss": 0.4241,
+      "step": 4236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047402798608635766,
+      "loss": 0.4604,
+      "step": 4237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004738252645637668,
+      "loss": 0.4956,
+      "step": 4238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047362254735561763,
+      "loss": 0.4311,
+      "step": 4239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004734198344953243,
+      "loss": 0.4473,
+      "step": 4240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047321712601630043,
+      "loss": 0.4081,
+      "step": 4241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000473014421951959,
+      "loss": 0.541,
+      "step": 4242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047281172233571214,
+      "loss": 0.4558,
+      "step": 4243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047260902720097143,
+      "loss": 0.0582,
+      "step": 4244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047240633658114753,
+      "loss": 0.4611,
+      "step": 4245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004722036505096501,
+      "loss": 0.38,
+      "step": 4246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047200096901988867,
+      "loss": 0.4749,
+      "step": 4247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004717982921452716,
+      "loss": 0.481,
+      "step": 4248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004715956199192065,
+      "loss": 0.494,
+      "step": 4249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047139295237510044,
+      "loss": 0.5388,
+      "step": 4250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047119028954635946,
+      "loss": 0.4329,
+      "step": 4251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047098763146638924,
+      "loss": 0.3969,
+      "step": 4252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004707849781685939,
+      "loss": 0.3839,
+      "step": 4253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004705823296863774,
+      "loss": 0.4626,
+      "step": 4254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004703796860531429,
+      "loss": 0.0574,
+      "step": 4255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047017704730229257,
+      "loss": 0.4165,
+      "step": 4256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004699744134672277,
+      "loss": 0.4138,
+      "step": 4257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000469771784581349,
+      "loss": 0.4392,
+      "step": 4258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046956916067805645,
+      "loss": 0.4117,
+      "step": 4259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046936654179074864,
+      "loss": 0.0572,
+      "step": 4260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004691639279528238,
+      "loss": 0.4797,
+      "step": 4261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004689613191976792,
+      "loss": 0.5117,
+      "step": 4262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004687587155587115,
+      "loss": 0.4766,
+      "step": 4263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004685561170693162,
+      "loss": 0.4996,
+      "step": 4264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000468353523762888,
+      "loss": 0.3397,
+      "step": 4265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046815093567282115,
+      "loss": 0.4033,
+      "step": 4266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004679483528325084,
+      "loss": 0.0562,
+      "step": 4267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004677457752753419,
+      "loss": 0.4701,
+      "step": 4268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046754320303471307,
+      "loss": 0.4384,
+      "step": 4269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004673406361440125,
+      "loss": 0.3774,
+      "step": 4270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004671380746366295,
+      "loss": 0.4403,
+      "step": 4271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046693551854595284,
+      "loss": 0.4301,
+      "step": 4272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046673296790537036,
+      "loss": 0.3804,
+      "step": 4273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004665304227482691,
+      "loss": 0.4277,
+      "step": 4274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046632788310803467,
+      "loss": 0.0555,
+      "step": 4275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004661253490180523,
+      "loss": 0.384,
+      "step": 4276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046592282051170623,
+      "loss": 0.4221,
+      "step": 4277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046572029762237963,
+      "loss": 0.4631,
+      "step": 4278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004655177803834548,
+      "loss": 0.5019,
+      "step": 4279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004653152688283132,
+      "loss": 0.4194,
+      "step": 4280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046511276299033536,
+      "loss": 0.4073,
+      "step": 4281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004649102629029004,
+      "loss": 0.5156,
+      "step": 4282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004647077685993872,
+      "loss": 0.4503,
+      "step": 4283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004645052801131732,
+      "loss": 0.3671,
+      "step": 4284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004643027974776352,
+      "loss": 0.3754,
+      "step": 4285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004641003207261487,
+      "loss": 0.4092,
+      "step": 4286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004638978498920887,
+      "loss": 0.4137,
+      "step": 4287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046369538500882855,
+      "loss": 0.4451,
+      "step": 4288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046349292610974133,
+      "loss": 0.5051,
+      "step": 4289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004632904732281985,
+      "loss": 0.481,
+      "step": 4290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004630880263975711,
+      "loss": 0.5807,
+      "step": 4291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004628855856512287,
+      "loss": 0.453,
+      "step": 4292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004626831510225401,
+      "loss": 0.4702,
+      "step": 4293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004624807225448732,
+      "loss": 0.4778,
+      "step": 4294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004622783002515946,
+      "loss": 0.4037,
+      "step": 4295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004620758841760703,
+      "loss": 0.4899,
+      "step": 4296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004618734743516647,
+      "loss": 0.392,
+      "step": 4297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004616710708117415,
+      "loss": 0.3839,
+      "step": 4298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004614686735896634,
+      "loss": 0.4309,
+      "step": 4299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046126628271879206,
+      "loss": 0.4233,
+      "step": 4300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000461063898232488,
+      "loss": 0.4899,
+      "step": 4301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046086152016411064,
+      "loss": 0.5226,
+      "step": 4302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004606591485470188,
+      "loss": 0.4868,
+      "step": 4303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046045678341456926,
+      "loss": 0.3771,
+      "step": 4304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004602544248001186,
+      "loss": 0.4288,
+      "step": 4305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004600520727370221,
+      "loss": 0.5249,
+      "step": 4306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004598497272586338,
+      "loss": 0.0542,
+      "step": 4307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045964738839830694,
+      "loss": 0.5135,
+      "step": 4308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004594450561893935,
+      "loss": 0.4932,
+      "step": 4309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004592427306652441,
+      "loss": 0.4832,
+      "step": 4310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004590404118592088,
+      "loss": 0.5796,
+      "step": 4311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045883809980463606,
+      "loss": 0.3793,
+      "step": 4312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045863579453487365,
+      "loss": 0.5104,
+      "step": 4313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004584334960832678,
+      "loss": 0.5442,
+      "step": 4314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004582312044831639,
+      "loss": 0.3784,
+      "step": 4315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004580289197679061,
+      "loss": 0.5492,
+      "step": 4316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004578266419708376,
+      "loss": 0.4294,
+      "step": 4317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004576243711253004,
+      "loss": 0.4071,
+      "step": 4318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004574221072646349,
+      "loss": 0.0537,
+      "step": 4319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045721985042218084,
+      "loss": 0.4587,
+      "step": 4320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045701760063127686,
+      "loss": 0.4468,
+      "step": 4321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045681535792526,
+      "loss": 0.0534,
+      "step": 4322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045661312233746663,
+      "loss": 0.4965,
+      "step": 4323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045641089390123155,
+      "loss": 0.4634,
+      "step": 4324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004562086726498888,
+      "loss": 0.5527,
+      "step": 4325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004560064586167706,
+      "loss": 0.3678,
+      "step": 4326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004558042518352085,
+      "loss": 0.429,
+      "step": 4327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045560205233853264,
+      "loss": 0.053,
+      "step": 4328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004553998601600721,
+      "loss": 0.4144,
+      "step": 4329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045519767533315475,
+      "loss": 0.4614,
+      "step": 4330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004549954978911072,
+      "loss": 0.4594,
+      "step": 4331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045479332786725466,
+      "loss": 0.4117,
+      "step": 4332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045459116529492137,
+      "loss": 0.55,
+      "step": 4333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004543890102074302,
+      "loss": 0.4557,
+      "step": 4334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045418686263810296,
+      "loss": 0.4139,
+      "step": 4335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045398472262025985,
+      "loss": 0.4907,
+      "step": 4336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045378259018722023,
+      "loss": 0.4565,
+      "step": 4337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004535804653723021,
+      "loss": 0.438,
+      "step": 4338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045337834820882207,
+      "loss": 0.4406,
+      "step": 4339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045317623873009576,
+      "loss": 0.5468,
+      "step": 4340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045297413696943707,
+      "loss": 0.4768,
+      "step": 4341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045277204296015897,
+      "loss": 0.4175,
+      "step": 4342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045256995673557303,
+      "loss": 0.4546,
+      "step": 4343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004523678783289898,
+      "loss": 0.4752,
+      "step": 4344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004521658077737181,
+      "loss": 0.3873,
+      "step": 4345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045196374510306583,
+      "loss": 0.4304,
+      "step": 4346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045176169035033957,
+      "loss": 0.5201,
+      "step": 4347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004515596435488441,
+      "loss": 0.3717,
+      "step": 4348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004513576047318835,
+      "loss": 0.4863,
+      "step": 4349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004511555739327601,
+      "loss": 0.4996,
+      "step": 4350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004509535511847753,
+      "loss": 0.5193,
+      "step": 4351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004507515365212291,
+      "loss": 0.3875,
+      "step": 4352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045054952997541963,
+      "loss": 0.3837,
+      "step": 4353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045034753158064433,
+      "loss": 0.4241,
+      "step": 4354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004501455413701992,
+      "loss": 0.4403,
+      "step": 4355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004499435593773785,
+      "loss": 0.4302,
+      "step": 4356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004497415856354755,
+      "loss": 0.4977,
+      "step": 4357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044953962017778184,
+      "loss": 0.4293,
+      "step": 4358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044933766303758806,
+      "loss": 0.402,
+      "step": 4359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044913571424818325,
+      "loss": 0.4497,
+      "step": 4360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044893377384285513,
+      "loss": 0.5533,
+      "step": 4361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044873184185489006,
+      "loss": 0.4933,
+      "step": 4362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004485299183175727,
+      "loss": 0.3779,
+      "step": 4363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044832800326418667,
+      "loss": 0.4433,
+      "step": 4364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004481260967280141,
+      "loss": 0.4257,
+      "step": 4365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004479241987423357,
+      "loss": 0.4696,
+      "step": 4366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044772230934043083,
+      "loss": 0.4526,
+      "step": 4367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004475204285555774,
+      "loss": 0.4429,
+      "step": 4368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000447318556421052,
+      "loss": 0.4509,
+      "step": 4369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004471166929701293,
+      "loss": 0.6022,
+      "step": 4370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044691483823608316,
+      "loss": 0.4706,
+      "step": 4371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004467129922521857,
+      "loss": 0.5089,
+      "step": 4372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004465111550517077,
+      "loss": 0.0519,
+      "step": 4373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004463093266679185,
+      "loss": 0.5062,
+      "step": 4374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004461075071340858,
+      "loss": 0.3878,
+      "step": 4375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004459056964834761,
+      "loss": 0.4503,
+      "step": 4376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004457038947493542,
+      "loss": 0.3858,
+      "step": 4377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004455021019649837,
+      "loss": 0.3992,
+      "step": 4378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044530031816362634,
+      "loss": 0.4542,
+      "step": 4379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004450985433785427,
+      "loss": 0.524,
+      "step": 4380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044489677764299174,
+      "loss": 0.4187,
+      "step": 4381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044469502099023114,
+      "loss": 0.4086,
+      "step": 4382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044449327345351674,
+      "loss": 0.3945,
+      "step": 4383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004442915350661034,
+      "loss": 0.4542,
+      "step": 4384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004440898058612435,
+      "loss": 0.4381,
+      "step": 4385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044388808587218895,
+      "loss": 0.5529,
+      "step": 4386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004436863751321896,
+      "loss": 0.3971,
+      "step": 4387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004434846736744939,
+      "loss": 0.4556,
+      "step": 4388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004432829815323488,
+      "loss": 0.3682,
+      "step": 4389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004430812987389996,
+      "loss": 0.4138,
+      "step": 4390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004428796253276905,
+      "loss": 0.3812,
+      "step": 4391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004426779613316632,
+      "loss": 0.399,
+      "step": 4392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004424763067841588,
+      "loss": 0.4752,
+      "step": 4393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004422746617184163,
+      "loss": 0.4573,
+      "step": 4394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004420730261676735,
+      "loss": 0.3939,
+      "step": 4395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044187140016516645,
+      "loss": 0.4179,
+      "step": 4396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004416697837441295,
+      "loss": 0.401,
+      "step": 4397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044146817693779556,
+      "loss": 0.4302,
+      "step": 4398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000441266579779396,
+      "loss": 0.4343,
+      "step": 4399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044106499230216054,
+      "loss": 0.4797,
+      "step": 4400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044086341453931715,
+      "loss": 0.4434,
+      "step": 4401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004406618465240926,
+      "loss": 0.4747,
+      "step": 4402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004404602882897116,
+      "loss": 0.4946,
+      "step": 4403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004402587398693976,
+      "loss": 0.4966,
+      "step": 4404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044005720129637214,
+      "loss": 0.4186,
+      "step": 4405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043985567260385565,
+      "loss": 0.3811,
+      "step": 4406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043965415382506606,
+      "loss": 0.3533,
+      "step": 4407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004394526449932204,
+      "loss": 0.4619,
+      "step": 4408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043925114614153386,
+      "loss": 0.4331,
+      "step": 4409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043904965730321994,
+      "loss": 0.4508,
+      "step": 4410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004388481785114905,
+      "loss": 0.454,
+      "step": 4411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004386467097995557,
+      "loss": 0.4493,
+      "step": 4412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004384452512006244,
+      "loss": 0.4403,
+      "step": 4413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000438243802747903,
+      "loss": 0.3679,
+      "step": 4414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043804236447459697,
+      "loss": 0.5186,
+      "step": 4415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043784093641390973,
+      "loss": 0.4619,
+      "step": 4416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043763951859904335,
+      "loss": 0.5448,
+      "step": 4417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004374381110631977,
+      "loss": 0.0515,
+      "step": 4418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043723671383957136,
+      "loss": 0.3906,
+      "step": 4419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043703532696136126,
+      "loss": 0.0515,
+      "step": 4420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004368339504617623,
+      "loss": 0.4154,
+      "step": 4421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004366325843739678,
+      "loss": 0.455,
+      "step": 4422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004364312287311694,
+      "loss": 0.5076,
+      "step": 4423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043622988356655705,
+      "loss": 0.4275,
+      "step": 4424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000436028548913319,
+      "loss": 0.3964,
+      "step": 4425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043582722480464155,
+      "loss": 0.5168,
+      "step": 4426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004356259112737095,
+      "loss": 0.469,
+      "step": 4427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004354246083537061,
+      "loss": 0.5031,
+      "step": 4428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004352233160778121,
+      "loss": 0.3906,
+      "step": 4429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043502203447920706,
+      "loss": 0.473,
+      "step": 4430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004348207635910689,
+      "loss": 0.436,
+      "step": 4431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043461950344657335,
+      "loss": 0.5774,
+      "step": 4432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043441825407889475,
+      "loss": 0.4513,
+      "step": 4433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043421701552120534,
+      "loss": 0.4893,
+      "step": 4434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004340157878066761,
+      "loss": 0.4094,
+      "step": 4435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043381457096847533,
+      "loss": 0.4376,
+      "step": 4436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043361336503977034,
+      "loss": 0.0508,
+      "step": 4437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043341217005372624,
+      "loss": 0.4713,
+      "step": 4438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043321098604350673,
+      "loss": 0.0508,
+      "step": 4439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004330098130422731,
+      "loss": 0.4371,
+      "step": 4440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043280865108318527,
+      "loss": 0.3979,
+      "step": 4441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043260750019940143,
+      "loss": 0.5277,
+      "step": 4442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043240636042407755,
+      "loss": 0.4011,
+      "step": 4443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043220523179036784,
+      "loss": 0.4451,
+      "step": 4444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043200411433142496,
+      "loss": 0.0506,
+      "step": 4445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004318030080803995,
+      "loss": 0.5045,
+      "step": 4446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043160191307044037,
+      "loss": 0.4766,
+      "step": 4447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043140082933469437,
+      "loss": 0.4874,
+      "step": 4448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004311997569063067,
+      "loss": 0.3914,
+      "step": 4449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043099869581842077,
+      "loss": 0.4349,
+      "step": 4450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004307976461041775,
+      "loss": 0.5286,
+      "step": 4451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004305966077967166,
+      "loss": 0.537,
+      "step": 4452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043039558092917567,
+      "loss": 0.4307,
+      "step": 4453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004301945655346905,
+      "loss": 0.4172,
+      "step": 4454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042999356164639474,
+      "loss": 0.4607,
+      "step": 4455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004297925692974206,
+      "loss": 0.4165,
+      "step": 4456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000429591588520898,
+      "loss": 0.4822,
+      "step": 4457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004293906193499548,
+      "loss": 0.438,
+      "step": 4458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042918966181771745,
+      "loss": 0.4934,
+      "step": 4459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042898871595731025,
+      "loss": 0.4163,
+      "step": 4460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042878778180185554,
+      "loss": 0.4091,
+      "step": 4461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042858685938447364,
+      "loss": 0.5542,
+      "step": 4462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004283859487382831,
+      "loss": 0.432,
+      "step": 4463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042818504989640085,
+      "loss": 0.4242,
+      "step": 4464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042798416289194094,
+      "loss": 0.5017,
+      "step": 4465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042778328775801625,
+      "loss": 0.4971,
+      "step": 4466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004275824245277375,
+      "loss": 0.4648,
+      "step": 4467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042738157323421347,
+      "loss": 0.4172,
+      "step": 4468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004271807339105509,
+      "loss": 0.4717,
+      "step": 4469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042697990658985464,
+      "loss": 0.4363,
+      "step": 4470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004267790913052275,
+      "loss": 0.446,
+      "step": 4471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004265782880897705,
+      "loss": 0.4061,
+      "step": 4472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042637749697658215,
+      "loss": 0.5404,
+      "step": 4473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042617671799875947,
+      "loss": 0.4333,
+      "step": 4474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042597595118939725,
+      "loss": 0.4659,
+      "step": 4475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042577519658158845,
+      "loss": 0.5228,
+      "step": 4476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004255744542084238,
+      "loss": 0.5746,
+      "step": 4477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042537372410299235,
+      "loss": 0.3813,
+      "step": 4478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004251730062983808,
+      "loss": 0.4213,
+      "step": 4479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042497230082767367,
+      "loss": 0.4281,
+      "step": 4480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000424771607723954,
+      "loss": 0.4656,
+      "step": 4481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042457092702030246,
+      "loss": 0.4852,
+      "step": 4482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042437025874979755,
+      "loss": 0.3732,
+      "step": 4483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000424169602945516,
+      "loss": 0.4475,
+      "step": 4484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004239689596405323,
+      "loss": 0.4691,
+      "step": 4485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004237683288679192,
+      "loss": 0.4237,
+      "step": 4486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042356771066074674,
+      "loss": 0.4783,
+      "step": 4487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004233671050520834,
+      "loss": 0.3708,
+      "step": 4488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004231665120749956,
+      "loss": 0.4546,
+      "step": 4489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004229659317625474,
+      "loss": 0.5497,
+      "step": 4490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042276536414780097,
+      "loss": 0.4028,
+      "step": 4491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042256480926381633,
+      "loss": 0.502,
+      "step": 4492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004223642671436516,
+      "loss": 0.4907,
+      "step": 4493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042216373782036224,
+      "loss": 0.4446,
+      "step": 4494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042196322132700205,
+      "loss": 0.4135,
+      "step": 4495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042176271769662273,
+      "loss": 0.4574,
+      "step": 4496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004215622269622737,
+      "loss": 0.4012,
+      "step": 4497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004213617491570023,
+      "loss": 0.4591,
+      "step": 4498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042116128431385374,
+      "loss": 0.468,
+      "step": 4499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042096083246587126,
+      "loss": 0.363,
+      "step": 4500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042076039364609565,
+      "loss": 0.4558,
+      "step": 4501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004205599678875656,
+      "loss": 0.4617,
+      "step": 4502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004203595552233179,
+      "loss": 0.381,
+      "step": 4503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000420159155686387,
+      "loss": 0.4877,
+      "step": 4504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004199587693098051,
+      "loss": 0.4138,
+      "step": 4505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004197583961266025,
+      "loss": 0.4304,
+      "step": 4506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041955803616980706,
+      "loss": 0.391,
+      "step": 4507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041935768947244477,
+      "loss": 0.4784,
+      "step": 4508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000419157356067539,
+      "loss": 0.4089,
+      "step": 4509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041895703598811117,
+      "loss": 0.384,
+      "step": 4510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041875672926718055,
+      "loss": 0.4402,
+      "step": 4511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004185564359377643,
+      "loss": 0.4841,
+      "step": 4512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000418356156032877,
+      "loss": 0.5243,
+      "step": 4513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041815588958553135,
+      "loss": 0.5312,
+      "step": 4514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041795563662873797,
+      "loss": 0.3755,
+      "step": 4515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004177553971955046,
+      "loss": 0.4365,
+      "step": 4516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004175551713188373,
+      "loss": 0.38,
+      "step": 4517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004173549590317398,
+      "loss": 0.4017,
+      "step": 4518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041715476036721357,
+      "loss": 0.4818,
+      "step": 4519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041695457535825776,
+      "loss": 0.4224,
+      "step": 4520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004167544040378695,
+      "loss": 0.4624,
+      "step": 4521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004165542464390432,
+      "loss": 0.5366,
+      "step": 4522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041635410259477146,
+      "loss": 0.4092,
+      "step": 4523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041615397253804445,
+      "loss": 0.4025,
+      "step": 4524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041595385630185006,
+      "loss": 0.4072,
+      "step": 4525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004157537539191738,
+      "loss": 0.4291,
+      "step": 4526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041555366542299913,
+      "loss": 0.4719,
+      "step": 4527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041535359084630697,
+      "loss": 0.5007,
+      "step": 4528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041515353022207623,
+      "loss": 0.4423,
+      "step": 4529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004149534835832835,
+      "loss": 0.4282,
+      "step": 4530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041475345096290246,
+      "loss": 0.4871,
+      "step": 4531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004145534323939052,
+      "loss": 0.4481,
+      "step": 4532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004143534279092612,
+      "loss": 0.3893,
+      "step": 4533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004141534375419378,
+      "loss": 0.3822,
+      "step": 4534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004139534613248997,
+      "loss": 0.4474,
+      "step": 4535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004137534992911095,
+      "loss": 0.4934,
+      "step": 4536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004135535514735276,
+      "loss": 0.4034,
+      "step": 4537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004133536179051114,
+      "loss": 0.5022,
+      "step": 4538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041315369861881675,
+      "loss": 0.4399,
+      "step": 4539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041295379364759667,
+      "loss": 0.4097,
+      "step": 4540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041275390302440197,
+      "loss": 0.05,
+      "step": 4541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004125540267821811,
+      "loss": 0.4753,
+      "step": 4542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004123541649538803,
+      "loss": 0.4042,
+      "step": 4543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004121543175724429,
+      "loss": 0.3921,
+      "step": 4544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004119544846708106,
+      "loss": 0.3745,
+      "step": 4545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041175466628192185,
+      "loss": 0.5004,
+      "step": 4546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041155486243871366,
+      "loss": 0.5331,
+      "step": 4547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041135507317411967,
+      "loss": 0.4211,
+      "step": 4548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004111552985210719,
+      "loss": 0.5339,
+      "step": 4549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004109555385124997,
+      "loss": 0.0497,
+      "step": 4550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004107557931813297,
+      "loss": 0.4003,
+      "step": 4551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041055606256048687,
+      "loss": 0.6439,
+      "step": 4552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004103563466828928,
+      "loss": 0.4929,
+      "step": 4553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041015664558146714,
+      "loss": 0.4461,
+      "step": 4554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004099569592891273,
+      "loss": 0.4839,
+      "step": 4555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040975728783878787,
+      "loss": 0.4513,
+      "step": 4556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040955763126336125,
+      "loss": 0.3914,
+      "step": 4557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004093579895957573,
+      "loss": 0.4923,
+      "step": 4558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040915836286888345,
+      "loss": 0.4811,
+      "step": 4559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040895875111564446,
+      "loss": 0.433,
+      "step": 4560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040875915436894283,
+      "loss": 0.4895,
+      "step": 4561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004085595726616787,
+      "loss": 0.4991,
+      "step": 4562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040836000602674933,
+      "loss": 0.4254,
+      "step": 4563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040816045449705,
+      "loss": 0.4254,
+      "step": 4564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004079609181054731,
+      "loss": 0.4158,
+      "step": 4565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040776139688490876,
+      "loss": 0.4585,
+      "step": 4566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040756189086824454,
+      "loss": 0.0493,
+      "step": 4567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004073624000883653,
+      "loss": 0.5289,
+      "step": 4568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040716292457815376,
+      "loss": 0.3812,
+      "step": 4569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004069634643704897,
+      "loss": 0.476,
+      "step": 4570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004067640194982508,
+      "loss": 0.4246,
+      "step": 4571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004065645899943119,
+      "loss": 0.4091,
+      "step": 4572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004063651758915454,
+      "loss": 0.4057,
+      "step": 4573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004061657772228216,
+      "loss": 0.5015,
+      "step": 4574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040596639402100724,
+      "loss": 0.416,
+      "step": 4575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004057670263189673,
+      "loss": 0.416,
+      "step": 4576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040556767414956415,
+      "loss": 0.4318,
+      "step": 4577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004053683375456573,
+      "loss": 0.4788,
+      "step": 4578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040516901654010407,
+      "loss": 0.4138,
+      "step": 4579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040496971116575875,
+      "loss": 0.457,
+      "step": 4580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004047704214554737,
+      "loss": 0.489,
+      "step": 4581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040457114744209786,
+      "loss": 0.4113,
+      "step": 4582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004043718891584781,
+      "loss": 0.4092,
+      "step": 4583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004041726466374588,
+      "loss": 0.4644,
+      "step": 4584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040397341991188143,
+      "loss": 0.4399,
+      "step": 4585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040377420901458507,
+      "loss": 0.406,
+      "step": 4586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040357501397840604,
+      "loss": 0.4149,
+      "step": 4587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004033758348361781,
+      "loss": 0.4551,
+      "step": 4588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004031766716207325,
+      "loss": 0.3795,
+      "step": 4589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040297752436489775,
+      "loss": 0.0506,
+      "step": 4590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040277839310149966,
+      "loss": 0.541,
+      "step": 4591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040257927786336147,
+      "loss": 0.5076,
+      "step": 4592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040238017868330386,
+      "loss": 0.4829,
+      "step": 4593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040218109559414484,
+      "loss": 0.5112,
+      "step": 4594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004019820286286997,
+      "loss": 0.3831,
+      "step": 4595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040178297781978136,
+      "loss": 0.4409,
+      "step": 4596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040158394320019927,
+      "loss": 0.394,
+      "step": 4597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040138492480276114,
+      "loss": 0.5149,
+      "step": 4598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004011859226602715,
+      "loss": 0.444,
+      "step": 4599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040098693680553233,
+      "loss": 0.4125,
+      "step": 4600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040078796727134304,
+      "loss": 0.4531,
+      "step": 4601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004005890140905001,
+      "loss": 0.4795,
+      "step": 4602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004003900772957978,
+      "loss": 0.5053,
+      "step": 4603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040019115692002677,
+      "loss": 0.4536,
+      "step": 4604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039999225299597573,
+      "loss": 0.474,
+      "step": 4605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003997933655564305,
+      "loss": 0.4037,
+      "step": 4606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039959449463417425,
+      "loss": 0.4128,
+      "step": 4607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003993956402619873,
+      "loss": 0.4575,
+      "step": 4608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039919680247264713,
+      "loss": 0.0502,
+      "step": 4609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003989979812989287,
+      "loss": 0.4254,
+      "step": 4610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039879917677360425,
+      "loss": 0.4793,
+      "step": 4611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039860038892944305,
+      "loss": 0.5176,
+      "step": 4612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039840161779921174,
+      "loss": 0.499,
+      "step": 4613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039820286341567436,
+      "loss": 0.4298,
+      "step": 4614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003980041258115919,
+      "loss": 0.4686,
+      "step": 4615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039780540501972283,
+      "loss": 0.0501,
+      "step": 4616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003976067010728227,
+      "loss": 0.4554,
+      "step": 4617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003974080140036447,
+      "loss": 0.4001,
+      "step": 4618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039720934384493824,
+      "loss": 0.4991,
+      "step": 4619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039701069062945087,
+      "loss": 0.453,
+      "step": 4620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003968120543899271,
+      "loss": 0.415,
+      "step": 4621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003966134351591086,
+      "loss": 0.0499,
+      "step": 4622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039641483296973414,
+      "loss": 0.5049,
+      "step": 4623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003962162478545398,
+      "loss": 0.4492,
+      "step": 4624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003960176798462591,
+      "loss": 0.4403,
+      "step": 4625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003958191289776221,
+      "loss": 0.4402,
+      "step": 4626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039562059528135634,
+      "loss": 0.4247,
+      "step": 4627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003954220787901868,
+      "loss": 0.4404,
+      "step": 4628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039522357953683545,
+      "loss": 0.4658,
+      "step": 4629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003950250975540212,
+      "loss": 0.4639,
+      "step": 4630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039482663287446033,
+      "loss": 0.4258,
+      "step": 4631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039462818553086644,
+      "loss": 0.4203,
+      "step": 4632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039442975555594974,
+      "loss": 0.4257,
+      "step": 4633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003942313429824181,
+      "loss": 0.3916,
+      "step": 4634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039403294784297625,
+      "loss": 0.4569,
+      "step": 4635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039383457017032607,
+      "loss": 0.5515,
+      "step": 4636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003936362099971666,
+      "loss": 0.4176,
+      "step": 4637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003934378673561942,
+      "loss": 0.4034,
+      "step": 4638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039323954228010193,
+      "loss": 0.397,
+      "step": 4639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003930412348015804,
+      "loss": 0.4148,
+      "step": 4640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039284294495331684,
+      "loss": 0.4119,
+      "step": 4641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003926446727679958,
+      "loss": 0.0493,
+      "step": 4642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039244641827829906,
+      "loss": 0.4553,
+      "step": 4643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003922481815169054,
+      "loss": 0.4734,
+      "step": 4644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039204996251649056,
+      "loss": 0.4125,
+      "step": 4645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003918517613097274,
+      "loss": 0.392,
+      "step": 4646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039165357792928624,
+      "loss": 0.4017,
+      "step": 4647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003914554124078335,
+      "loss": 0.4425,
+      "step": 4648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003912572647780336,
+      "loss": 0.4429,
+      "step": 4649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039105913507254757,
+      "loss": 0.4971,
+      "step": 4650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003908610233240338,
+      "loss": 0.4893,
+      "step": 4651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039066292956514714,
+      "loss": 0.4556,
+      "step": 4652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039046485382854013,
+      "loss": 0.3999,
+      "step": 4653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000390266796146862,
+      "loss": 0.405,
+      "step": 4654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039006875655275895,
+      "loss": 0.3629,
+      "step": 4655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038987073507887435,
+      "loss": 0.4788,
+      "step": 4656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038967273175784847,
+      "loss": 0.4314,
+      "step": 4657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003894747466223187,
+      "loss": 0.4595,
+      "step": 4658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003892767797049194,
+      "loss": 0.3857,
+      "step": 4659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000389078831038282,
+      "loss": 0.4673,
+      "step": 4660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038888090065503463,
+      "loss": 0.4091,
+      "step": 4661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038868298858780293,
+      "loss": 0.4005,
+      "step": 4662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003884850948692088,
+      "loss": 0.4575,
+      "step": 4663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003882872195318717,
+      "loss": 0.3966,
+      "step": 4664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038808936260840786,
+      "loss": 0.3876,
+      "step": 4665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003878915241314305,
+      "loss": 0.4913,
+      "step": 4666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003876937041335498,
+      "loss": 0.451,
+      "step": 4667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003874959026473729,
+      "loss": 0.0488,
+      "step": 4668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000387298119705504,
+      "loss": 0.4553,
+      "step": 4669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038710035534054375,
+      "loss": 0.4955,
+      "step": 4670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038690260958509036,
+      "loss": 0.4804,
+      "step": 4671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038670488247173867,
+      "loss": 0.3632,
+      "step": 4672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003865071740330806,
+      "loss": 0.4414,
+      "step": 4673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038630948430170463,
+      "loss": 0.3784,
+      "step": 4674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003861118133101966,
+      "loss": 0.5094,
+      "step": 4675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003859141610911391,
+      "loss": 0.4554,
+      "step": 4676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038571652767711156,
+      "loss": 0.4081,
+      "step": 4677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003855189131006902,
+      "loss": 0.4052,
+      "step": 4678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003853213173944484,
+      "loss": 0.4242,
+      "step": 4679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003851237405909563,
+      "loss": 0.3244,
+      "step": 4680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003849261827227809,
+      "loss": 0.5229,
+      "step": 4681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038472864382248624,
+      "loss": 0.4326,
+      "step": 4682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038453112392263303,
+      "loss": 0.413,
+      "step": 4683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038433362305577906,
+      "loss": 0.4337,
+      "step": 4684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038413614125447853,
+      "loss": 0.5012,
+      "step": 4685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038393867855128304,
+      "loss": 0.4814,
+      "step": 4686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003837412349787408,
+      "loss": 0.3668,
+      "step": 4687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003835438105693968,
+      "loss": 0.4418,
+      "step": 4688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003833464053557931,
+      "loss": 0.5103,
+      "step": 4689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003831490193704685,
+      "loss": 0.3942,
+      "step": 4690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003829516526459584,
+      "loss": 0.4574,
+      "step": 4691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003827543052147952,
+      "loss": 0.5529,
+      "step": 4692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003825569771095082,
+      "loss": 0.4319,
+      "step": 4693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003823596683626235,
+      "loss": 0.5342,
+      "step": 4694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038216237900666385,
+      "loss": 0.475,
+      "step": 4695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003819651090741489,
+      "loss": 0.4669,
+      "step": 4696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003817678585975953,
+      "loss": 0.3856,
+      "step": 4697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038157062760951624,
+      "loss": 0.4452,
+      "step": 4698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003813734161424215,
+      "loss": 0.4628,
+      "step": 4699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000381176224228818,
+      "loss": 0.4185,
+      "step": 4700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003809790519012095,
+      "loss": 0.4611,
+      "step": 4701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003807818991920963,
+      "loss": 0.51,
+      "step": 4702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003805847661339754,
+      "loss": 0.3998,
+      "step": 4703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038038765275934076,
+      "loss": 0.38,
+      "step": 4704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038019055910068315,
+      "loss": 0.0487,
+      "step": 4705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037999348519049007,
+      "loss": 0.4938,
+      "step": 4706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003797964310612453,
+      "loss": 0.428,
+      "step": 4707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037959939674542985,
+      "loss": 0.4398,
+      "step": 4708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003794023822755214,
+      "loss": 0.4808,
+      "step": 4709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037920538768399425,
+      "loss": 0.3888,
+      "step": 4710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037900841300331955,
+      "loss": 0.4543,
+      "step": 4711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037881145826596516,
+      "loss": 0.3523,
+      "step": 4712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003786145235043954,
+      "loss": 0.3785,
+      "step": 4713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037841760875107145,
+      "loss": 0.4406,
+      "step": 4714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003782207140384514,
+      "loss": 0.4691,
+      "step": 4715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003780238393989897,
+      "loss": 0.4229,
+      "step": 4716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037782698486513774,
+      "loss": 0.4456,
+      "step": 4717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003776301504693434,
+      "loss": 0.3795,
+      "step": 4718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003774333362440514,
+      "loss": 0.3776,
+      "step": 4719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037723654222170333,
+      "loss": 0.4683,
+      "step": 4720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003770397684347368,
+      "loss": 0.4814,
+      "step": 4721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003768430149155866,
+      "loss": 0.4231,
+      "step": 4722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003766462816966841,
+      "loss": 0.4053,
+      "step": 4723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037644956881045734,
+      "loss": 0.4348,
+      "step": 4724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003762528762893309,
+      "loss": 0.4205,
+      "step": 4725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037605620416572604,
+      "loss": 0.4061,
+      "step": 4726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037585955247206076,
+      "loss": 0.4311,
+      "step": 4727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003756629212407497,
+      "loss": 0.4456,
+      "step": 4728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037546631050420376,
+      "loss": 0.5055,
+      "step": 4729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037526972029483085,
+      "loss": 0.5042,
+      "step": 4730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037507315064503536,
+      "loss": 0.424,
+      "step": 4731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003748766015872184,
+      "loss": 0.383,
+      "step": 4732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003746800731537776,
+      "loss": 0.4183,
+      "step": 4733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037448356537710716,
+      "loss": 0.4331,
+      "step": 4734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037428707828959795,
+      "loss": 0.5254,
+      "step": 4735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037409061192363724,
+      "loss": 0.4712,
+      "step": 4736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003738941663116091,
+      "loss": 0.4406,
+      "step": 4737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037369774148589426,
+      "loss": 0.4711,
+      "step": 4738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003735013374788696,
+      "loss": 0.4512,
+      "step": 4739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003733049543229091,
+      "loss": 0.4165,
+      "step": 4740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003731085920503828,
+      "loss": 0.5345,
+      "step": 4741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000372912250693658,
+      "loss": 0.3865,
+      "step": 4742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037271593028509746,
+      "loss": 0.5317,
+      "step": 4743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037251963085706155,
+      "loss": 0.401,
+      "step": 4744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037232335244190653,
+      "loss": 0.4355,
+      "step": 4745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003721270950719856,
+      "loss": 0.5284,
+      "step": 4746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003719308587796482,
+      "loss": 0.4317,
+      "step": 4747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003717346435972405,
+      "loss": 0.4202,
+      "step": 4748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003715384495571052,
+      "loss": 0.4417,
+      "step": 4749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003713422766915811,
+      "loss": 0.4272,
+      "step": 4750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037114612503300396,
+      "loss": 0.3667,
+      "step": 4751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037094999461370593,
+      "loss": 0.4862,
+      "step": 4752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037075388546601565,
+      "loss": 0.4659,
+      "step": 4753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037055779762225816,
+      "loss": 0.4377,
+      "step": 4754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003703617311147553,
+      "loss": 0.4131,
+      "step": 4755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003701656859758249,
+      "loss": 0.4655,
+      "step": 4756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036996966223778164,
+      "loss": 0.3933,
+      "step": 4757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036977365993293653,
+      "loss": 0.4165,
+      "step": 4758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036957767909359717,
+      "loss": 0.4546,
+      "step": 4759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036938171975206736,
+      "loss": 0.5145,
+      "step": 4760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036918578194064755,
+      "loss": 0.4755,
+      "step": 4761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036898986569163476,
+      "loss": 0.4036,
+      "step": 4762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036879397103732224,
+      "loss": 0.5042,
+      "step": 4763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036859809800999995,
+      "loss": 0.575,
+      "step": 4764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036840224664195365,
+      "loss": 0.4332,
+      "step": 4765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003682064169654663,
+      "loss": 0.4053,
+      "step": 4766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003680106090128168,
+      "loss": 0.5342,
+      "step": 4767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036781482281628066,
+      "loss": 0.502,
+      "step": 4768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003676190584081298,
+      "loss": 0.4288,
+      "step": 4769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003674233158206326,
+      "loss": 0.3969,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036722759508605387,
+      "loss": 0.5208,
+      "step": 4771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003670318962366542,
+      "loss": 0.5278,
+      "step": 4772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003668362193046915,
+      "loss": 0.4077,
+      "step": 4773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003666405643224194,
+      "loss": 0.4751,
+      "step": 4774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036644493132208846,
+      "loss": 0.5997,
+      "step": 4775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000366249320335945,
+      "loss": 0.3765,
+      "step": 4776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036605373139623234,
+      "loss": 0.3718,
+      "step": 4777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003658581645351896,
+      "loss": 0.4807,
+      "step": 4778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036566261978505265,
+      "loss": 0.4064,
+      "step": 4779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003654670971780535,
+      "loss": 0.4825,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036527159674642064,
+      "loss": 0.4795,
+      "step": 4781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036507611852237875,
+      "loss": 0.4419,
+      "step": 4782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003648806625381491,
+      "loss": 0.4235,
+      "step": 4783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036468522882594903,
+      "loss": 0.3806,
+      "step": 4784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036448981741799236,
+      "loss": 0.3513,
+      "step": 4785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003642944283464894,
+      "loss": 0.5254,
+      "step": 4786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036409906164364624,
+      "loss": 0.4818,
+      "step": 4787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036390371734166583,
+      "loss": 0.5085,
+      "step": 4788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000363708395472747,
+      "loss": 0.4758,
+      "step": 4789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036351309606908545,
+      "loss": 0.4047,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036331781916287256,
+      "loss": 0.4414,
+      "step": 4791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003631225647862963,
+      "loss": 0.4629,
+      "step": 4792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003629273329715412,
+      "loss": 0.4564,
+      "step": 4793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036273212375078734,
+      "loss": 0.4595,
+      "step": 4794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036253693715621163,
+      "loss": 0.4694,
+      "step": 4795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003623417732199872,
+      "loss": 0.4474,
+      "step": 4796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003621466319742833,
+      "loss": 0.3984,
+      "step": 4797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036195151345126556,
+      "loss": 0.5023,
+      "step": 4798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036175641768309587,
+      "loss": 0.3813,
+      "step": 4799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003615613447019322,
+      "loss": 0.3942,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036136629453992897,
+      "loss": 0.0494,
+      "step": 4801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036117126722923665,
+      "loss": 0.3972,
+      "step": 4802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003609762628020021,
+      "loss": 0.4237,
+      "step": 4803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036078128129036824,
+      "loss": 0.4875,
+      "step": 4804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003605863227264745,
+      "loss": 0.4484,
+      "step": 4805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003603913871424561,
+      "loss": 0.3777,
+      "step": 4806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000360196474570445,
+      "loss": 0.416,
+      "step": 4807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036000158504256906,
+      "loss": 0.4287,
+      "step": 4808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035980671859095215,
+      "loss": 0.4099,
+      "step": 4809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003596118752477146,
+      "loss": 0.4266,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000359417055044973,
+      "loss": 0.4519,
+      "step": 4811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003592222580148399,
+      "loss": 0.3714,
+      "step": 4812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003590274841894242,
+      "loss": 0.4305,
+      "step": 4813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035883273360083097,
+      "loss": 0.3733,
+      "step": 4814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003586380062811615,
+      "loss": 0.468,
+      "step": 4815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035844330226251277,
+      "loss": 0.4275,
+      "step": 4816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003582486215769785,
+      "loss": 0.4028,
+      "step": 4817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003580539642566484,
+      "loss": 0.5076,
+      "step": 4818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035785933033360815,
+      "loss": 0.465,
+      "step": 4819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035766471983994,
+      "loss": 0.4576,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003574701328077216,
+      "loss": 0.4001,
+      "step": 4821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003572755692690275,
+      "loss": 0.0493,
+      "step": 4822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000357081029255928,
+      "loss": 0.4,
+      "step": 4823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035688651280048957,
+      "loss": 0.4102,
+      "step": 4824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003566920199347747,
+      "loss": 0.5032,
+      "step": 4825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003564975506908421,
+      "loss": 0.4301,
+      "step": 4826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003563031051007467,
+      "loss": 0.4327,
+      "step": 4827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035610868319653927,
+      "loss": 0.5806,
+      "step": 4828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035591428501026694,
+      "loss": 0.5173,
+      "step": 4829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000355719910573973,
+      "loss": 0.414,
+      "step": 4830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003555255599196962,
+      "loss": 0.4297,
+      "step": 4831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000355331233079472,
+      "loss": 0.4745,
+      "step": 4832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003551369300853317,
+      "loss": 0.3922,
+      "step": 4833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003549426509693028,
+      "loss": 0.4576,
+      "step": 4834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003547483957634087,
+      "loss": 0.4891,
+      "step": 4835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003545541644996689,
+      "loss": 0.4656,
+      "step": 4836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003543599572100994,
+      "loss": 0.459,
+      "step": 4837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003541657739267111,
+      "loss": 0.4613,
+      "step": 4838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003539716146815122,
+      "loss": 0.4337,
+      "step": 4839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035377747950650606,
+      "loss": 0.4479,
+      "step": 4840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035358336843369277,
+      "loss": 0.5139,
+      "step": 4841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000353389281495068,
+      "loss": 0.3531,
+      "step": 4842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003531952187226234,
+      "loss": 0.4159,
+      "step": 4843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035300118014834685,
+      "loss": 0.5668,
+      "step": 4844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035280716580422236,
+      "loss": 0.4205,
+      "step": 4845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003526131757222296,
+      "loss": 0.4307,
+      "step": 4846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003524192099343443,
+      "loss": 0.5321,
+      "step": 4847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003522252684725383,
+      "loss": 0.4026,
+      "step": 4848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035203135136877955,
+      "loss": 0.439,
+      "step": 4849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003518374586550318,
+      "loss": 0.4711,
+      "step": 4850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003516435903632548,
+      "loss": 0.4752,
+      "step": 4851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035144974652540463,
+      "loss": 0.437,
+      "step": 4852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003512559271734324,
+      "loss": 0.5759,
+      "step": 4853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003510621323392862,
+      "loss": 0.6102,
+      "step": 4854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003508683620549095,
+      "loss": 0.5643,
+      "step": 4855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035067461635224207,
+      "loss": 0.4391,
+      "step": 4856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035048089526321937,
+      "loss": 0.4964,
+      "step": 4857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035028719881977284,
+      "loss": 0.4449,
+      "step": 4858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003500935270538302,
+      "loss": 0.4838,
+      "step": 4859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003498998799973143,
+      "loss": 0.4604,
+      "step": 4860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003497062576821448,
+      "loss": 0.4998,
+      "step": 4861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003495126601402368,
+      "loss": 0.5089,
+      "step": 4862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003493190874035015,
+      "loss": 0.4258,
+      "step": 4863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034912553950384583,
+      "loss": 0.4149,
+      "step": 4864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034893201647317275,
+      "loss": 0.55,
+      "step": 4865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003487385183433812,
+      "loss": 0.4734,
+      "step": 4866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003485450451463659,
+      "loss": 0.4071,
+      "step": 4867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003483515969140173,
+      "loss": 0.4692,
+      "step": 4868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003481581736782221,
+      "loss": 0.4178,
+      "step": 4869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003479647754708626,
+      "loss": 0.3692,
+      "step": 4870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003477714023238171,
+      "loss": 0.4663,
+      "step": 4871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034757805426895974,
+      "loss": 0.4778,
+      "step": 4872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003473847313381605,
+      "loss": 0.4114,
+      "step": 4873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034719143356328554,
+      "loss": 0.0491,
+      "step": 4874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003469981609761961,
+      "loss": 0.3748,
+      "step": 4875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003468049136087499,
+      "loss": 0.049,
+      "step": 4876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003466116914928005,
+      "loss": 0.4454,
+      "step": 4877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034641849466019695,
+      "loss": 0.4908,
+      "step": 4878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003462253231427844,
+      "loss": 0.4155,
+      "step": 4879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003460321769724038,
+      "loss": 0.3748,
+      "step": 4880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034583905618089204,
+      "loss": 0.4555,
+      "step": 4881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034564596080008126,
+      "loss": 0.4261,
+      "step": 4882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034545289086180007,
+      "loss": 0.4496,
+      "step": 4883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003452598463978726,
+      "loss": 0.4591,
+      "step": 4884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034506682744011875,
+      "loss": 0.4258,
+      "step": 4885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034487383402035434,
+      "loss": 0.0485,
+      "step": 4886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003446808661703908,
+      "loss": 0.4886,
+      "step": 4887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034448792392203576,
+      "loss": 0.4749,
+      "step": 4888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034429500730709205,
+      "loss": 0.4865,
+      "step": 4889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034410211635735853,
+      "loss": 0.4387,
+      "step": 4890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034390925110463,
+      "loss": 0.3945,
+      "step": 4891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034371641158069687,
+      "loss": 0.3948,
+      "step": 4892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034352359781734525,
+      "loss": 0.3683,
+      "step": 4893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003433308098463571,
+      "loss": 0.3467,
+      "step": 4894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003431380476995102,
+      "loss": 0.4847,
+      "step": 4895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034294531140857806,
+      "loss": 0.5275,
+      "step": 4896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003427526010053295,
+      "loss": 0.433,
+      "step": 4897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003425599165215295,
+      "loss": 0.3917,
+      "step": 4898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034236725798893886,
+      "loss": 0.4646,
+      "step": 4899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003421746254393138,
+      "loss": 0.418,
+      "step": 4900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003419820189044064,
+      "loss": 0.616,
+      "step": 4901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034178943841596456,
+      "loss": 0.4122,
+      "step": 4902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034159688400573184,
+      "loss": 0.3863,
+      "step": 4903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034140435570544704,
+      "loss": 0.3461,
+      "step": 4904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034121185354684525,
+      "loss": 0.4534,
+      "step": 4905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003410193775616571,
+      "loss": 0.4225,
+      "step": 4906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034082692778160883,
+      "loss": 0.4318,
+      "step": 4907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034063450423842226,
+      "loss": 0.3375,
+      "step": 4908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003404421069638151,
+      "loss": 0.4082,
+      "step": 4909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003402497359895007,
+      "loss": 0.4232,
+      "step": 4910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034005739134718795,
+      "loss": 0.4781,
+      "step": 4911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033986507306858125,
+      "loss": 0.4343,
+      "step": 4912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033967278118538115,
+      "loss": 0.4056,
+      "step": 4913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003394805157292834,
+      "loss": 0.453,
+      "step": 4914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003392882767319796,
+      "loss": 0.4373,
+      "step": 4915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033909606422515703,
+      "loss": 0.4819,
+      "step": 4916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003389038782404984,
+      "loss": 0.4457,
+      "step": 4917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003387117188096823,
+      "loss": 0.4399,
+      "step": 4918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003385195859643827,
+      "loss": 0.4285,
+      "step": 4919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003383274797362692,
+      "loss": 0.4716,
+      "step": 4920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003381354001570073,
+      "loss": 0.416,
+      "step": 4921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003379433472582577,
+      "loss": 0.5067,
+      "step": 4922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003377513210716772,
+      "loss": 0.4165,
+      "step": 4923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003375593216289178,
+      "loss": 0.4396,
+      "step": 4924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003373673489616272,
+      "loss": 0.3931,
+      "step": 4925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033717540310144854,
+      "loss": 0.418,
+      "step": 4926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033698348408002087,
+      "loss": 0.4435,
+      "step": 4927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033679159192897873,
+      "loss": 0.4363,
+      "step": 4928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033659972667995187,
+      "loss": 0.4301,
+      "step": 4929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033640788836456604,
+      "loss": 0.4346,
+      "step": 4930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033621607701444237,
+      "loss": 0.4031,
+      "step": 4931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003360242926611978,
+      "loss": 0.397,
+      "step": 4932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003358325353364441,
+      "loss": 0.4529,
+      "step": 4933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003356408050717893,
+      "loss": 0.3838,
+      "step": 4934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033544910189883684,
+      "loss": 0.4431,
+      "step": 4935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003352574258491854,
+      "loss": 0.4972,
+      "step": 4936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003350657769544295,
+      "loss": 0.5159,
+      "step": 4937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033487415524615917,
+      "loss": 0.4529,
+      "step": 4938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033468256075595967,
+      "loss": 0.442,
+      "step": 4939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033449099351541225,
+      "loss": 0.5005,
+      "step": 4940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033429945355609294,
+      "loss": 0.4729,
+      "step": 4941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003341079409095739,
+      "loss": 0.4203,
+      "step": 4942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033391645560742263,
+      "loss": 0.3762,
+      "step": 4943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003337249976812021,
+      "loss": 0.4445,
+      "step": 4944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033353356716247075,
+      "loss": 0.3721,
+      "step": 4945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003333421640827826,
+      "loss": 0.4678,
+      "step": 4946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033315078847368694,
+      "loss": 0.4379,
+      "step": 4947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033295944036672867,
+      "loss": 0.4246,
+      "step": 4948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000332768119793448,
+      "loss": 0.4006,
+      "step": 4949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033257682678538105,
+      "loss": 0.4663,
+      "step": 4950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033238556137405877,
+      "loss": 0.4417,
+      "step": 4951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033219432359100806,
+      "loss": 0.424,
+      "step": 4952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000332003113467751,
+      "loss": 0.3928,
+      "step": 4953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003318119310358054,
+      "loss": 0.5938,
+      "step": 4954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033162077632668395,
+      "loss": 0.4241,
+      "step": 4955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033142964937189534,
+      "loss": 0.4392,
+      "step": 4956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003312385502029434,
+      "loss": 0.3895,
+      "step": 4957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033104747885132746,
+      "loss": 0.4686,
+      "step": 4958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033085643534854224,
+      "loss": 0.4597,
+      "step": 4959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003306654197260779,
+      "loss": 0.4713,
+      "step": 4960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033047443201541995,
+      "loss": 0.4041,
+      "step": 4961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033028347224804954,
+      "loss": 0.4158,
+      "step": 4962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003300925404554427,
+      "loss": 0.4628,
+      "step": 4963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032990163666907115,
+      "loss": 0.3635,
+      "step": 4964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003297107609204022,
+      "loss": 0.4728,
+      "step": 4965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032951991324089823,
+      "loss": 0.424,
+      "step": 4966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003293290936620171,
+      "loss": 0.3872,
+      "step": 4967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032913830221521225,
+      "loss": 0.4453,
+      "step": 4968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032894753893193197,
+      "loss": 0.448,
+      "step": 4969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003287568038436203,
+      "loss": 0.4366,
+      "step": 4970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003285660969817165,
+      "loss": 0.4747,
+      "step": 4971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032837541837765544,
+      "loss": 0.4734,
+      "step": 4972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003281847680628668,
+      "loss": 0.3649,
+      "step": 4973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032799414606877607,
+      "loss": 0.5847,
+      "step": 4974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032780355242680395,
+      "loss": 0.4454,
+      "step": 4975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032761298716836655,
+      "loss": 0.4473,
+      "step": 4976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032742245032487484,
+      "loss": 0.5132,
+      "step": 4977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032723194192773556,
+      "loss": 0.5432,
+      "step": 4978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003270414620083507,
+      "loss": 0.4832,
+      "step": 4979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003268510105981175,
+      "loss": 0.4255,
+      "step": 4980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003266605877284284,
+      "loss": 0.0525,
+      "step": 4981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032647019343067153,
+      "loss": 0.4224,
+      "step": 4982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003262798277362299,
+      "loss": 0.5188,
+      "step": 4983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032608949067648165,
+      "loss": 0.4019,
+      "step": 4984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032589918228280066,
+      "loss": 0.4495,
+      "step": 4985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032570890258655597,
+      "loss": 0.4567,
+      "step": 4986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032551865161911165,
+      "loss": 0.442,
+      "step": 4987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003253284294118274,
+      "loss": 0.4127,
+      "step": 4988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032513823599605797,
+      "loss": 0.422,
+      "step": 4989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003249480714031531,
+      "loss": 0.4594,
+      "step": 4990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032475793566445846,
+      "loss": 0.0508,
+      "step": 4991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003245678288113142,
+      "loss": 0.4375,
+      "step": 4992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003243777508750563,
+      "loss": 0.4882,
+      "step": 4993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003241877018870156,
+      "loss": 0.4202,
+      "step": 4994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003239976818785183,
+      "loss": 0.5071,
+      "step": 4995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032380769088088604,
+      "loss": 0.3965,
+      "step": 4996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032361772892543515,
+      "loss": 0.3671,
+      "step": 4997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003234277960434779,
+      "loss": 0.4399,
+      "step": 4998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032323789226632103,
+      "loss": 0.4247,
+      "step": 4999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003230480176252669,
+      "loss": 0.4189,
+      "step": 5000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000322858172151613,
+      "loss": 0.4189,
+      "step": 5001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000322668355876652,
+      "loss": 0.4545,
+      "step": 5002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003224785688316717,
+      "loss": 0.3977,
+      "step": 5003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032228881104795514,
+      "loss": 0.478,
+      "step": 5004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032209908255678086,
+      "loss": 0.3691,
+      "step": 5005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032190938338942177,
+      "loss": 0.4774,
+      "step": 5006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003217197135771465,
+      "loss": 0.4645,
+      "step": 5007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032153007315121896,
+      "loss": 0.4442,
+      "step": 5008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003213404621428978,
+      "loss": 0.3787,
+      "step": 5009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003211508805834372,
+      "loss": 0.4075,
+      "step": 5010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032096132850408643,
+      "loss": 0.4379,
+      "step": 5011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003207718059360895,
+      "loss": 0.4532,
+      "step": 5012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000320582312910686,
+      "loss": 0.4882,
+      "step": 5013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003203928494591105,
+      "loss": 0.4089,
+      "step": 5014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003202034156125927,
+      "loss": 0.4004,
+      "step": 5015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032001401140235733,
+      "loss": 0.4285,
+      "step": 5016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031982463685962436,
+      "loss": 0.4757,
+      "step": 5017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003196352920156088,
+      "loss": 0.3783,
+      "step": 5018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031944597690152096,
+      "loss": 0.4548,
+      "step": 5019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031925669154856605,
+      "loss": 0.5152,
+      "step": 5020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003190674359879442,
+      "loss": 0.4319,
+      "step": 5021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031887821025085094,
+      "loss": 0.3838,
+      "step": 5022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003186890143684769,
+      "loss": 0.5863,
+      "step": 5023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003184998483720074,
+      "loss": 0.4634,
+      "step": 5024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031831071229262336,
+      "loss": 0.4139,
+      "step": 5025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031812160616150043,
+      "loss": 0.4132,
+      "step": 5026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003179325300098096,
+      "loss": 0.4578,
+      "step": 5027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031774348386871636,
+      "loss": 0.4401,
+      "step": 5028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031755446776938177,
+      "loss": 0.4578,
+      "step": 5029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031736548174296183,
+      "loss": 0.5036,
+      "step": 5030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031717652582060745,
+      "loss": 0.4501,
+      "step": 5031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003169876000334649,
+      "loss": 0.5502,
+      "step": 5032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003167987044126749,
+      "loss": 0.5245,
+      "step": 5033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031660983898937383,
+      "loss": 0.4768,
+      "step": 5034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031642100379469275,
+      "loss": 0.376,
+      "step": 5035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003162321988597577,
+      "loss": 0.425,
+      "step": 5036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031604342421569,
+      "loss": 0.4645,
+      "step": 5037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003158546798936055,
+      "loss": 0.4612,
+      "step": 5038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003156659659246156,
+      "loss": 0.5502,
+      "step": 5039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003154772823398264,
+      "loss": 0.5035,
+      "step": 5040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031528862917033914,
+      "loss": 0.3796,
+      "step": 5041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031510000644725003,
+      "loss": 0.449,
+      "step": 5042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003149114142016497,
+      "loss": 0.5188,
+      "step": 5043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031472285246462474,
+      "loss": 0.4545,
+      "step": 5044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003145343212672559,
+      "loss": 0.406,
+      "step": 5045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031434582064061937,
+      "loss": 0.4575,
+      "step": 5046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003141573506157862,
+      "loss": 0.3873,
+      "step": 5047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031396891122382223,
+      "loss": 0.3674,
+      "step": 5048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031378050249578846,
+      "loss": 0.5109,
+      "step": 5049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031359212446274053,
+      "loss": 0.4295,
+      "step": 5050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031340377715572944,
+      "loss": 0.5692,
+      "step": 5051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003132154606058006,
+      "loss": 0.4392,
+      "step": 5052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003130271748439951,
+      "loss": 0.4753,
+      "step": 5053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003128389199013482,
+      "loss": 0.6003,
+      "step": 5054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031265069580889047,
+      "loss": 0.4382,
+      "step": 5055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003124625025976473,
+      "loss": 0.3848,
+      "step": 5056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003122743402986391,
+      "loss": 0.4984,
+      "step": 5057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031208620894288106,
+      "loss": 0.4199,
+      "step": 5058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003118981085613831,
+      "loss": 0.0514,
+      "step": 5059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003117100391851505,
+      "loss": 0.406,
+      "step": 5060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031152200084518303,
+      "loss": 0.4369,
+      "step": 5061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031133399357247544,
+      "loss": 0.491,
+      "step": 5062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031114601739801753,
+      "loss": 0.4196,
+      "step": 5063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000310958072352794,
+      "loss": 0.4085,
+      "step": 5064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031077015846778385,
+      "loss": 0.5151,
+      "step": 5065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031058227577396156,
+      "loss": 0.5483,
+      "step": 5066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031039442430229623,
+      "loss": 0.4094,
+      "step": 5067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003102066040837519,
+      "loss": 0.4788,
+      "step": 5068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031001881514928733,
+      "loss": 0.4843,
+      "step": 5069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003098310575298563,
+      "loss": 0.4143,
+      "step": 5070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003096433312564075,
+      "loss": 0.4268,
+      "step": 5071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003094556363598838,
+      "loss": 0.4675,
+      "step": 5072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030926797287122374,
+      "loss": 0.5081,
+      "step": 5073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030908034082136016,
+      "loss": 0.5084,
+      "step": 5074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003088927402412209,
+      "loss": 0.5098,
+      "step": 5075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003087051711617289,
+      "loss": 0.4921,
+      "step": 5076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003085176336138011,
+      "loss": 0.3867,
+      "step": 5077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030833012762835,
+      "loss": 0.4544,
+      "step": 5078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003081426532362828,
+      "loss": 0.475,
+      "step": 5079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003079552104685011,
+      "loss": 0.4241,
+      "step": 5080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003077677993559014,
+      "loss": 0.4071,
+      "step": 5081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030758041992937535,
+      "loss": 0.5802,
+      "step": 5082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030739307221980906,
+      "loss": 0.5504,
+      "step": 5083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003072057562580835,
+      "loss": 0.45,
+      "step": 5084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003070184720750744,
+      "loss": 0.4125,
+      "step": 5085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003068312197016523,
+      "loss": 0.4086,
+      "step": 5086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030664399916868225,
+      "loss": 0.4197,
+      "step": 5087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003064568105070242,
+      "loss": 0.4463,
+      "step": 5088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003062696537475332,
+      "loss": 0.4204,
+      "step": 5089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003060825289210585,
+      "loss": 0.4829,
+      "step": 5090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003058954360584443,
+      "loss": 0.3747,
+      "step": 5091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030570837519052965,
+      "loss": 0.4339,
+      "step": 5092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003055213463481483,
+      "loss": 0.4105,
+      "step": 5093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030533434956212836,
+      "loss": 0.5342,
+      "step": 5094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003051473848632931,
+      "loss": 0.3636,
+      "step": 5095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003049604522824604,
+      "loss": 0.3985,
+      "step": 5096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003047735518504428,
+      "loss": 0.401,
+      "step": 5097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003045866835980473,
+      "loss": 0.4315,
+      "step": 5098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030439984755607595,
+      "loss": 0.4633,
+      "step": 5099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030421304375532544,
+      "loss": 0.4137,
+      "step": 5100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003040262722265872,
+      "loss": 0.4431,
+      "step": 5101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003038395330006468,
+      "loss": 0.4283,
+      "step": 5102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030365282610828514,
+      "loss": 0.371,
+      "step": 5103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030346615158027753,
+      "loss": 0.4406,
+      "step": 5104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030327950944739393,
+      "loss": 0.4465,
+      "step": 5105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000303092899740399,
+      "loss": 0.4934,
+      "step": 5106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000302906322490052,
+      "loss": 0.5804,
+      "step": 5107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003027197777271072,
+      "loss": 0.4677,
+      "step": 5108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030253326548231274,
+      "loss": 0.4225,
+      "step": 5109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000302346785786412,
+      "loss": 0.4202,
+      "step": 5110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003021603386701429,
+      "loss": 0.4189,
+      "step": 5111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030197392416423806,
+      "loss": 0.6204,
+      "step": 5112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030178754229942447,
+      "loss": 0.4503,
+      "step": 5113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003016011931064239,
+      "loss": 0.5479,
+      "step": 5114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003014148766159529,
+      "loss": 0.4373,
+      "step": 5115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030122859285872214,
+      "loss": 0.4486,
+      "step": 5116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003010423418654373,
+      "loss": 0.5212,
+      "step": 5117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003008561236667986,
+      "loss": 0.4554,
+      "step": 5118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003006699382935009,
+      "loss": 0.3654,
+      "step": 5119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003004837857762333,
+      "loss": 0.507,
+      "step": 5120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030029766614568,
+      "loss": 0.4048,
+      "step": 5121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003001115794325194,
+      "loss": 0.5486,
+      "step": 5122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029992552566742483,
+      "loss": 0.3965,
+      "step": 5123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002997395048810635,
+      "loss": 0.5438,
+      "step": 5124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029955351710409797,
+      "loss": 0.0507,
+      "step": 5125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029936756236718487,
+      "loss": 0.4221,
+      "step": 5126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029918164070097576,
+      "loss": 0.4232,
+      "step": 5127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029899575213611633,
+      "loss": 0.504,
+      "step": 5128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029880989670324717,
+      "loss": 0.4888,
+      "step": 5129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002986240744330033,
+      "loss": 0.4205,
+      "step": 5130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000298438285356014,
+      "loss": 0.495,
+      "step": 5131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002982525295029034,
+      "loss": 0.4634,
+      "step": 5132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029806680690429,
+      "loss": 0.4127,
+      "step": 5133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029788111759078705,
+      "loss": 0.4257,
+      "step": 5134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029769546159300205,
+      "loss": 0.5072,
+      "step": 5135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002975098389415371,
+      "loss": 0.4348,
+      "step": 5136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029732424966698903,
+      "loss": 0.4005,
+      "step": 5137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002971386937999485,
+      "loss": 0.5293,
+      "step": 5138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969531713710014,
+      "loss": 0.4438,
+      "step": 5139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002967676824107278,
+      "loss": 0.4258,
+      "step": 5140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029658222694970224,
+      "loss": 0.4142,
+      "step": 5141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029639680501849364,
+      "loss": 0.0501,
+      "step": 5142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002962114166476657,
+      "loss": 0.4476,
+      "step": 5143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002960260618677765,
+      "loss": 0.4352,
+      "step": 5144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029584074070937823,
+      "loss": 0.4683,
+      "step": 5145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002956554532030179,
+      "loss": 0.4523,
+      "step": 5146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002954701993792368,
+      "loss": 0.41,
+      "step": 5147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002952849792685709,
+      "loss": 0.6597,
+      "step": 5148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002950997929015503,
+      "loss": 0.4647,
+      "step": 5149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029491464030869985,
+      "loss": 0.4818,
+      "step": 5150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029472952152053854,
+      "loss": 0.4785,
+      "step": 5151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002945444365675803,
+      "loss": 0.5103,
+      "step": 5152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002943593854803325,
+      "loss": 0.4061,
+      "step": 5153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029417436828929784,
+      "loss": 0.587,
+      "step": 5154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002939893850249731,
+      "loss": 0.4106,
+      "step": 5155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029380443571784954,
+      "loss": 0.474,
+      "step": 5156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029361952039841274,
+      "loss": 0.4403,
+      "step": 5157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029343463909714274,
+      "loss": 0.4913,
+      "step": 5158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029324979184451404,
+      "loss": 0.4235,
+      "step": 5159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029306497867099515,
+      "loss": 0.4484,
+      "step": 5160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002928801996070495,
+      "loss": 0.4294,
+      "step": 5161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029269545468313456,
+      "loss": 0.5554,
+      "step": 5162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029251074392970224,
+      "loss": 0.4808,
+      "step": 5163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002923260673771988,
+      "loss": 0.5281,
+      "step": 5164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000292141425056065,
+      "loss": 0.4815,
+      "step": 5165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029195681699673606,
+      "loss": 0.3345,
+      "step": 5166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002917722432296409,
+      "loss": 0.3972,
+      "step": 5167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002915877037852034,
+      "loss": 0.4742,
+      "step": 5168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002914031986938417,
+      "loss": 0.4924,
+      "step": 5169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029121872798596805,
+      "loss": 0.4343,
+      "step": 5170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029103429169198943,
+      "loss": 0.0494,
+      "step": 5171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029084988984230664,
+      "loss": 0.4249,
+      "step": 5172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029066552246731524,
+      "loss": 0.5195,
+      "step": 5173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029048118959740504,
+      "loss": 0.5576,
+      "step": 5174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029029689126295965,
+      "loss": 0.4352,
+      "step": 5175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029011262749435767,
+      "loss": 0.447,
+      "step": 5176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002899283983219716,
+      "loss": 0.3639,
+      "step": 5177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002897442037761684,
+      "loss": 0.5,
+      "step": 5178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002895600438873093,
+      "loss": 0.4878,
+      "step": 5179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028937591868574984,
+      "loss": 0.4332,
+      "step": 5180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028919182820183985,
+      "loss": 0.4244,
+      "step": 5181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028900777246592313,
+      "loss": 0.4421,
+      "step": 5182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028882375150833815,
+      "loss": 0.4684,
+      "step": 5183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002886397653594175,
+      "loss": 0.5608,
+      "step": 5184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028845581404948804,
+      "loss": 0.4487,
+      "step": 5185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028827189760887096,
+      "loss": 0.4233,
+      "step": 5186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002880880160678815,
+      "loss": 0.4572,
+      "step": 5187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028790416945682955,
+      "loss": 0.5565,
+      "step": 5188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002877203578060186,
+      "loss": 0.4305,
+      "step": 5189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028753658114574683,
+      "loss": 0.4636,
+      "step": 5190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028735283950630664,
+      "loss": 0.451,
+      "step": 5191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028716913291798466,
+      "loss": 0.4722,
+      "step": 5192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002869854614110617,
+      "loss": 0.4038,
+      "step": 5193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028680182501581255,
+      "loss": 0.4637,
+      "step": 5194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002866182237625064,
+      "loss": 0.4213,
+      "step": 5195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028643465768140686,
+      "loss": 0.4814,
+      "step": 5196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002862511268027716,
+      "loss": 0.5063,
+      "step": 5197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028606763115685213,
+      "loss": 0.0491,
+      "step": 5198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002858841707738946,
+      "loss": 0.3915,
+      "step": 5199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002857007456841393,
+      "loss": 0.5226,
+      "step": 5200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028551735591782044,
+      "loss": 0.5215,
+      "step": 5201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028533400150516675,
+      "loss": 0.4277,
+      "step": 5202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002851506824764011,
+      "loss": 0.5157,
+      "step": 5203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002849673988617399,
+      "loss": 0.3809,
+      "step": 5204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028478415069139453,
+      "loss": 0.4321,
+      "step": 5205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002846009379955701,
+      "loss": 0.4899,
+      "step": 5206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000284417760804466,
+      "loss": 0.4293,
+      "step": 5207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028423461914827585,
+      "loss": 0.4025,
+      "step": 5208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028405151305718724,
+      "loss": 0.4473,
+      "step": 5209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002838684425613821,
+      "loss": 0.3992,
+      "step": 5210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028368540769103603,
+      "loss": 0.453,
+      "step": 5211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002835024084763193,
+      "loss": 0.509,
+      "step": 5212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000283319444947396,
+      "loss": 0.4594,
+      "step": 5213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028313651713442455,
+      "loss": 0.5262,
+      "step": 5214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002829536250675573,
+      "loss": 0.4918,
+      "step": 5215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028277076877694077,
+      "loss": 0.3911,
+      "step": 5216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002825879482927155,
+      "loss": 0.473,
+      "step": 5217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028240516364501654,
+      "loss": 0.4447,
+      "step": 5218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002822224148639723,
+      "loss": 0.4365,
+      "step": 5219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028203970197970574,
+      "loss": 0.4361,
+      "step": 5220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000281857025022334,
+      "loss": 0.4484,
+      "step": 5221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000281674384021968,
+      "loss": 0.4864,
+      "step": 5222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028149177900871304,
+      "loss": 0.4775,
+      "step": 5223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028130921001266816,
+      "loss": 0.457,
+      "step": 5224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028112667706392693,
+      "loss": 0.504,
+      "step": 5225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002809441801925762,
+      "loss": 0.5169,
+      "step": 5226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002807617194286976,
+      "loss": 0.3882,
+      "step": 5227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028057929480236653,
+      "loss": 0.0487,
+      "step": 5228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002803969063436525,
+      "loss": 0.608,
+      "step": 5229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028021455408261896,
+      "loss": 0.0485,
+      "step": 5230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002800322380493235,
+      "loss": 0.0484,
+      "step": 5231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027984995827381786,
+      "loss": 0.5126,
+      "step": 5232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002796677147861472,
+      "loss": 0.4645,
+      "step": 5233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002794855076163514,
+      "loss": 0.4106,
+      "step": 5234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027930333679446407,
+      "loss": 0.4406,
+      "step": 5235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027912120235051304,
+      "loss": 0.4255,
+      "step": 5236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027893910431451945,
+      "loss": 0.4491,
+      "step": 5237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027875704271649926,
+      "loss": 0.47,
+      "step": 5238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002785750175864621,
+      "loss": 0.4571,
+      "step": 5239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002783930289544118,
+      "loss": 0.4165,
+      "step": 5240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002782110768503454,
+      "loss": 0.5366,
+      "step": 5241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027802916130425484,
+      "loss": 0.4025,
+      "step": 5242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027784728234612555,
+      "loss": 0.473,
+      "step": 5243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027766544000593725,
+      "loss": 0.4568,
+      "step": 5244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027748363431366325,
+      "loss": 0.4074,
+      "step": 5245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027730186529927107,
+      "loss": 0.403,
+      "step": 5246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002771201329927223,
+      "loss": 0.4822,
+      "step": 5247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027693843742397195,
+      "loss": 0.4122,
+      "step": 5248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027675677862296945,
+      "loss": 0.4102,
+      "step": 5249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000276575156619658,
+      "loss": 0.4081,
+      "step": 5250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002763935714439748,
+      "loss": 0.4019,
+      "step": 5251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000276212023125851,
+      "loss": 0.4769,
+      "step": 5252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027603051169521167,
+      "loss": 0.5181,
+      "step": 5253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002758490371819758,
+      "loss": 0.4811,
+      "step": 5254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002756675996160559,
+      "loss": 0.428,
+      "step": 5255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027548619902735907,
+      "loss": 0.4696,
+      "step": 5256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027530483544578577,
+      "loss": 0.4968,
+      "step": 5257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002751235089012308,
+      "loss": 0.4039,
+      "step": 5258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027494221942358246,
+      "loss": 0.0506,
+      "step": 5259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002747609670427231,
+      "loss": 0.3893,
+      "step": 5260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002745797517885293,
+      "loss": 0.5507,
+      "step": 5261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002743985736908707,
+      "loss": 0.5376,
+      "step": 5262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002742174327796115,
+      "loss": 0.4363,
+      "step": 5263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002740363290846096,
+      "loss": 0.5332,
+      "step": 5264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027385526263571676,
+      "loss": 0.0489,
+      "step": 5265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027367423346277847,
+      "loss": 0.4725,
+      "step": 5266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027349324159563426,
+      "loss": 0.4456,
+      "step": 5267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002733122870641175,
+      "loss": 0.4399,
+      "step": 5268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002731313698980554,
+      "loss": 0.4814,
+      "step": 5269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002729504901272686,
+      "loss": 0.5984,
+      "step": 5270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002727696477815722,
+      "loss": 0.0494,
+      "step": 5271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002725888428907748,
+      "loss": 0.4015,
+      "step": 5272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002724080754846788,
+      "loss": 0.0494,
+      "step": 5273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027222734559308073,
+      "loss": 0.4098,
+      "step": 5274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002720466532457707,
+      "loss": 0.4376,
+      "step": 5275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002718659984725323,
+      "loss": 0.491,
+      "step": 5276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002716853813031435,
+      "loss": 0.5282,
+      "step": 5277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002715048017673758,
+      "loss": 0.4647,
+      "step": 5278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002713242598949948,
+      "loss": 0.4268,
+      "step": 5279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002711437557157591,
+      "loss": 0.4368,
+      "step": 5280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027096328925942207,
+      "loss": 0.4308,
+      "step": 5281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002707828605557301,
+      "loss": 0.4803,
+      "step": 5282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002706024696344241,
+      "loss": 0.4289,
+      "step": 5283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027042211652523784,
+      "loss": 0.4399,
+      "step": 5284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002702418012578995,
+      "loss": 0.4675,
+      "step": 5285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002700615238621309,
+      "loss": 0.4178,
+      "step": 5286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026988128436764757,
+      "loss": 0.4627,
+      "step": 5287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002697010828041587,
+      "loss": 0.4456,
+      "step": 5288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026952091920136756,
+      "loss": 0.374,
+      "step": 5289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002693407935889707,
+      "loss": 0.5281,
+      "step": 5290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000269160705996659,
+      "loss": 0.4695,
+      "step": 5291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026898065645411615,
+      "loss": 0.5032,
+      "step": 5292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002688006449910204,
+      "loss": 0.4407,
+      "step": 5293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002686206716370436,
+      "loss": 0.4316,
+      "step": 5294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002684407364218509,
+      "loss": 0.3912,
+      "step": 5295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002682608393751016,
+      "loss": 0.4077,
+      "step": 5296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002680809805264486,
+      "loss": 0.4052,
+      "step": 5297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026790115990553843,
+      "loss": 0.5065,
+      "step": 5298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002677213775420111,
+      "loss": 0.4301,
+      "step": 5299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026754163346550075,
+      "loss": 0.4701,
+      "step": 5300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002673619277056349,
+      "loss": 0.4375,
+      "step": 5301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000267182260292035,
+      "loss": 0.4495,
+      "step": 5302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026700263125431587,
+      "loss": 0.457,
+      "step": 5303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026682304062208633,
+      "loss": 0.5016,
+      "step": 5304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002666434884249489,
+      "loss": 0.3864,
+      "step": 5305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026646397469249906,
+      "loss": 0.4489,
+      "step": 5306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026628449945432673,
+      "loss": 0.444,
+      "step": 5307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002661050627400153,
+      "loss": 0.462,
+      "step": 5308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002659256645791417,
+      "loss": 0.4998,
+      "step": 5309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002657463050012765,
+      "loss": 0.5094,
+      "step": 5310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026556698403598395,
+      "loss": 0.3954,
+      "step": 5311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026538770171282187,
+      "loss": 0.4124,
+      "step": 5312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002652084580613422,
+      "loss": 0.399,
+      "step": 5313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002650292531110895,
+      "loss": 0.4821,
+      "step": 5314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002648500868916027,
+      "loss": 0.4913,
+      "step": 5315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026467095943241427,
+      "loss": 0.3694,
+      "step": 5316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002644918707630502,
+      "loss": 0.4536,
+      "step": 5317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026431282091303026,
+      "loss": 0.4584,
+      "step": 5318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026413380991186723,
+      "loss": 0.3556,
+      "step": 5319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026395483778906816,
+      "loss": 0.4308,
+      "step": 5320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026377590457413337,
+      "loss": 0.4409,
+      "step": 5321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002635970102965568,
+      "loss": 0.4464,
+      "step": 5322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026341815498582633,
+      "loss": 0.5338,
+      "step": 5323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002632393386714226,
+      "loss": 0.4679,
+      "step": 5324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002630605613828205,
+      "loss": 0.3751,
+      "step": 5325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026288182314948837,
+      "loss": 0.5159,
+      "step": 5326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026270312400088826,
+      "loss": 0.5123,
+      "step": 5327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000262524463966475,
+      "loss": 0.4275,
+      "step": 5328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026234584307569794,
+      "loss": 0.4337,
+      "step": 5329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002621672613579994,
+      "loss": 0.5262,
+      "step": 5330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002619887188428155,
+      "loss": 0.4293,
+      "step": 5331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026181021555957574,
+      "loss": 0.4674,
+      "step": 5332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026163175153770334,
+      "loss": 0.51,
+      "step": 5333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026145332680661484,
+      "loss": 0.4281,
+      "step": 5334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026127494139572063,
+      "loss": 0.4684,
+      "step": 5335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000261096595334424,
+      "loss": 0.4183,
+      "step": 5336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002609182886521223,
+      "loss": 0.4694,
+      "step": 5337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002607400213782062,
+      "loss": 0.4391,
+      "step": 5338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026056179354206,
+      "loss": 0.4054,
+      "step": 5339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002603836051730614,
+      "loss": 0.5249,
+      "step": 5340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002602054563005815,
+      "loss": 0.418,
+      "step": 5341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002600273469539852,
+      "loss": 0.5398,
+      "step": 5342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598492771626304,
+      "loss": 0.3993,
+      "step": 5343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025967124695586873,
+      "loss": 0.4891,
+      "step": 5344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025949325636304556,
+      "loss": 0.4036,
+      "step": 5345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002593153054134993,
+      "loss": 0.3463,
+      "step": 5346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002591373941365622,
+      "loss": 0.4812,
+      "step": 5347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002589595225615596,
+      "loss": 0.4232,
+      "step": 5348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002587816907178107,
+      "loss": 0.4218,
+      "step": 5349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025860389863462764,
+      "loss": 0.3816,
+      "step": 5350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025842614634131637,
+      "loss": 0.4619,
+      "step": 5351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025824843386717633,
+      "loss": 0.5952,
+      "step": 5352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002580707612415002,
+      "loss": 0.4689,
+      "step": 5353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025789312849357416,
+      "loss": 0.5068,
+      "step": 5354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025771553565267786,
+      "loss": 0.421,
+      "step": 5355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002575379827480843,
+      "loss": 0.0492,
+      "step": 5356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025736046980906013,
+      "loss": 0.3789,
+      "step": 5357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002571829968648649,
+      "loss": 0.4769,
+      "step": 5358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025700556394475194,
+      "loss": 0.5017,
+      "step": 5359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025682817107796805,
+      "loss": 0.4596,
+      "step": 5360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025665081829375326,
+      "loss": 0.4351,
+      "step": 5361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025647350562134114,
+      "loss": 0.5165,
+      "step": 5362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002562962330899583,
+      "loss": 0.049,
+      "step": 5363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002561190007288251,
+      "loss": 0.4835,
+      "step": 5364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002559418085671551,
+      "loss": 0.4574,
+      "step": 5365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002557646566341556,
+      "loss": 0.4957,
+      "step": 5366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002555875449590265,
+      "loss": 0.5198,
+      "step": 5367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002554104735709618,
+      "loss": 0.4164,
+      "step": 5368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025523344249914843,
+      "loss": 0.5084,
+      "step": 5369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002550564517727669,
+      "loss": 0.4469,
+      "step": 5370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025487950142099127,
+      "loss": 0.0487,
+      "step": 5371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002547025914729882,
+      "loss": 0.4229,
+      "step": 5372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025452572195791833,
+      "loss": 0.4679,
+      "step": 5373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002543488929049356,
+      "loss": 0.394,
+      "step": 5374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002541721043431869,
+      "loss": 0.4733,
+      "step": 5375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002539953563018129,
+      "loss": 0.3791,
+      "step": 5376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025381864880994745,
+      "loss": 0.4194,
+      "step": 5377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002536419818967176,
+      "loss": 0.4365,
+      "step": 5378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025346535559124354,
+      "loss": 0.4374,
+      "step": 5379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025328876992263916,
+      "loss": 0.4001,
+      "step": 5380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002531122249200114,
+      "loss": 0.3894,
+      "step": 5381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025293572061246074,
+      "loss": 0.484,
+      "step": 5382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025275925702908066,
+      "loss": 0.3804,
+      "step": 5383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000252582834198958,
+      "loss": 0.3823,
+      "step": 5384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002524064521511731,
+      "loss": 0.4644,
+      "step": 5385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002522301109147995,
+      "loss": 0.4872,
+      "step": 5386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025205381051890364,
+      "loss": 0.436,
+      "step": 5387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002518775509925456,
+      "loss": 0.5341,
+      "step": 5388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025170133236477867,
+      "loss": 0.4211,
+      "step": 5389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002515251546646494,
+      "loss": 0.4153,
+      "step": 5390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025134901792119756,
+      "loss": 0.413,
+      "step": 5391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025117292216345614,
+      "loss": 0.4546,
+      "step": 5392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002509968674204517,
+      "loss": 0.4918,
+      "step": 5393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025082085372120323,
+      "loss": 0.4127,
+      "step": 5394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002506448810947237,
+      "loss": 0.4249,
+      "step": 5395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002504689495700191,
+      "loss": 0.6299,
+      "step": 5396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025029305917608863,
+      "loss": 0.3578,
+      "step": 5397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025011720994192474,
+      "loss": 0.4358,
+      "step": 5398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000249941401896513,
+      "loss": 0.5215,
+      "step": 5399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002497656350688324,
+      "loss": 0.4279,
+      "step": 5400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024958990948785475,
+      "loss": 0.4641,
+      "step": 5401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002494142251825454,
+      "loss": 0.4378,
+      "step": 5402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002492385821818628,
+      "loss": 0.4828,
+      "step": 5403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002490629805147586,
+      "loss": 0.4463,
+      "step": 5404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024888742021017784,
+      "loss": 0.4846,
+      "step": 5405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024871190129705813,
+      "loss": 0.5713,
+      "step": 5406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024853642380433076,
+      "loss": 0.379,
+      "step": 5407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002483609877609202,
+      "loss": 0.4158,
+      "step": 5408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024818559319574415,
+      "loss": 0.4563,
+      "step": 5409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024801024013771283,
+      "loss": 0.3774,
+      "step": 5410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024783492861573037,
+      "loss": 0.4757,
+      "step": 5411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024765965865869373,
+      "loss": 0.6133,
+      "step": 5412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024748443029549305,
+      "loss": 0.4544,
+      "step": 5413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002473092435550117,
+      "loss": 0.4791,
+      "step": 5414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002471340984661262,
+      "loss": 0.3851,
+      "step": 5415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024695899505770575,
+      "loss": 0.6381,
+      "step": 5416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024678393335861326,
+      "loss": 0.4788,
+      "step": 5417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002466089133977046,
+      "loss": 0.4738,
+      "step": 5418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002464339352038286,
+      "loss": 0.3823,
+      "step": 5419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002462589988058274,
+      "loss": 0.3577,
+      "step": 5420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002460841042325362,
+      "loss": 0.4808,
+      "step": 5421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002459092515127833,
+      "loss": 0.4052,
+      "step": 5422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024573444067538984,
+      "loss": 0.4027,
+      "step": 5423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002455596717491705,
+      "loss": 0.4473,
+      "step": 5424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002453849447629328,
+      "loss": 0.401,
+      "step": 5425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002452102597454773,
+      "loss": 0.379,
+      "step": 5426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002450356167255978,
+      "loss": 0.5,
+      "step": 5427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024486101573208114,
+      "loss": 0.4658,
+      "step": 5428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024468645679370717,
+      "loss": 0.4784,
+      "step": 5429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000244511939939249,
+      "loss": 0.0486,
+      "step": 5430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002443374651974723,
+      "loss": 0.4307,
+      "step": 5431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024416303259713635,
+      "loss": 0.4379,
+      "step": 5432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002439886421669933,
+      "loss": 0.4772,
+      "step": 5433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002438142939357882,
+      "loss": 0.39,
+      "step": 5434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024363998793225939,
+      "loss": 0.4498,
+      "step": 5435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024346572418513802,
+      "loss": 0.4949,
+      "step": 5436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002432915027231487,
+      "loss": 0.5101,
+      "step": 5437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024311732357500826,
+      "loss": 0.0483,
+      "step": 5438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024294318676942728,
+      "loss": 0.5074,
+      "step": 5439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002427690923351092,
+      "loss": 0.417,
+      "step": 5440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024259504030075025,
+      "loss": 0.4391,
+      "step": 5441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024242103069503995,
+      "loss": 0.4161,
+      "step": 5442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024224706354666065,
+      "loss": 0.4628,
+      "step": 5443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024207313888428795,
+      "loss": 0.5774,
+      "step": 5444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024189925673658992,
+      "loss": 0.4449,
+      "step": 5445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024172541713222807,
+      "loss": 0.4714,
+      "step": 5446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024155162009985675,
+      "loss": 0.4475,
+      "step": 5447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024137786566812364,
+      "loss": 0.5101,
+      "step": 5448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024120415386566862,
+      "loss": 0.4451,
+      "step": 5449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002410304847211252,
+      "loss": 0.5042,
+      "step": 5450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024085685826311966,
+      "loss": 0.4081,
+      "step": 5451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024068327452027145,
+      "loss": 0.3951,
+      "step": 5452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024050973352119238,
+      "loss": 0.4265,
+      "step": 5453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002403362352944878,
+      "loss": 0.4353,
+      "step": 5454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024016277986875585,
+      "loss": 0.422,
+      "step": 5455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023998936727258753,
+      "loss": 0.4368,
+      "step": 5456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002398159975345669,
+      "loss": 0.4659,
+      "step": 5457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023964267068327084,
+      "loss": 0.4861,
+      "step": 5458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023946938674726943,
+      "loss": 0.4548,
+      "step": 5459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002392961457551251,
+      "loss": 0.4716,
+      "step": 5460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023912294773539367,
+      "loss": 0.5063,
+      "step": 5461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002389497927166238,
+      "loss": 0.3826,
+      "step": 5462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023877668072735708,
+      "loss": 0.4802,
+      "step": 5463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023860361179612794,
+      "loss": 0.495,
+      "step": 5464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023843058595146365,
+      "loss": 0.5162,
+      "step": 5465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023825760322188473,
+      "loss": 0.4581,
+      "step": 5466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002380846636359039,
+      "loss": 0.4789,
+      "step": 5467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002379117672220274,
+      "loss": 0.5619,
+      "step": 5468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023773891400875413,
+      "loss": 0.5455,
+      "step": 5469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002375661040245759,
+      "loss": 0.4369,
+      "step": 5470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002373933372979773,
+      "loss": 0.4285,
+      "step": 5471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023722061385743592,
+      "loss": 0.3597,
+      "step": 5472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023704793373142225,
+      "loss": 0.0484,
+      "step": 5473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023687529694839954,
+      "loss": 0.4232,
+      "step": 5474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023670270353682373,
+      "loss": 0.479,
+      "step": 5475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023653015352514384,
+      "loss": 0.4914,
+      "step": 5476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023635764694180173,
+      "loss": 0.4467,
+      "step": 5477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023618518381523203,
+      "loss": 0.3972,
+      "step": 5478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002360127641738623,
+      "loss": 0.4215,
+      "step": 5479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023584038804611279,
+      "loss": 0.4755,
+      "step": 5480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023566805546039688,
+      "loss": 0.5176,
+      "step": 5481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023549576644512016,
+      "loss": 0.5219,
+      "step": 5482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002353235210286816,
+      "loss": 0.4348,
+      "step": 5483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023515131923947281,
+      "loss": 0.4662,
+      "step": 5484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002349791611058782,
+      "loss": 0.5063,
+      "step": 5485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000234807046656275,
+      "loss": 0.5549,
+      "step": 5486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002346349759190332,
+      "loss": 0.4231,
+      "step": 5487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002344629489225159,
+      "loss": 0.3795,
+      "step": 5488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002342909656950782,
+      "loss": 0.4071,
+      "step": 5489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023411902626506876,
+      "loss": 0.4472,
+      "step": 5490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023394713066082868,
+      "loss": 0.4059,
+      "step": 5491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023377527891069218,
+      "loss": 0.4341,
+      "step": 5492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023360347104298552,
+      "loss": 0.5105,
+      "step": 5493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023343170708602846,
+      "loss": 0.5061,
+      "step": 5494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023325998706813316,
+      "loss": 0.3633,
+      "step": 5495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023308831101760485,
+      "loss": 0.048,
+      "step": 5496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023291667896274088,
+      "loss": 0.4203,
+      "step": 5497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000232745090931832,
+      "loss": 0.4241,
+      "step": 5498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023257354695316147,
+      "loss": 0.5211,
+      "step": 5499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023240204705500517,
+      "loss": 0.4686,
+      "step": 5500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002322305912656319,
+      "loss": 0.475,
+      "step": 5501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023205917961330308,
+      "loss": 0.4343,
+      "step": 5502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023188781212627314,
+      "loss": 0.4924,
+      "step": 5503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023171648883278846,
+      "loss": 0.5017,
+      "step": 5504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000231545209761089,
+      "loss": 0.5071,
+      "step": 5505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023137397493940698,
+      "loss": 0.4957,
+      "step": 5506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002312027843959675,
+      "loss": 0.45,
+      "step": 5507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023103163815898814,
+      "loss": 0.4555,
+      "step": 5508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023086053625667952,
+      "loss": 0.4703,
+      "step": 5509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023068947871724484,
+      "loss": 0.3884,
+      "step": 5510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023051846556887951,
+      "loss": 0.4689,
+      "step": 5511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023034749683977225,
+      "loss": 0.4802,
+      "step": 5512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023017657255810427,
+      "loss": 0.5076,
+      "step": 5513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002300056927520493,
+      "loss": 0.4333,
+      "step": 5514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022983485744977388,
+      "loss": 0.4347,
+      "step": 5515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022966406667943728,
+      "loss": 0.4867,
+      "step": 5516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022949332046919146,
+      "loss": 0.5678,
+      "step": 5517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002293226188471805,
+      "loss": 0.4515,
+      "step": 5518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022915196184154175,
+      "loss": 0.4162,
+      "step": 5519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000228981349480405,
+      "loss": 0.5341,
+      "step": 5520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002288107817918927,
+      "loss": 0.4193,
+      "step": 5521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022864025880411982,
+      "loss": 0.3997,
+      "step": 5522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022846978054519418,
+      "loss": 0.5083,
+      "step": 5523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022829934704321603,
+      "loss": 0.3992,
+      "step": 5524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002281289583262785,
+      "loss": 0.5033,
+      "step": 5525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002279586144224668,
+      "loss": 0.4436,
+      "step": 5526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022778831535985927,
+      "loss": 0.4159,
+      "step": 5527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022761806116652677,
+      "loss": 0.3828,
+      "step": 5528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022744785187053256,
+      "loss": 0.4563,
+      "step": 5529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022727768749993273,
+      "loss": 0.4301,
+      "step": 5530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000227107568082776,
+      "loss": 0.3652,
+      "step": 5531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002269374936471032,
+      "loss": 0.4172,
+      "step": 5532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022676746422094823,
+      "loss": 0.4841,
+      "step": 5533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022659747983233748,
+      "loss": 0.4702,
+      "step": 5534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022642754050929004,
+      "loss": 0.4167,
+      "step": 5535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000226257646279817,
+      "loss": 0.4257,
+      "step": 5536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022608779717192264,
+      "loss": 0.4414,
+      "step": 5537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022591799321360352,
+      "loss": 0.485,
+      "step": 5538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022574823443284904,
+      "loss": 0.4473,
+      "step": 5539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022557852085764053,
+      "loss": 0.3688,
+      "step": 5540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002254088525159525,
+      "loss": 0.3677,
+      "step": 5541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022523922943575175,
+      "loss": 0.4125,
+      "step": 5542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022506965164499764,
+      "loss": 0.4616,
+      "step": 5543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022490011917164206,
+      "loss": 0.3848,
+      "step": 5544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022473063204362943,
+      "loss": 0.3877,
+      "step": 5545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022456119028889672,
+      "loss": 0.3917,
+      "step": 5546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022439179393537358,
+      "loss": 0.4471,
+      "step": 5547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002242224430109816,
+      "loss": 0.444,
+      "step": 5548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002240531375436355,
+      "loss": 0.4111,
+      "step": 5549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022388387756124223,
+      "loss": 0.5042,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002237146630917014,
+      "loss": 0.4612,
+      "step": 5551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022354549416290497,
+      "loss": 0.5092,
+      "step": 5552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022337637080273743,
+      "loss": 0.4156,
+      "step": 5553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022320729303907595,
+      "loss": 0.4277,
+      "step": 5554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022303826089978962,
+      "loss": 0.387,
+      "step": 5555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022286927441274063,
+      "loss": 0.3933,
+      "step": 5556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002227003336057834,
+      "loss": 0.3862,
+      "step": 5557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022253143850676482,
+      "loss": 0.6559,
+      "step": 5558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022236258914352426,
+      "loss": 0.4648,
+      "step": 5559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022219378554389353,
+      "loss": 0.408,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022202502773569706,
+      "loss": 0.4606,
+      "step": 5561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002218563157467513,
+      "loss": 0.3856,
+      "step": 5562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002216876496048656,
+      "loss": 0.4648,
+      "step": 5563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022151902933784153,
+      "loss": 0.4028,
+      "step": 5564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022135045497347322,
+      "loss": 0.4156,
+      "step": 5565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002211819265395471,
+      "loss": 0.3636,
+      "step": 5566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022101344406384222,
+      "loss": 0.5044,
+      "step": 5567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022084500757412978,
+      "loss": 0.5818,
+      "step": 5568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022067661709817383,
+      "loss": 0.4137,
+      "step": 5569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022050827266373015,
+      "loss": 0.4777,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022033997429854753,
+      "loss": 0.3931,
+      "step": 5571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022017172203036702,
+      "loss": 0.5327,
+      "step": 5572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022000351588692197,
+      "loss": 0.4786,
+      "step": 5573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021983535589593838,
+      "loss": 0.0482,
+      "step": 5574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002196672420851341,
+      "loss": 0.4354,
+      "step": 5575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021949917448221983,
+      "loss": 0.4304,
+      "step": 5576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021933115311489865,
+      "loss": 0.4768,
+      "step": 5577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000219163178010866,
+      "loss": 0.048,
+      "step": 5578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021899524919780927,
+      "loss": 0.4537,
+      "step": 5579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021882736670340874,
+      "loss": 0.3903,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002186595305553369,
+      "loss": 0.454,
+      "step": 5581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021849174078125843,
+      "loss": 0.4683,
+      "step": 5582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021832399740883085,
+      "loss": 0.3583,
+      "step": 5583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021815630046570318,
+      "loss": 0.4879,
+      "step": 5584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021798864997951757,
+      "loss": 0.4193,
+      "step": 5585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021782104597790818,
+      "loss": 0.3855,
+      "step": 5586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021765348848850163,
+      "loss": 0.4177,
+      "step": 5587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021748597753891674,
+      "loss": 0.4137,
+      "step": 5588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002173185131567647,
+      "loss": 0.4378,
+      "step": 5589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021715109536964912,
+      "loss": 0.4197,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000216983724205166,
+      "loss": 0.413,
+      "step": 5591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021681639969090317,
+      "loss": 0.4624,
+      "step": 5592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021664912185444124,
+      "loss": 0.4325,
+      "step": 5593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021648189072335312,
+      "loss": 0.4371,
+      "step": 5594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021631470632520373,
+      "loss": 0.4252,
+      "step": 5595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021614756868755059,
+      "loss": 0.4644,
+      "step": 5596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021598047783794334,
+      "loss": 0.5066,
+      "step": 5597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002158134338039241,
+      "loss": 0.3864,
+      "step": 5598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021564643661302674,
+      "loss": 0.4108,
+      "step": 5599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021547948629277807,
+      "loss": 0.4084,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021531258287069683,
+      "loss": 0.0483,
+      "step": 5601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021514572637429407,
+      "loss": 0.4998,
+      "step": 5602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021497891683107317,
+      "loss": 0.5785,
+      "step": 5603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021481215426852974,
+      "loss": 0.4406,
+      "step": 5604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021464543871415177,
+      "loss": 0.4999,
+      "step": 5605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021447877019541901,
+      "loss": 0.0482,
+      "step": 5606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021431214873980403,
+      "loss": 0.4414,
+      "step": 5607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021414557437477138,
+      "loss": 0.0482,
+      "step": 5608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021397904712777794,
+      "loss": 0.4768,
+      "step": 5609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021381256702627272,
+      "loss": 0.483,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021364613409769712,
+      "loss": 0.5026,
+      "step": 5611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002134797483694847,
+      "loss": 0.0477,
+      "step": 5612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021331340986906094,
+      "loss": 0.4656,
+      "step": 5613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021314711862384394,
+      "loss": 0.5052,
+      "step": 5614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021298087466124394,
+      "loss": 0.4388,
+      "step": 5615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021281467800866316,
+      "loss": 0.4215,
+      "step": 5616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002126485286934965,
+      "loss": 0.4664,
+      "step": 5617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021248242674313028,
+      "loss": 0.454,
+      "step": 5618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021231637218494372,
+      "loss": 0.4907,
+      "step": 5619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021215036504630786,
+      "loss": 0.4308,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021198440535458634,
+      "loss": 0.4685,
+      "step": 5621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021181849313713425,
+      "loss": 0.4783,
+      "step": 5622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002116526284212995,
+      "loss": 0.4802,
+      "step": 5623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021148681123442194,
+      "loss": 0.4823,
+      "step": 5624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021132104160383354,
+      "loss": 0.4756,
+      "step": 5625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021115531955685852,
+      "loss": 0.4165,
+      "step": 5626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021098964512081343,
+      "loss": 0.5341,
+      "step": 5627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021082401832300641,
+      "loss": 0.405,
+      "step": 5628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021065843919073824,
+      "loss": 0.3975,
+      "step": 5629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021049290775130176,
+      "loss": 0.4287,
+      "step": 5630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002103274240319818,
+      "loss": 0.4774,
+      "step": 5631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021016198806005554,
+      "loss": 0.048,
+      "step": 5632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020999659986279212,
+      "loss": 0.5011,
+      "step": 5633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000209831259467453,
+      "loss": 0.3823,
+      "step": 5634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020966596690129137,
+      "loss": 0.4232,
+      "step": 5635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002095007221915528,
+      "loss": 0.4819,
+      "step": 5636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020933552536547506,
+      "loss": 0.4146,
+      "step": 5637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020917037645028786,
+      "loss": 0.4825,
+      "step": 5638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020900527547321318,
+      "loss": 0.5001,
+      "step": 5639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002088402224614649,
+      "loss": 0.4628,
+      "step": 5640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020867521744224904,
+      "loss": 0.5427,
+      "step": 5641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020851026044276406,
+      "loss": 0.4506,
+      "step": 5642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002083453514901998,
+      "loss": 0.3726,
+      "step": 5643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020818049061173866,
+      "loss": 0.3846,
+      "step": 5644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020801567783455523,
+      "loss": 0.514,
+      "step": 5645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002078509131858158,
+      "loss": 0.5043,
+      "step": 5646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020768619669267897,
+      "loss": 0.3719,
+      "step": 5647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002075215283822953,
+      "loss": 0.4762,
+      "step": 5648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002073569082818077,
+      "loss": 0.3835,
+      "step": 5649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020719233641835046,
+      "loss": 0.4519,
+      "step": 5650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020702781281905052,
+      "loss": 0.4222,
+      "step": 5651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020686333751102675,
+      "loss": 0.5889,
+      "step": 5652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002066989105213899,
+      "loss": 0.4374,
+      "step": 5653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002065345318772429,
+      "loss": 0.5184,
+      "step": 5654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020637020160568065,
+      "loss": 0.0477,
+      "step": 5655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020620591973379026,
+      "loss": 0.4071,
+      "step": 5656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020604168628865027,
+      "loss": 0.4868,
+      "step": 5657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020587750129733197,
+      "loss": 0.3372,
+      "step": 5658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020571336478689817,
+      "loss": 0.4559,
+      "step": 5659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020554927678440404,
+      "loss": 0.5123,
+      "step": 5660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020538523731689667,
+      "loss": 0.4756,
+      "step": 5661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020522124641141472,
+      "loss": 0.4217,
+      "step": 5662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002050573040949893,
+      "loss": 0.5109,
+      "step": 5663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020489341039464355,
+      "loss": 0.406,
+      "step": 5664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020472956533739244,
+      "loss": 0.3989,
+      "step": 5665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002045657689502427,
+      "loss": 0.4541,
+      "step": 5666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002044020212601934,
+      "loss": 0.5095,
+      "step": 5667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002042383222942355,
+      "loss": 0.4302,
+      "step": 5668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020407467207935178,
+      "loss": 0.4677,
+      "step": 5669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020391107064251707,
+      "loss": 0.4103,
+      "step": 5670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020374751801069846,
+      "loss": 0.5397,
+      "step": 5671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020358401421085426,
+      "loss": 0.5632,
+      "step": 5672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002034205592699354,
+      "loss": 0.4054,
+      "step": 5673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020325715321488453,
+      "loss": 0.4138,
+      "step": 5674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020309379607263623,
+      "loss": 0.4382,
+      "step": 5675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020293048787011698,
+      "loss": 0.4867,
+      "step": 5676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020276722863424534,
+      "loss": 0.5413,
+      "step": 5677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020260401839193183,
+      "loss": 0.4498,
+      "step": 5678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020244085717007837,
+      "loss": 0.4657,
+      "step": 5679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002022777449955795,
+      "loss": 0.5236,
+      "step": 5680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020211468189532122,
+      "loss": 0.3933,
+      "step": 5681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020195166789618168,
+      "loss": 0.4744,
+      "step": 5682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020178870302503088,
+      "loss": 0.4735,
+      "step": 5683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020162578730873067,
+      "loss": 0.3866,
+      "step": 5684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020146292077413474,
+      "loss": 0.4501,
+      "step": 5685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020130010344808903,
+      "loss": 0.4982,
+      "step": 5686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020113733535743073,
+      "loss": 0.4259,
+      "step": 5687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020097461652898935,
+      "loss": 0.4304,
+      "step": 5688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020081194698958632,
+      "loss": 0.437,
+      "step": 5689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002006493267660348,
+      "loss": 0.4432,
+      "step": 5690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020048675588513987,
+      "loss": 0.5094,
+      "step": 5691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020032423437369844,
+      "loss": 0.4769,
+      "step": 5692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020016176225849948,
+      "loss": 0.5544,
+      "step": 5693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019999933956632327,
+      "loss": 0.4376,
+      "step": 5694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001998369663239425,
+      "loss": 0.4713,
+      "step": 5695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001996746425581215,
+      "loss": 0.4747,
+      "step": 5696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019951236829561658,
+      "loss": 0.4926,
+      "step": 5697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019935014356317566,
+      "loss": 0.5319,
+      "step": 5698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019918796838753862,
+      "loss": 0.4684,
+      "step": 5699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001990258427954374,
+      "loss": 0.4033,
+      "step": 5700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019886376681359515,
+      "loss": 0.4364,
+      "step": 5701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001987017404687274,
+      "loss": 0.4633,
+      "step": 5702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019853976378754134,
+      "loss": 0.469,
+      "step": 5703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001983778367967361,
+      "loss": 0.5227,
+      "step": 5704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019821595952300214,
+      "loss": 0.4734,
+      "step": 5705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019805413199302225,
+      "loss": 0.4581,
+      "step": 5706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019789235423347084,
+      "loss": 0.5161,
+      "step": 5707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019773062627101423,
+      "loss": 0.4337,
+      "step": 5708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019756894813231012,
+      "loss": 0.4316,
+      "step": 5709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019740731984400846,
+      "loss": 0.3986,
+      "step": 5710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019724574143275075,
+      "loss": 0.4163,
+      "step": 5711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019708421292517043,
+      "loss": 0.048,
+      "step": 5712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019692273434789248,
+      "loss": 0.4637,
+      "step": 5713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019676130572753391,
+      "loss": 0.4818,
+      "step": 5714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019659992709070346,
+      "loss": 0.6699,
+      "step": 5715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019643859846400126,
+      "loss": 0.4895,
+      "step": 5716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019627731987401954,
+      "loss": 0.3648,
+      "step": 5717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019611609134734237,
+      "loss": 0.3795,
+      "step": 5718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019595491291054528,
+      "loss": 0.0484,
+      "step": 5719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019579378459019575,
+      "loss": 0.432,
+      "step": 5720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019563270641285296,
+      "loss": 0.516,
+      "step": 5721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019547167840506792,
+      "loss": 0.048,
+      "step": 5722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019531070059338284,
+      "loss": 0.5605,
+      "step": 5723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019514977300433235,
+      "loss": 0.449,
+      "step": 5724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019498889566444246,
+      "loss": 0.5209,
+      "step": 5725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019482806860023094,
+      "loss": 0.4954,
+      "step": 5726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001946672918382073,
+      "loss": 0.4946,
+      "step": 5727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019450656540487277,
+      "loss": 0.4119,
+      "step": 5728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019434588932672021,
+      "loss": 0.4655,
+      "step": 5729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019418526363023435,
+      "loss": 0.4823,
+      "step": 5730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001940246883418913,
+      "loss": 0.4517,
+      "step": 5731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019386416348815905,
+      "loss": 0.5043,
+      "step": 5732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001937036890954974,
+      "loss": 0.4622,
+      "step": 5733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001935432651903577,
+      "loss": 0.4231,
+      "step": 5734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019338289179918295,
+      "loss": 0.3976,
+      "step": 5735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019322256894840784,
+      "loss": 0.4688,
+      "step": 5736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019306229666445895,
+      "loss": 0.4357,
+      "step": 5737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019290207497375407,
+      "loss": 0.3723,
+      "step": 5738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019274190390270292,
+      "loss": 0.4471,
+      "step": 5739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000192581783477707,
+      "loss": 0.4614,
+      "step": 5740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019242171372515926,
+      "loss": 0.5036,
+      "step": 5741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019226169467144439,
+      "loss": 0.5552,
+      "step": 5742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001921017263429388,
+      "loss": 0.3616,
+      "step": 5743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001919418087660102,
+      "loss": 0.5226,
+      "step": 5744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019178194196701825,
+      "loss": 0.0483,
+      "step": 5745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019162212597231427,
+      "loss": 0.47,
+      "step": 5746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019146236080824113,
+      "loss": 0.4505,
+      "step": 5747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019130264650113306,
+      "loss": 0.3816,
+      "step": 5748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019114298307731621,
+      "loss": 0.4823,
+      "step": 5749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001909833705631084,
+      "loss": 0.3966,
+      "step": 5750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019082380898481893,
+      "loss": 0.0481,
+      "step": 5751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019066429836874844,
+      "loss": 0.3989,
+      "step": 5752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019050483874118962,
+      "loss": 0.4883,
+      "step": 5753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019034543012842658,
+      "loss": 0.542,
+      "step": 5754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001901860725567349,
+      "loss": 0.4128,
+      "step": 5755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000190026766052382,
+      "loss": 0.3704,
+      "step": 5756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018986751064162667,
+      "loss": 0.3605,
+      "step": 5757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018970830635071928,
+      "loss": 0.384,
+      "step": 5758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001895491532059021,
+      "loss": 0.5042,
+      "step": 5759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018939005123340836,
+      "loss": 0.4313,
+      "step": 5760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001892310004594633,
+      "loss": 0.4896,
+      "step": 5761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018907200091028366,
+      "loss": 0.4707,
+      "step": 5762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018891305261207775,
+      "loss": 0.48,
+      "step": 5763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001887541555910453,
+      "loss": 0.5337,
+      "step": 5764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001885953098733777,
+      "loss": 0.5244,
+      "step": 5765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001884365154852581,
+      "loss": 0.4653,
+      "step": 5766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018827777245286042,
+      "loss": 0.4047,
+      "step": 5767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018811908080235102,
+      "loss": 0.3934,
+      "step": 5768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018796044055988725,
+      "loss": 0.5461,
+      "step": 5769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018780185175161824,
+      "loss": 0.4727,
+      "step": 5770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001876433144036845,
+      "loss": 0.5029,
+      "step": 5771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001874848285422181,
+      "loss": 0.4296,
+      "step": 5772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018732639419334285,
+      "loss": 0.4041,
+      "step": 5773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001871680113831734,
+      "loss": 0.451,
+      "step": 5774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018700968013781667,
+      "loss": 0.389,
+      "step": 5775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018685140048337062,
+      "loss": 0.0479,
+      "step": 5776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018669317244592493,
+      "loss": 0.4794,
+      "step": 5777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018653499605156065,
+      "loss": 0.436,
+      "step": 5778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001863768713263504,
+      "loss": 0.4601,
+      "step": 5779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018621879829635817,
+      "loss": 0.4939,
+      "step": 5780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018606077698763975,
+      "loss": 0.411,
+      "step": 5781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018590280742624178,
+      "loss": 0.3641,
+      "step": 5782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018574488963820284,
+      "loss": 0.4095,
+      "step": 5783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018558702364955304,
+      "loss": 0.4121,
+      "step": 5784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018542920948631363,
+      "loss": 0.4686,
+      "step": 5785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001852714471744978,
+      "loss": 0.4844,
+      "step": 5786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018511373674010945,
+      "loss": 0.3773,
+      "step": 5787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018495607820914452,
+      "loss": 0.4406,
+      "step": 5788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018479847160759018,
+      "loss": 0.4697,
+      "step": 5789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018464091696142538,
+      "loss": 0.4841,
+      "step": 5790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018448341429661973,
+      "loss": 0.5247,
+      "step": 5791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018432596363913506,
+      "loss": 0.4778,
+      "step": 5792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018416856501492424,
+      "loss": 0.4285,
+      "step": 5793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018401121844993168,
+      "loss": 0.4854,
+      "step": 5794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018385392397009332,
+      "loss": 0.5232,
+      "step": 5795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018369668160133607,
+      "loss": 0.4878,
+      "step": 5796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018353949136957866,
+      "loss": 0.4695,
+      "step": 5797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018338235330073117,
+      "loss": 0.4408,
+      "step": 5798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018322526742069496,
+      "loss": 0.4921,
+      "step": 5799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018306823375536296,
+      "loss": 0.4797,
+      "step": 5800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018291125233061927,
+      "loss": 0.5045,
+      "step": 5801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001827543231723396,
+      "loss": 0.5097,
+      "step": 5802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018259744630639101,
+      "loss": 0.4424,
+      "step": 5803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018244062175863164,
+      "loss": 0.4507,
+      "step": 5804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018228384955491135,
+      "loss": 0.4999,
+      "step": 5805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018212712972107126,
+      "loss": 0.4332,
+      "step": 5806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001819704622829439,
+      "loss": 0.4077,
+      "step": 5807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018181384726635304,
+      "loss": 0.432,
+      "step": 5808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018165728469711395,
+      "loss": 0.4675,
+      "step": 5809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018150077460103337,
+      "loss": 0.4439,
+      "step": 5810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001813443170039089,
+      "loss": 0.3838,
+      "step": 5811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018118791193152988,
+      "loss": 0.0483,
+      "step": 5812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018103155940967704,
+      "loss": 0.4399,
+      "step": 5813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001808752594641222,
+      "loss": 0.4857,
+      "step": 5814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018071901212062875,
+      "loss": 0.4525,
+      "step": 5815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018056281740495123,
+      "loss": 0.4307,
+      "step": 5816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018040667534283572,
+      "loss": 0.4554,
+      "step": 5817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018025058596001927,
+      "loss": 0.4897,
+      "step": 5818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001800945492822304,
+      "loss": 0.4984,
+      "step": 5819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001799385653351892,
+      "loss": 0.4883,
+      "step": 5820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017978263414460672,
+      "loss": 0.4176,
+      "step": 5821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017962675573618552,
+      "loss": 0.4427,
+      "step": 5822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017947093013561945,
+      "loss": 0.4808,
+      "step": 5823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017931515736859343,
+      "loss": 0.5109,
+      "step": 5824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017915943746078423,
+      "loss": 0.3978,
+      "step": 5825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017900377043785897,
+      "loss": 0.4878,
+      "step": 5826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001788481563254769,
+      "loss": 0.5126,
+      "step": 5827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017869259514928827,
+      "loss": 0.5488,
+      "step": 5828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017853708693493447,
+      "loss": 0.4636,
+      "step": 5829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017838163170804855,
+      "loss": 0.4077,
+      "step": 5830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001782262294942541,
+      "loss": 0.4127,
+      "step": 5831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017807088031916675,
+      "loss": 0.562,
+      "step": 5832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017791558420839293,
+      "loss": 0.4617,
+      "step": 5833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017776034118753065,
+      "loss": 0.4105,
+      "step": 5834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017760515128216865,
+      "loss": 0.4653,
+      "step": 5835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017745001451788738,
+      "loss": 0.5023,
+      "step": 5836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001772949309202585,
+      "loss": 0.505,
+      "step": 5837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017713990051484464,
+      "loss": 0.4584,
+      "step": 5838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017698492332720006,
+      "loss": 0.4376,
+      "step": 5839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017682999938286966,
+      "loss": 0.5005,
+      "step": 5840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017667512870739004,
+      "loss": 0.4832,
+      "step": 5841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000176520311326289,
+      "loss": 0.5265,
+      "step": 5842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001763655472650853,
+      "loss": 0.441,
+      "step": 5843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017621083654928915,
+      "loss": 0.4393,
+      "step": 5844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017605617920440187,
+      "loss": 0.4696,
+      "step": 5845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017590157525591594,
+      "loss": 0.4025,
+      "step": 5846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001757470247293152,
+      "loss": 0.3956,
+      "step": 5847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017559252765007437,
+      "loss": 0.4956,
+      "step": 5848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017543808404365963,
+      "loss": 0.5514,
+      "step": 5849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017528369393552835,
+      "loss": 0.4716,
+      "step": 5850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017512935735112888,
+      "loss": 0.4811,
+      "step": 5851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000174975074315901,
+      "loss": 0.4209,
+      "step": 5852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017482084485527538,
+      "loss": 0.4392,
+      "step": 5853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017466666899467438,
+      "loss": 0.4971,
+      "step": 5854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017451254675951067,
+      "loss": 0.5345,
+      "step": 5855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017435847817518884,
+      "loss": 0.4564,
+      "step": 5856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017420446326710433,
+      "loss": 0.506,
+      "step": 5857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001740505020606437,
+      "loss": 0.5134,
+      "step": 5858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017389659458118484,
+      "loss": 0.5547,
+      "step": 5859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017374274085409663,
+      "loss": 0.4734,
+      "step": 5860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017358894090473926,
+      "loss": 0.0482,
+      "step": 5861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017343519475846364,
+      "loss": 0.4479,
+      "step": 5862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017328150244061224,
+      "loss": 0.4883,
+      "step": 5863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017312786397651853,
+      "loss": 0.4984,
+      "step": 5864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001729742793915071,
+      "loss": 0.5125,
+      "step": 5865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017282074871089365,
+      "loss": 0.4137,
+      "step": 5866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017266727195998493,
+      "loss": 0.4985,
+      "step": 5867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017251384916407915,
+      "loss": 0.4269,
+      "step": 5868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001723604803484649,
+      "loss": 0.4133,
+      "step": 5869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017220716553842253,
+      "loss": 0.4127,
+      "step": 5870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017205390475922322,
+      "loss": 0.4985,
+      "step": 5871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017190069803612934,
+      "loss": 0.4735,
+      "step": 5872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017174754539439448,
+      "loss": 0.4211,
+      "step": 5873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001715944468592628,
+      "loss": 0.4,
+      "step": 5874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017144140245597007,
+      "loss": 0.4648,
+      "step": 5875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017128841220974284,
+      "loss": 0.5424,
+      "step": 5876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017113547614579916,
+      "loss": 0.4273,
+      "step": 5877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017098259428934736,
+      "loss": 0.4648,
+      "step": 5878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017082976666558757,
+      "loss": 0.5149,
+      "step": 5879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017067699329971072,
+      "loss": 0.5011,
+      "step": 5880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017052427421689875,
+      "loss": 0.3927,
+      "step": 5881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017037160944232465,
+      "loss": 0.4921,
+      "step": 5882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017021899900115277,
+      "loss": 0.4183,
+      "step": 5883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017006644291853791,
+      "loss": 0.387,
+      "step": 5884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016991394121962627,
+      "loss": 0.5145,
+      "step": 5885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001697614939295552,
+      "loss": 0.4523,
+      "step": 5886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016960910107345284,
+      "loss": 0.449,
+      "step": 5887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001694567626764385,
+      "loss": 0.483,
+      "step": 5888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016930447876362247,
+      "loss": 0.5175,
+      "step": 5889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001691522493601062,
+      "loss": 0.4467,
+      "step": 5890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001690000744909817,
+      "loss": 0.5671,
+      "step": 5891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001688479541813324,
+      "loss": 0.4658,
+      "step": 5892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001686958884562328,
+      "loss": 0.4327,
+      "step": 5893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016854387734074804,
+      "loss": 0.4384,
+      "step": 5894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001683919208599346,
+      "loss": 0.343,
+      "step": 5895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001682400190388398,
+      "loss": 0.0476,
+      "step": 5896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016808817190250186,
+      "loss": 0.5066,
+      "step": 5897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016793637947595035,
+      "loss": 0.4177,
+      "step": 5898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001677846417842052,
+      "loss": 0.4127,
+      "step": 5899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016763295885227793,
+      "loss": 0.4391,
+      "step": 5900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001674813307051707,
+      "loss": 0.4896,
+      "step": 5901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001673297573678767,
+      "loss": 0.4484,
+      "step": 5902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016717823886538014,
+      "loss": 0.4739,
+      "step": 5903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001670267752226562,
+      "loss": 0.3944,
+      "step": 5904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001668753664646711,
+      "loss": 0.4002,
+      "step": 5905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016672401261638158,
+      "loss": 0.4415,
+      "step": 5906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001665727137027358,
+      "loss": 0.4492,
+      "step": 5907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016642146974867272,
+      "loss": 0.4999,
+      "step": 5908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001662702807791222,
+      "loss": 0.4451,
+      "step": 5909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016611914681900514,
+      "loss": 0.5101,
+      "step": 5910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016596806789323314,
+      "loss": 0.4677,
+      "step": 5911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016581704402670921,
+      "loss": 0.4875,
+      "step": 5912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016566607524432658,
+      "loss": 0.4918,
+      "step": 5913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001655151615709699,
+      "loss": 0.4758,
+      "step": 5914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016536430303151466,
+      "loss": 0.457,
+      "step": 5915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016521349965082738,
+      "loss": 0.4375,
+      "step": 5916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000165062751453765,
+      "loss": 0.5029,
+      "step": 5917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016491205846517576,
+      "loss": 0.4329,
+      "step": 5918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016476142070989886,
+      "loss": 0.455,
+      "step": 5919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001646108382127644,
+      "loss": 0.3767,
+      "step": 5920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016446031099859288,
+      "loss": 0.4615,
+      "step": 5921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016430983909219627,
+      "loss": 0.4342,
+      "step": 5922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016415942251837707,
+      "loss": 0.5177,
+      "step": 5923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016400906130192887,
+      "loss": 0.3753,
+      "step": 5924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001638587554676361,
+      "loss": 0.5629,
+      "step": 5925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016370850504027396,
+      "loss": 0.5065,
+      "step": 5926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016355831004460876,
+      "loss": 0.3896,
+      "step": 5927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016340817050539708,
+      "loss": 0.4259,
+      "step": 5928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016325808644738704,
+      "loss": 0.4642,
+      "step": 5929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016310805789531725,
+      "loss": 0.525,
+      "step": 5930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016295808487391732,
+      "loss": 0.4684,
+      "step": 5931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001628081674079076,
+      "loss": 0.4435,
+      "step": 5932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001626583055219994,
+      "loss": 0.4286,
+      "step": 5933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016250849924089484,
+      "loss": 0.5364,
+      "step": 5934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016235874858928662,
+      "loss": 0.4834,
+      "step": 5935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016220905359185862,
+      "loss": 0.5752,
+      "step": 5936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016205941427328534,
+      "loss": 0.5034,
+      "step": 5937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016190983065823229,
+      "loss": 0.3976,
+      "step": 5938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016176030277135556,
+      "loss": 0.403,
+      "step": 5939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016161083063730226,
+      "loss": 0.4367,
+      "step": 5940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016146141428071014,
+      "loss": 0.4961,
+      "step": 5941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001613120537262081,
+      "loss": 0.451,
+      "step": 5942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001611627489984152,
+      "loss": 0.5236,
+      "step": 5943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016101350012194182,
+      "loss": 0.4344,
+      "step": 5944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016086430712138904,
+      "loss": 0.4819,
+      "step": 5945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016071517002134872,
+      "loss": 0.4485,
+      "step": 5946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016056608884640339,
+      "loss": 0.4789,
+      "step": 5947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016041706362112645,
+      "loss": 0.4021,
+      "step": 5948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016026809437008228,
+      "loss": 0.4285,
+      "step": 5949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016011918111782548,
+      "loss": 0.4962,
+      "step": 5950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015997032388890187,
+      "loss": 0.4141,
+      "step": 5951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000159821522707848,
+      "loss": 0.4065,
+      "step": 5952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015967277759919106,
+      "loss": 0.4691,
+      "step": 5953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015952408858744914,
+      "loss": 0.4993,
+      "step": 5954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015937545569713103,
+      "loss": 0.5726,
+      "step": 5955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015922687895273598,
+      "loss": 0.5384,
+      "step": 5956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015907835837875435,
+      "loss": 0.4226,
+      "step": 5957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001589298939996671,
+      "loss": 0.4669,
+      "step": 5958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015878148583994622,
+      "loss": 0.3817,
+      "step": 5959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015863313392405382,
+      "loss": 0.4616,
+      "step": 5960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015848483827644316,
+      "loss": 0.5021,
+      "step": 5961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001583365989215582,
+      "loss": 0.5081,
+      "step": 5962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015818841588383358,
+      "loss": 0.5109,
+      "step": 5963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015804028918769487,
+      "loss": 0.483,
+      "step": 5964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015789221885755773,
+      "loss": 0.439,
+      "step": 5965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015774420491782904,
+      "loss": 0.5188,
+      "step": 5966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001575962473929064,
+      "loss": 0.4548,
+      "step": 5967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015744834630717798,
+      "loss": 0.402,
+      "step": 5968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015730050168502252,
+      "loss": 0.0482,
+      "step": 5969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015715271355080972,
+      "loss": 0.4851,
+      "step": 5970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015700498192889996,
+      "loss": 0.4824,
+      "step": 5971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015685730684364386,
+      "loss": 0.4855,
+      "step": 5972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015670968831938315,
+      "loss": 0.5352,
+      "step": 5973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015656212638045025,
+      "loss": 0.4766,
+      "step": 5974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000156414621051168,
+      "loss": 0.4967,
+      "step": 5975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015626717235585008,
+      "loss": 0.4065,
+      "step": 5976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015611978031880087,
+      "loss": 0.4742,
+      "step": 5977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015597244496431546,
+      "loss": 0.4755,
+      "step": 5978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015582516631667908,
+      "loss": 0.5256,
+      "step": 5979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015567794440016824,
+      "loss": 0.5529,
+      "step": 5980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015553077923904986,
+      "loss": 0.451,
+      "step": 5981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001553836708575815,
+      "loss": 0.5376,
+      "step": 5982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001552366192800113,
+      "loss": 0.4294,
+      "step": 5983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015508962453057823,
+      "loss": 0.4733,
+      "step": 5984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001549426866335119,
+      "loss": 0.4216,
+      "step": 5985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001547958056130321,
+      "loss": 0.447,
+      "step": 5986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001546489814933496,
+      "loss": 0.5674,
+      "step": 5987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015450221429866596,
+      "loss": 0.4722,
+      "step": 5988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000154355504053173,
+      "loss": 0.5216,
+      "step": 5989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015420885078105335,
+      "loss": 0.5403,
+      "step": 5990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001540622545064802,
+      "loss": 0.3829,
+      "step": 5991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015391571525361742,
+      "loss": 0.3865,
+      "step": 5992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001537692330466195,
+      "loss": 0.4384,
+      "step": 5993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015362280790963106,
+      "loss": 0.4749,
+      "step": 5994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000153476439866788,
+      "loss": 0.4778,
+      "step": 5995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015333012894221638,
+      "loss": 0.048,
+      "step": 5996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015318387516003303,
+      "loss": 0.5096,
+      "step": 5997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001530376785443452,
+      "loss": 0.4403,
+      "step": 5998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015289153911925108,
+      "loss": 0.5416,
+      "step": 5999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015274545690883884,
+      "loss": 0.048,
+      "step": 6000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001525994319371876,
+      "loss": 0.5786,
+      "step": 6001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015245346422836713,
+      "loss": 0.4705,
+      "step": 6002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015230755380643768,
+      "loss": 0.4772,
+      "step": 6003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015216170069544976,
+      "loss": 0.3636,
+      "step": 6004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015201590491944472,
+      "loss": 0.4667,
+      "step": 6005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001518701665024545,
+      "loss": 0.4629,
+      "step": 6006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015172448546850166,
+      "loss": 0.4026,
+      "step": 6007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015157886184159874,
+      "loss": 0.4946,
+      "step": 6008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015143329564574947,
+      "loss": 0.4949,
+      "step": 6009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001512877869049478,
+      "loss": 0.5085,
+      "step": 6010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015114233564317831,
+      "loss": 0.3995,
+      "step": 6011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015099694188441598,
+      "loss": 0.3549,
+      "step": 6012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015085160565262652,
+      "loss": 0.0479,
+      "step": 6013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015070632697176595,
+      "loss": 0.5289,
+      "step": 6014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015056110586578102,
+      "loss": 0.4524,
+      "step": 6015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015041594235860867,
+      "loss": 0.5939,
+      "step": 6016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015027083647417656,
+      "loss": 0.4904,
+      "step": 6017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015012578823640283,
+      "loss": 0.5669,
+      "step": 6018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001499807976691962,
+      "loss": 0.5354,
+      "step": 6019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001498358647964557,
+      "loss": 0.4201,
+      "step": 6020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014969098964207113,
+      "loss": 0.42,
+      "step": 6021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001495461722299225,
+      "loss": 0.3976,
+      "step": 6022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014940141258388024,
+      "loss": 0.5493,
+      "step": 6023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001492567107278055,
+      "loss": 0.4451,
+      "step": 6024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001491120666855499,
+      "loss": 0.5278,
+      "step": 6025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014896748048095538,
+      "loss": 0.5891,
+      "step": 6026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001488229521378544,
+      "loss": 0.4856,
+      "step": 6027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014867848168006998,
+      "loss": 0.4446,
+      "step": 6028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001485340691314156,
+      "loss": 0.4252,
+      "step": 6029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014838971451569484,
+      "loss": 0.3686,
+      "step": 6030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001482454178567021,
+      "loss": 0.5212,
+      "step": 6031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014810117917822218,
+      "loss": 0.6208,
+      "step": 6032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014795699850403028,
+      "loss": 0.5977,
+      "step": 6033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000147812875857892,
+      "loss": 0.6367,
+      "step": 6034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001476688112635634,
+      "loss": 0.5063,
+      "step": 6035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014752480474479102,
+      "loss": 0.453,
+      "step": 6036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001473808563253119,
+      "loss": 0.4542,
+      "step": 6037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014723696602885305,
+      "loss": 0.3873,
+      "step": 6038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014709313387913247,
+      "loss": 0.423,
+      "step": 6039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001469493598998583,
+      "loss": 0.4998,
+      "step": 6040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014680564411472918,
+      "loss": 0.4383,
+      "step": 6041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014666198654743418,
+      "loss": 0.4706,
+      "step": 6042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014651838722165244,
+      "loss": 0.4514,
+      "step": 6043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001463748461610539,
+      "loss": 0.4598,
+      "step": 6044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014623136338929883,
+      "loss": 0.4833,
+      "step": 6045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014608793893003784,
+      "loss": 0.465,
+      "step": 6046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001459445728069118,
+      "loss": 0.4318,
+      "step": 6047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014580126504355206,
+      "loss": 0.4038,
+      "step": 6048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014565801566358044,
+      "loss": 0.5056,
+      "step": 6049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014551482469060906,
+      "loss": 0.4597,
+      "step": 6050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001453716921482406,
+      "loss": 0.4393,
+      "step": 6051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014522861806006755,
+      "loss": 0.5631,
+      "step": 6052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014508560244967328,
+      "loss": 0.426,
+      "step": 6053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014494264534063145,
+      "loss": 0.4601,
+      "step": 6054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014479974675650603,
+      "loss": 0.4637,
+      "step": 6055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014465690672085118,
+      "loss": 0.5394,
+      "step": 6056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014451412525721174,
+      "loss": 0.5566,
+      "step": 6057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014437140238912256,
+      "loss": 0.3825,
+      "step": 6058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014422873814010916,
+      "loss": 0.4907,
+      "step": 6059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014408613253368696,
+      "loss": 0.4873,
+      "step": 6060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014394358559336208,
+      "loss": 0.4973,
+      "step": 6061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001438010973426308,
+      "loss": 0.0479,
+      "step": 6062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014365866780497995,
+      "loss": 0.4561,
+      "step": 6063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014351629700388636,
+      "loss": 0.4507,
+      "step": 6064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001433739849628174,
+      "loss": 0.3875,
+      "step": 6065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001432317317052308,
+      "loss": 0.4829,
+      "step": 6066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014308953725457424,
+      "loss": 0.4893,
+      "step": 6067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001429474016342861,
+      "loss": 0.4639,
+      "step": 6068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001428053248677949,
+      "loss": 0.4388,
+      "step": 6069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014266330697851955,
+      "loss": 0.4659,
+      "step": 6070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014252134798986904,
+      "loss": 0.551,
+      "step": 6071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001423794479252429,
+      "loss": 0.4409,
+      "step": 6072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014223760680803093,
+      "loss": 0.3676,
+      "step": 6073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001420958246616129,
+      "loss": 0.4575,
+      "step": 6074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014195410150935923,
+      "loss": 0.4376,
+      "step": 6075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014181243737463039,
+      "loss": 0.4814,
+      "step": 6076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014167083228077727,
+      "loss": 0.4761,
+      "step": 6077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014152928625114092,
+      "loss": 0.3812,
+      "step": 6078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014138779930905276,
+      "loss": 0.4156,
+      "step": 6079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014124637147783432,
+      "loss": 0.5461,
+      "step": 6080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014110500278079774,
+      "loss": 0.5126,
+      "step": 6081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001409636932412447,
+      "loss": 0.4396,
+      "step": 6082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014082244288246787,
+      "loss": 0.4639,
+      "step": 6083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014068125172774976,
+      "loss": 0.452,
+      "step": 6084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001405401198003634,
+      "loss": 0.5544,
+      "step": 6085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014039904712357166,
+      "loss": 0.5017,
+      "step": 6086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014025803372062795,
+      "loss": 0.4834,
+      "step": 6087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014011707961477589,
+      "loss": 0.3864,
+      "step": 6088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013997618482924933,
+      "loss": 0.4982,
+      "step": 6089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013983534938727212,
+      "loss": 0.4427,
+      "step": 6090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013969457331205844,
+      "loss": 0.4547,
+      "step": 6091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001395538566268129,
+      "loss": 0.4772,
+      "step": 6092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013941319935473006,
+      "loss": 0.048,
+      "step": 6093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001392726015189949,
+      "loss": 0.4653,
+      "step": 6094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001391320631427825,
+      "loss": 0.5063,
+      "step": 6095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013899158424925784,
+      "loss": 0.4695,
+      "step": 6096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013885116486157662,
+      "loss": 0.4525,
+      "step": 6097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013871080500288434,
+      "loss": 0.5581,
+      "step": 6098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013857050469631704,
+      "loss": 0.5757,
+      "step": 6099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001384302639650006,
+      "loss": 0.6628,
+      "step": 6100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001382900828320512,
+      "loss": 0.5118,
+      "step": 6101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001381499613205753,
+      "loss": 0.4799,
+      "step": 6102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013800989945366948,
+      "loss": 0.4207,
+      "step": 6103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013786989725442033,
+      "loss": 0.4459,
+      "step": 6104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013772995474590477,
+      "loss": 0.048,
+      "step": 6105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001375900719511899,
+      "loss": 0.5011,
+      "step": 6106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013745024889333284,
+      "loss": 0.505,
+      "step": 6107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013731048559538102,
+      "loss": 0.5203,
+      "step": 6108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013717078208037192,
+      "loss": 0.4249,
+      "step": 6109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001370311383713333,
+      "loss": 0.5425,
+      "step": 6110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001368915544912827,
+      "loss": 0.4894,
+      "step": 6111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013675203046322826,
+      "loss": 0.4857,
+      "step": 6112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013661256631016793,
+      "loss": 0.5044,
+      "step": 6113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001364731620550899,
+      "loss": 0.4056,
+      "step": 6114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013633381772097265,
+      "loss": 0.4596,
+      "step": 6115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013619453333078447,
+      "loss": 0.5623,
+      "step": 6116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001360553089074841,
+      "loss": 0.4181,
+      "step": 6117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013591614447401996,
+      "loss": 0.4344,
+      "step": 6118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013577704005333097,
+      "loss": 0.4601,
+      "step": 6119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013563799566834606,
+      "loss": 0.4014,
+      "step": 6120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013549901134198418,
+      "loss": 0.4095,
+      "step": 6121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013536008709715447,
+      "loss": 0.4564,
+      "step": 6122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013522122295675615,
+      "loss": 0.4493,
+      "step": 6123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013508241894367868,
+      "loss": 0.4214,
+      "step": 6124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013494367508080102,
+      "loss": 0.5563,
+      "step": 6125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001348049913909929,
+      "loss": 0.4273,
+      "step": 6126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013466636789711384,
+      "loss": 0.4995,
+      "step": 6127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013452780462201348,
+      "loss": 0.5071,
+      "step": 6128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013438930158853169,
+      "loss": 0.5038,
+      "step": 6129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001342508588194979,
+      "loss": 0.4926,
+      "step": 6130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013411247633773204,
+      "loss": 0.5072,
+      "step": 6131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001339741541660442,
+      "loss": 0.4796,
+      "step": 6132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001338358923272343,
+      "loss": 0.4135,
+      "step": 6133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013369769084409211,
+      "loss": 0.3931,
+      "step": 6134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013355954973939794,
+      "loss": 0.4044,
+      "step": 6135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013342146903592178,
+      "loss": 0.4948,
+      "step": 6136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013328344875642383,
+      "loss": 0.4611,
+      "step": 6137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001331454889236543,
+      "loss": 0.5782,
+      "step": 6138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001330075895603536,
+      "loss": 0.4796,
+      "step": 6139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013286975068925168,
+      "loss": 0.5103,
+      "step": 6140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013273197233306893,
+      "loss": 0.5721,
+      "step": 6141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013259425451451573,
+      "loss": 0.4578,
+      "step": 6142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013245659725629244,
+      "loss": 0.4904,
+      "step": 6143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013231900058108937,
+      "loss": 0.472,
+      "step": 6144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013218146451158692,
+      "loss": 0.4459,
+      "step": 6145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013204398907045562,
+      "loss": 0.4971,
+      "step": 6146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013190657428035558,
+      "loss": 0.4795,
+      "step": 6147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013176922016393727,
+      "loss": 0.6119,
+      "step": 6148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013163192674384123,
+      "loss": 0.3936,
+      "step": 6149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001314946940426977,
+      "loss": 0.4415,
+      "step": 6150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013135752208312713,
+      "loss": 0.5436,
+      "step": 6151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013122041088773988,
+      "loss": 0.4647,
+      "step": 6152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013108336047913633,
+      "loss": 0.0477,
+      "step": 6153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001309463708799069,
+      "loss": 0.4901,
+      "step": 6154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013080944211263162,
+      "loss": 0.5206,
+      "step": 6155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013067257419988095,
+      "loss": 0.4415,
+      "step": 6156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001305357671642151,
+      "loss": 0.0478,
+      "step": 6157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001303990210281844,
+      "loss": 0.5186,
+      "step": 6158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001302623358143289,
+      "loss": 0.4523,
+      "step": 6159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001301257115451787,
+      "loss": 0.5039,
+      "step": 6160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012998914824325418,
+      "loss": 0.4252,
+      "step": 6161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012985264593106505,
+      "loss": 0.4122,
+      "step": 6162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001297162046311114,
+      "loss": 0.5266,
+      "step": 6163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012957982436588317,
+      "loss": 0.0478,
+      "step": 6164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012944350515786018,
+      "loss": 0.467,
+      "step": 6165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001293072470295124,
+      "loss": 0.5182,
+      "step": 6166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012917105000329943,
+      "loss": 0.5339,
+      "step": 6167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012903491410167113,
+      "loss": 0.5432,
+      "step": 6168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012889883934706674,
+      "loss": 0.5466,
+      "step": 6169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000128762825761916,
+      "loss": 0.4288,
+      "step": 6170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012862687336863833,
+      "loss": 0.4448,
+      "step": 6171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012849098218964321,
+      "loss": 0.4502,
+      "step": 6172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012835515224732957,
+      "loss": 0.5909,
+      "step": 6173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012821938356408674,
+      "loss": 0.4229,
+      "step": 6174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012808367616229377,
+      "loss": 0.4987,
+      "step": 6175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001279480300643198,
+      "loss": 0.4564,
+      "step": 6176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012781244529252338,
+      "loss": 0.5156,
+      "step": 6177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001276769218692534,
+      "loss": 0.4426,
+      "step": 6178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001275414598168485,
+      "loss": 0.4199,
+      "step": 6179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012740605915763714,
+      "loss": 0.4189,
+      "step": 6180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012727071991393784,
+      "loss": 0.4615,
+      "step": 6181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012713544210805878,
+      "loss": 0.426,
+      "step": 6182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001270002257622983,
+      "loss": 0.4213,
+      "step": 6183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012686507089894412,
+      "loss": 0.474,
+      "step": 6184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012672997754027427,
+      "loss": 0.4705,
+      "step": 6185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001265949457085565,
+      "loss": 0.5093,
+      "step": 6186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012645997542604838,
+      "loss": 0.4137,
+      "step": 6187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012632506671499743,
+      "loss": 0.4822,
+      "step": 6188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012619021959764094,
+      "loss": 0.4329,
+      "step": 6189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001260554340962062,
+      "loss": 0.4182,
+      "step": 6190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001259207102329099,
+      "loss": 0.5121,
+      "step": 6191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012578604802995912,
+      "loss": 0.3719,
+      "step": 6192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012565144750955042,
+      "loss": 0.4938,
+      "step": 6193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001255169086938704,
+      "loss": 0.5286,
+      "step": 6194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012538243160509539,
+      "loss": 0.447,
+      "step": 6195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012524801626539152,
+      "loss": 0.4568,
+      "step": 6196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012511366269691475,
+      "loss": 0.4181,
+      "step": 6197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001249793709218111,
+      "loss": 0.4547,
+      "step": 6198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001248451409622159,
+      "loss": 0.4601,
+      "step": 6199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001247109728402547,
+      "loss": 0.4229,
+      "step": 6200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012457686657804274,
+      "loss": 0.5792,
+      "step": 6201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012444282219768504,
+      "loss": 0.0479,
+      "step": 6202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012430883972127648,
+      "loss": 0.3944,
+      "step": 6203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012417491917090167,
+      "loss": 0.4662,
+      "step": 6204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012404106056863523,
+      "loss": 0.4642,
+      "step": 6205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000123907263936541,
+      "loss": 0.4155,
+      "step": 6206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012377352929667323,
+      "loss": 0.4481,
+      "step": 6207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001236398566710757,
+      "loss": 0.6287,
+      "step": 6208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001235062460817819,
+      "loss": 0.6359,
+      "step": 6209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012337269755081522,
+      "loss": 0.4016,
+      "step": 6210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012323921110018893,
+      "loss": 0.3926,
+      "step": 6211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012310578675190555,
+      "loss": 0.4481,
+      "step": 6212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012297242452795793,
+      "loss": 0.4404,
+      "step": 6213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001228391244503284,
+      "loss": 0.4189,
+      "step": 6214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001227058865409894,
+      "loss": 0.4684,
+      "step": 6215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012257271082190242,
+      "loss": 0.4507,
+      "step": 6216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012243959731501923,
+      "loss": 0.4249,
+      "step": 6217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001223065460422814,
+      "loss": 0.549,
+      "step": 6218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012217355702562,
+      "loss": 0.4897,
+      "step": 6219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000122040630286956,
+      "loss": 0.4406,
+      "step": 6220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012190776584819974,
+      "loss": 0.4365,
+      "step": 6221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001217749637312518,
+      "loss": 0.475,
+      "step": 6222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012164222395800218,
+      "loss": 0.407,
+      "step": 6223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001215095465503307,
+      "loss": 0.4581,
+      "step": 6224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012137693153010687,
+      "loss": 0.432,
+      "step": 6225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012124437891918994,
+      "loss": 0.5032,
+      "step": 6226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012111188873942903,
+      "loss": 0.0476,
+      "step": 6227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001209794610126625,
+      "loss": 0.5567,
+      "step": 6228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012084709576071884,
+      "loss": 0.4298,
+      "step": 6229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001207147930054161,
+      "loss": 0.4835,
+      "step": 6230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012058255276856206,
+      "loss": 0.5392,
+      "step": 6231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012045037507195422,
+      "loss": 0.4369,
+      "step": 6232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012031825993737965,
+      "loss": 0.4825,
+      "step": 6233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012018620738661545,
+      "loss": 0.4875,
+      "step": 6234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012005421744142775,
+      "loss": 0.452,
+      "step": 6235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011992229012357297,
+      "loss": 0.4586,
+      "step": 6236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011979042545479695,
+      "loss": 0.4614,
+      "step": 6237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011965862345683526,
+      "loss": 0.4681,
+      "step": 6238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011952688415141317,
+      "loss": 0.4368,
+      "step": 6239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011939520756024547,
+      "loss": 0.416,
+      "step": 6240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011926359370503692,
+      "loss": 0.42,
+      "step": 6241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011913204260748145,
+      "loss": 0.4218,
+      "step": 6242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011900055428926304,
+      "loss": 0.4614,
+      "step": 6243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011886912877205525,
+      "loss": 0.5436,
+      "step": 6244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011873776607752119,
+      "loss": 0.4349,
+      "step": 6245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011860646622731375,
+      "loss": 0.4377,
+      "step": 6246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011847522924307536,
+      "loss": 0.3813,
+      "step": 6247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011834405514643804,
+      "loss": 0.4092,
+      "step": 6248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011821294395902377,
+      "loss": 0.4492,
+      "step": 6249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011808189570244354,
+      "loss": 0.4756,
+      "step": 6250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011795091039829853,
+      "loss": 0.5723,
+      "step": 6251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011781998806817934,
+      "loss": 0.5404,
+      "step": 6252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011768912873366616,
+      "loss": 0.4204,
+      "step": 6253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011755833241632902,
+      "loss": 0.4966,
+      "step": 6254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011742759913772705,
+      "loss": 0.3951,
+      "step": 6255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011729692891940946,
+      "loss": 0.467,
+      "step": 6256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011716632178291498,
+      "loss": 0.4359,
+      "step": 6257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011703577774977193,
+      "loss": 0.488,
+      "step": 6258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011690529684149798,
+      "loss": 0.4795,
+      "step": 6259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011677487907960072,
+      "loss": 0.386,
+      "step": 6260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011664452448557722,
+      "loss": 0.4961,
+      "step": 6261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011651423308091408,
+      "loss": 0.4817,
+      "step": 6262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011638400488708767,
+      "loss": 0.4563,
+      "step": 6263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011625383992556355,
+      "loss": 0.467,
+      "step": 6264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011612373821779726,
+      "loss": 0.5078,
+      "step": 6265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011599369978523377,
+      "loss": 0.5601,
+      "step": 6266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011586372464930756,
+      "loss": 0.4613,
+      "step": 6267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001157338128314428,
+      "loss": 0.4673,
+      "step": 6268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011560396435305303,
+      "loss": 0.4279,
+      "step": 6269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011547417923554165,
+      "loss": 0.5858,
+      "step": 6270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011534445750030142,
+      "loss": 0.4467,
+      "step": 6271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011521479916871447,
+      "loss": 0.4574,
+      "step": 6272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011508520426215279,
+      "loss": 0.4534,
+      "step": 6273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011495567280197783,
+      "loss": 0.4928,
+      "step": 6274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011482620480954053,
+      "loss": 0.5004,
+      "step": 6275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001146968003061814,
+      "loss": 0.4124,
+      "step": 6276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011456745931323053,
+      "loss": 0.4198,
+      "step": 6277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011443818185200755,
+      "loss": 0.4657,
+      "step": 6278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011430896794382134,
+      "loss": 0.5341,
+      "step": 6279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011417981760997065,
+      "loss": 0.4261,
+      "step": 6280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011405073087174361,
+      "loss": 0.483,
+      "step": 6281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011392170775041788,
+      "loss": 0.0473,
+      "step": 6282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011379274826726066,
+      "loss": 0.0474,
+      "step": 6283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011366385244352862,
+      "loss": 0.5275,
+      "step": 6284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011353502030046808,
+      "loss": 0.4276,
+      "step": 6285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011340625185931452,
+      "loss": 0.4434,
+      "step": 6286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011327754714129323,
+      "loss": 0.4425,
+      "step": 6287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011314890616761886,
+      "loss": 0.4264,
+      "step": 6288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011302032895949571,
+      "loss": 0.3524,
+      "step": 6289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011289181553811733,
+      "loss": 0.4557,
+      "step": 6290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011276336592466696,
+      "loss": 0.4025,
+      "step": 6291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011263498014031726,
+      "loss": 0.4319,
+      "step": 6292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001125066582062304,
+      "loss": 0.5209,
+      "step": 6293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011237840014355777,
+      "loss": 0.4918,
+      "step": 6294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011225020597344055,
+      "loss": 0.4509,
+      "step": 6295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011212207571700933,
+      "loss": 0.5432,
+      "step": 6296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000111994009395384,
+      "loss": 0.4948,
+      "step": 6297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011186600702967432,
+      "loss": 0.5189,
+      "step": 6298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011173806864097885,
+      "loss": 0.5737,
+      "step": 6299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011161019425038604,
+      "loss": 0.5044,
+      "step": 6300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011148238387897386,
+      "loss": 0.4088,
+      "step": 6301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011135463754780961,
+      "loss": 0.5089,
+      "step": 6302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011122695527794985,
+      "loss": 0.4201,
+      "step": 6303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011109933709044073,
+      "loss": 0.5099,
+      "step": 6304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011097178300631794,
+      "loss": 0.4677,
+      "step": 6305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011084429304660648,
+      "loss": 0.4744,
+      "step": 6306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011071686723232099,
+      "loss": 0.4866,
+      "step": 6307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011058950558446506,
+      "loss": 0.4843,
+      "step": 6308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011046220812403213,
+      "loss": 0.4559,
+      "step": 6309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011033497487200494,
+      "loss": 0.0475,
+      "step": 6310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011020780584935564,
+      "loss": 0.4241,
+      "step": 6311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011008070107704583,
+      "loss": 0.4903,
+      "step": 6312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010995366057602641,
+      "loss": 0.4027,
+      "step": 6313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010982668436723775,
+      "loss": 0.4536,
+      "step": 6314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010969977247160984,
+      "loss": 0.4952,
+      "step": 6315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010957292491006154,
+      "loss": 0.4923,
+      "step": 6316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010944614170350153,
+      "loss": 0.5951,
+      "step": 6317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010931942287282786,
+      "loss": 0.4315,
+      "step": 6318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010919276843892784,
+      "loss": 0.4753,
+      "step": 6319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010906617842267819,
+      "loss": 0.3671,
+      "step": 6320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010893965284494501,
+      "loss": 0.4313,
+      "step": 6321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010881319172658405,
+      "loss": 0.5116,
+      "step": 6322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010868679508843976,
+      "loss": 0.5631,
+      "step": 6323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001085604629513467,
+      "loss": 0.4201,
+      "step": 6324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010843419533612831,
+      "loss": 0.0473,
+      "step": 6325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010830799226359767,
+      "loss": 0.5243,
+      "step": 6326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010818185375455714,
+      "loss": 0.4446,
+      "step": 6327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001080557798297983,
+      "loss": 0.4418,
+      "step": 6328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001079297705101025,
+      "loss": 0.4907,
+      "step": 6329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010780382581623977,
+      "loss": 0.5027,
+      "step": 6330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010767794576897,
+      "loss": 0.4536,
+      "step": 6331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010755213038904233,
+      "loss": 0.5514,
+      "step": 6332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010742637969719515,
+      "loss": 0.4254,
+      "step": 6333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010730069371415623,
+      "loss": 0.3875,
+      "step": 6334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010717507246064273,
+      "loss": 0.4407,
+      "step": 6335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010704951595736112,
+      "loss": 0.4913,
+      "step": 6336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001069240242250072,
+      "loss": 0.5753,
+      "step": 6337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010679859728426583,
+      "loss": 0.4735,
+      "step": 6338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010667323515581157,
+      "loss": 0.5573,
+      "step": 6339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010654793786030809,
+      "loss": 0.4454,
+      "step": 6340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010642270541840866,
+      "loss": 0.5371,
+      "step": 6341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001062975378507553,
+      "loss": 0.4255,
+      "step": 6342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001061724351779798,
+      "loss": 0.5492,
+      "step": 6343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010604739742070313,
+      "loss": 0.4515,
+      "step": 6344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010592242459953571,
+      "loss": 0.5322,
+      "step": 6345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010579751673507681,
+      "loss": 0.4044,
+      "step": 6346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010567267384791535,
+      "loss": 0.4189,
+      "step": 6347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010554789595862956,
+      "loss": 0.501,
+      "step": 6348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010542318308778687,
+      "loss": 0.4562,
+      "step": 6349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001052985352559439,
+      "loss": 0.4962,
+      "step": 6350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010517395248364686,
+      "loss": 0.4788,
+      "step": 6351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010504943479143065,
+      "loss": 0.4032,
+      "step": 6352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010492498219982005,
+      "loss": 0.3792,
+      "step": 6353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010480059472932879,
+      "loss": 0.4769,
+      "step": 6354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010467627240045996,
+      "loss": 0.4674,
+      "step": 6355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010455201523370588,
+      "loss": 0.4276,
+      "step": 6356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001044278232495482,
+      "loss": 0.3774,
+      "step": 6357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010430369646845784,
+      "loss": 0.3881,
+      "step": 6358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001041796349108946,
+      "loss": 0.4902,
+      "step": 6359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001040556385973081,
+      "loss": 0.4838,
+      "step": 6360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010393170754813685,
+      "loss": 0.4321,
+      "step": 6361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010380784178380864,
+      "loss": 0.4653,
+      "step": 6362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010368404132474069,
+      "loss": 0.3802,
+      "step": 6363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010356030619133916,
+      "loss": 0.4238,
+      "step": 6364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010343663640399969,
+      "loss": 0.5708,
+      "step": 6365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010331303198310715,
+      "loss": 0.4553,
+      "step": 6366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010318949294903523,
+      "loss": 0.0476,
+      "step": 6367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010306601932214743,
+      "loss": 0.4536,
+      "step": 6368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010294261112279601,
+      "loss": 0.4523,
+      "step": 6369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010281926837132277,
+      "loss": 0.4984,
+      "step": 6370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001026959910880585,
+      "loss": 0.4719,
+      "step": 6371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010257277929332332,
+      "loss": 0.4258,
+      "step": 6372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010244963300742661,
+      "loss": 0.4921,
+      "step": 6373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010232655225066662,
+      "loss": 0.5037,
+      "step": 6374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010220353704333107,
+      "loss": 0.5206,
+      "step": 6375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010208058740569698,
+      "loss": 0.4337,
+      "step": 6376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010195770335803035,
+      "loss": 0.4768,
+      "step": 6377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010183488492058646,
+      "loss": 0.0473,
+      "step": 6378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010171213211360969,
+      "loss": 0.5381,
+      "step": 6379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010158944495733386,
+      "loss": 0.4652,
+      "step": 6380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010146682347198156,
+      "loss": 0.5006,
+      "step": 6381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010134426767776478,
+      "loss": 0.5107,
+      "step": 6382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010122177759488471,
+      "loss": 0.4607,
+      "step": 6383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010109935324353187,
+      "loss": 0.4934,
+      "step": 6384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010097699464388538,
+      "loss": 0.4106,
+      "step": 6385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010085470181611411,
+      "loss": 0.5023,
+      "step": 6386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010073247478037583,
+      "loss": 0.5221,
+      "step": 6387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010061031355681765,
+      "loss": 0.5256,
+      "step": 6388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010048821816557541,
+      "loss": 0.5143,
+      "step": 6389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010036618862677449,
+      "loss": 0.0474,
+      "step": 6390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010024422496052931,
+      "loss": 0.4414,
+      "step": 6391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010012232718694342,
+      "loss": 0.5002,
+      "step": 6392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010000049532610949,
+      "loss": 0.4355,
+      "step": 6393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.987872939810938e-05,
+      "loss": 0.5376,
+      "step": 6394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.975702942301418e-05,
+      "loss": 0.4742,
+      "step": 6395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.963539542088368e-05,
+      "loss": 0.406,
+      "step": 6396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.951382741176723e-05,
+      "loss": 0.5249,
+      "step": 6397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.939232541570315e-05,
+      "loss": 0.5304,
+      "step": 6398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.927088945271895e-05,
+      "loss": 0.498,
+      "step": 6399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.914951954283113e-05,
+      "loss": 0.4991,
+      "step": 6400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.902821570604542e-05,
+      "loss": 0.4619,
+      "step": 6401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.89069779623567e-05,
+      "loss": 0.522,
+      "step": 6402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.878580633174867e-05,
+      "loss": 0.4385,
+      "step": 6403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.866470083419432e-05,
+      "loss": 0.4072,
+      "step": 6404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.854366148965588e-05,
+      "loss": 0.509,
+      "step": 6405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.842268831808454e-05,
+      "loss": 0.514,
+      "step": 6406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.830178133942047e-05,
+      "loss": 0.5469,
+      "step": 6407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.818094057359317e-05,
+      "loss": 0.4142,
+      "step": 6408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.806016604052103e-05,
+      "loss": 0.4171,
+      "step": 6409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.793945776011176e-05,
+      "loss": 0.4933,
+      "step": 6410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78188157522617e-05,
+      "loss": 0.4639,
+      "step": 6411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.769824003685663e-05,
+      "loss": 0.5161,
+      "step": 6412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.75777306337714e-05,
+      "loss": 0.5296,
+      "step": 6413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.745728756286986e-05,
+      "loss": 0.4107,
+      "step": 6414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.733691084400481e-05,
+      "loss": 0.4817,
+      "step": 6415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.721660049701836e-05,
+      "loss": 0.4875,
+      "step": 6416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.709635654174153e-05,
+      "loss": 0.4777,
+      "step": 6417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.697617899799427e-05,
+      "loss": 0.5812,
+      "step": 6418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.685606788558577e-05,
+      "loss": 0.475,
+      "step": 6419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.673602322431424e-05,
+      "loss": 0.5128,
+      "step": 6420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.661604503396692e-05,
+      "loss": 0.5674,
+      "step": 6421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.649613333432011e-05,
+      "loss": 0.3763,
+      "step": 6422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.637628814513927e-05,
+      "loss": 0.4388,
+      "step": 6423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.625650948617842e-05,
+      "loss": 0.6062,
+      "step": 6424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.613679737718112e-05,
+      "loss": 0.5,
+      "step": 6425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.601715183787979e-05,
+      "loss": 0.4203,
+      "step": 6426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.589757288799606e-05,
+      "loss": 0.4651,
+      "step": 6427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.577806054724009e-05,
+      "loss": 0.4259,
+      "step": 6428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.565861483531151e-05,
+      "loss": 0.4574,
+      "step": 6429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.553923577189883e-05,
+      "loss": 0.4066,
+      "step": 6430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.541992337667954e-05,
+      "loss": 0.5443,
+      "step": 6431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.530067766932038e-05,
+      "loss": 0.5262,
+      "step": 6432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.518149866947656e-05,
+      "loss": 0.43,
+      "step": 6433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.506238639679282e-05,
+      "loss": 0.5333,
+      "step": 6434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.494334087090267e-05,
+      "loss": 0.481,
+      "step": 6435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.482436211142869e-05,
+      "loss": 0.0475,
+      "step": 6436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.470545013798237e-05,
+      "loss": 0.4474,
+      "step": 6437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.458660497016425e-05,
+      "loss": 0.4407,
+      "step": 6438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.446782662756404e-05,
+      "loss": 0.5095,
+      "step": 6439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.43491151297599e-05,
+      "loss": 0.3815,
+      "step": 6440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.423047049631955e-05,
+      "loss": 0.3865,
+      "step": 6441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.411189274679937e-05,
+      "loss": 0.4425,
+      "step": 6442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.399338190074485e-05,
+      "loss": 0.5431,
+      "step": 6443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.387493797769037e-05,
+      "loss": 0.4441,
+      "step": 6444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.375656099715935e-05,
+      "loss": 0.4738,
+      "step": 6445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.363825097866424e-05,
+      "loss": 0.3964,
+      "step": 6446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.352000794170606e-05,
+      "loss": 0.4841,
+      "step": 6447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.340183190577534e-05,
+      "loss": 0.4221,
+      "step": 6448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.328372289035114e-05,
+      "loss": 0.5164,
+      "step": 6449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.316568091490174e-05,
+      "loss": 0.4534,
+      "step": 6450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.304770599888424e-05,
+      "loss": 0.4182,
+      "step": 6451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.292979816174474e-05,
+      "loss": 0.5514,
+      "step": 6452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28119574229182e-05,
+      "loss": 0.5867,
+      "step": 6453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.269418380182875e-05,
+      "loss": 0.4596,
+      "step": 6454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.257647731788905e-05,
+      "loss": 0.4593,
+      "step": 6455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.245883799050103e-05,
+      "loss": 0.3982,
+      "step": 6456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.234126583905544e-05,
+      "loss": 0.5574,
+      "step": 6457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.222376088293199e-05,
+      "loss": 0.5139,
+      "step": 6458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.210632314149925e-05,
+      "loss": 0.4434,
+      "step": 6459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.198895263411484e-05,
+      "loss": 0.5121,
+      "step": 6460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.187164938012521e-05,
+      "loss": 0.6655,
+      "step": 6461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.175441339886558e-05,
+      "loss": 0.4713,
+      "step": 6462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.163724470966028e-05,
+      "loss": 0.4587,
+      "step": 6463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.152014333182257e-05,
+      "loss": 0.4517,
+      "step": 6464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.140310928465445e-05,
+      "loss": 0.5549,
+      "step": 6465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.1286142587447e-05,
+      "loss": 0.4673,
+      "step": 6466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.116924325948011e-05,
+      "loss": 0.4691,
+      "step": 6467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.105241132002246e-05,
+      "loss": 0.0477,
+      "step": 6468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.093564678833171e-05,
+      "loss": 0.4463,
+      "step": 6469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.081894968365451e-05,
+      "loss": 0.3699,
+      "step": 6470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.070232002522644e-05,
+      "loss": 0.0476,
+      "step": 6471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.058575783227157e-05,
+      "loss": 0.4463,
+      "step": 6472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.046926312400322e-05,
+      "loss": 0.6089,
+      "step": 6473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.035283591962351e-05,
+      "loss": 0.494,
+      "step": 6474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.023647623832332e-05,
+      "loss": 0.5869,
+      "step": 6475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.012018409928275e-05,
+      "loss": 0.4999,
+      "step": 6476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.000395952167013e-05,
+      "loss": 0.5081,
+      "step": 6477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.988780252464318e-05,
+      "loss": 0.4242,
+      "step": 6478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.977171312734833e-05,
+      "loss": 0.4506,
+      "step": 6479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.965569134892088e-05,
+      "loss": 0.4506,
+      "step": 6480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.953973720848496e-05,
+      "loss": 0.4799,
+      "step": 6481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.942385072515352e-05,
+      "loss": 0.4279,
+      "step": 6482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.930803191802856e-05,
+      "loss": 0.4346,
+      "step": 6483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.919228080620046e-05,
+      "loss": 0.5393,
+      "step": 6484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.907659740874891e-05,
+      "loss": 0.4358,
+      "step": 6485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.896098174474227e-05,
+      "loss": 0.4517,
+      "step": 6486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.884543383323773e-05,
+      "loss": 0.538,
+      "step": 6487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.87299536932813e-05,
+      "loss": 0.4675,
+      "step": 6488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.861454134390784e-05,
+      "loss": 0.4882,
+      "step": 6489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.849919680414115e-05,
+      "loss": 0.5195,
+      "step": 6490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.83839200929935e-05,
+      "loss": 0.5275,
+      "step": 6491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.82687112294664e-05,
+      "loss": 0.4648,
+      "step": 6492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.815357023254983e-05,
+      "loss": 0.0474,
+      "step": 6493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.803849712122292e-05,
+      "loss": 0.5281,
+      "step": 6494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.792349191445331e-05,
+      "loss": 0.4932,
+      "step": 6495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.780855463119757e-05,
+      "loss": 0.3956,
+      "step": 6496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.769368529040133e-05,
+      "loss": 0.5164,
+      "step": 6497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.757888391099838e-05,
+      "loss": 0.5853,
+      "step": 6498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.746415051191187e-05,
+      "loss": 0.4048,
+      "step": 6499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.734948511205359e-05,
+      "loss": 0.4959,
+      "step": 6500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.723488773032406e-05,
+      "loss": 0.5908,
+      "step": 6501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.712035838561272e-05,
+      "loss": 0.422,
+      "step": 6502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.700589709679758e-05,
+      "loss": 0.4129,
+      "step": 6503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.689150388274558e-05,
+      "loss": 0.3931,
+      "step": 6504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.677717876231261e-05,
+      "loss": 0.3967,
+      "step": 6505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.666292175434282e-05,
+      "loss": 0.4089,
+      "step": 6506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.654873287766963e-05,
+      "loss": 0.4482,
+      "step": 6507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.643461215111503e-05,
+      "loss": 0.4901,
+      "step": 6508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.632055959348978e-05,
+      "loss": 0.4066,
+      "step": 6509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.620657522359359e-05,
+      "loss": 0.4066,
+      "step": 6510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.609265906021451e-05,
+      "loss": 0.444,
+      "step": 6511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.597881112212974e-05,
+      "loss": 0.6019,
+      "step": 6512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.586503142810504e-05,
+      "loss": 0.4929,
+      "step": 6513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.575131999689511e-05,
+      "loss": 0.4845,
+      "step": 6514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.563767684724305e-05,
+      "loss": 0.5387,
+      "step": 6515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.552410199788108e-05,
+      "loss": 0.536,
+      "step": 6516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.541059546752988e-05,
+      "loss": 0.4652,
+      "step": 6517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.529715727489911e-05,
+      "loss": 0.5675,
+      "step": 6518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.518378743868716e-05,
+      "loss": 0.4638,
+      "step": 6519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.507048597758071e-05,
+      "loss": 0.5841,
+      "step": 6520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.495725291025569e-05,
+      "loss": 0.538,
+      "step": 6521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.484408825537653e-05,
+      "loss": 0.4258,
+      "step": 6522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.473099203159646e-05,
+      "loss": 0.4778,
+      "step": 6523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.461796425755735e-05,
+      "loss": 0.479,
+      "step": 6524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.45050049518899e-05,
+      "loss": 0.5869,
+      "step": 6525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.439211413321335e-05,
+      "loss": 0.4456,
+      "step": 6526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.427929182013588e-05,
+      "loss": 0.4839,
+      "step": 6527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.416653803125412e-05,
+      "loss": 0.4125,
+      "step": 6528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.405385278515355e-05,
+      "loss": 0.344,
+      "step": 6529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.394123610040838e-05,
+      "loss": 0.5127,
+      "step": 6530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.382868799558142e-05,
+      "loss": 0.5319,
+      "step": 6531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.371620848922435e-05,
+      "loss": 0.4354,
+      "step": 6532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.360379759987729e-05,
+      "loss": 0.5233,
+      "step": 6533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.34914553460694e-05,
+      "loss": 0.4525,
+      "step": 6534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.337918174631798e-05,
+      "loss": 0.541,
+      "step": 6535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.32669768191296e-05,
+      "loss": 0.4281,
+      "step": 6536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.31548405829991e-05,
+      "loss": 0.4183,
+      "step": 6537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.304277305641028e-05,
+      "loss": 0.6257,
+      "step": 6538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.293077425783541e-05,
+      "loss": 0.4742,
+      "step": 6539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.281884420573554e-05,
+      "loss": 0.6049,
+      "step": 6540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.270698291856049e-05,
+      "loss": 0.4313,
+      "step": 6541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.259519041474833e-05,
+      "loss": 0.395,
+      "step": 6542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.248346671272622e-05,
+      "loss": 0.4119,
+      "step": 6543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.23718118309098e-05,
+      "loss": 0.5095,
+      "step": 6544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.226022578770348e-05,
+      "loss": 0.4216,
+      "step": 6545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.214870860150015e-05,
+      "loss": 0.4654,
+      "step": 6546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.203726029068148e-05,
+      "loss": 0.5469,
+      "step": 6547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.19258808736178e-05,
+      "loss": 0.3989,
+      "step": 6548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.181457036866807e-05,
+      "loss": 0.4954,
+      "step": 6549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.170332879417968e-05,
+      "loss": 0.4338,
+      "step": 6550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.159215616848892e-05,
+      "loss": 0.4893,
+      "step": 6551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.148105250992066e-05,
+      "loss": 0.5394,
+      "step": 6552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.137001783678843e-05,
+      "loss": 0.4018,
+      "step": 6553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.12590521673941e-05,
+      "loss": 0.5995,
+      "step": 6554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.114815552002858e-05,
+      "loss": 0.5122,
+      "step": 6555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.10373279129712e-05,
+      "loss": 0.5471,
+      "step": 6556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.092656936449e-05,
+      "loss": 0.4763,
+      "step": 6557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.081587989284133e-05,
+      "loss": 0.521,
+      "step": 6558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.070525951627061e-05,
+      "loss": 0.4639,
+      "step": 6559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.059470825301152e-05,
+      "loss": 0.4503,
+      "step": 6560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.048422612128653e-05,
+      "loss": 0.4631,
+      "step": 6561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.037381313930665e-05,
+      "loss": 0.3835,
+      "step": 6562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.026346932527163e-05,
+      "loss": 0.4553,
+      "step": 6563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.015319469736948e-05,
+      "loss": 0.0474,
+      "step": 6564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.004298927377712e-05,
+      "loss": 0.5139,
+      "step": 6565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.993285307265996e-05,
+      "loss": 0.5203,
+      "step": 6566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.982278611217198e-05,
+      "loss": 0.4379,
+      "step": 6567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.971278841045582e-05,
+      "loss": 0.3878,
+      "step": 6568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.96028599856426e-05,
+      "loss": 0.6246,
+      "step": 6569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.94930008558521e-05,
+      "loss": 0.5004,
+      "step": 6570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.938321103919278e-05,
+      "loss": 0.5084,
+      "step": 6571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.927349055376132e-05,
+      "loss": 0.5394,
+      "step": 6572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.916383941764326e-05,
+      "loss": 0.5094,
+      "step": 6573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.905425764891272e-05,
+      "loss": 0.4837,
+      "step": 6574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.894474526563223e-05,
+      "loss": 0.5698,
+      "step": 6575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.883530228585306e-05,
+      "loss": 0.5405,
+      "step": 6576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.872592872761486e-05,
+      "loss": 0.4641,
+      "step": 6577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.861662460894614e-05,
+      "loss": 0.5298,
+      "step": 6578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.850738994786339e-05,
+      "loss": 0.5459,
+      "step": 6579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.839822476237224e-05,
+      "loss": 0.3602,
+      "step": 6580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.828912907046659e-05,
+      "loss": 0.4845,
+      "step": 6581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.818010289012889e-05,
+      "loss": 0.5267,
+      "step": 6582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.807114623933026e-05,
+      "loss": 0.3936,
+      "step": 6583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.796225913603016e-05,
+      "loss": 0.4406,
+      "step": 6584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.785344159817697e-05,
+      "loss": 0.4391,
+      "step": 6585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.774469364370701e-05,
+      "loss": 0.5326,
+      "step": 6586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.763601529054554e-05,
+      "loss": 0.4835,
+      "step": 6587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.752740655660634e-05,
+      "loss": 0.4829,
+      "step": 6588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.741886745979155e-05,
+      "loss": 0.4584,
+      "step": 6589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.7310398017992e-05,
+      "loss": 0.5265,
+      "step": 6590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.720199824908692e-05,
+      "loss": 0.5175,
+      "step": 6591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.709366817094421e-05,
+      "loss": 0.5188,
+      "step": 6592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.698540780141988e-05,
+      "loss": 0.563,
+      "step": 6593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.687721715835899e-05,
+      "loss": 0.0476,
+      "step": 6594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.676909625959472e-05,
+      "loss": 0.5299,
+      "step": 6595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.666104512294903e-05,
+      "loss": 0.4344,
+      "step": 6596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.655306376623205e-05,
+      "loss": 0.5079,
+      "step": 6597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.644515220724264e-05,
+      "loss": 0.4983,
+      "step": 6598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.633731046376819e-05,
+      "loss": 0.502,
+      "step": 6599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.622953855358456e-05,
+      "loss": 0.5145,
+      "step": 6600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.612183649445581e-05,
+      "loss": 0.4448,
+      "step": 6601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.601420430413486e-05,
+      "loss": 0.5852,
+      "step": 6602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.590664200036296e-05,
+      "loss": 0.0476,
+      "step": 6603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.579914960086987e-05,
+      "loss": 0.542,
+      "step": 6604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.569172712337375e-05,
+      "loss": 0.4885,
+      "step": 6605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.558437458558132e-05,
+      "loss": 0.4812,
+      "step": 6606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.547709200518793e-05,
+      "loss": 0.4167,
+      "step": 6607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.536987939987688e-05,
+      "loss": 0.5101,
+      "step": 6608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.526273678732043e-05,
+      "loss": 0.4133,
+      "step": 6609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.515566418517911e-05,
+      "loss": 0.4341,
+      "step": 6610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.504866161110202e-05,
+      "loss": 0.5001,
+      "step": 6611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.494172908272656e-05,
+      "loss": 0.4539,
+      "step": 6612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.483486661767868e-05,
+      "loss": 0.431,
+      "step": 6613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.472807423357298e-05,
+      "loss": 0.0476,
+      "step": 6614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.46213519480119e-05,
+      "loss": 0.4078,
+      "step": 6615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.451469977858688e-05,
+      "loss": 0.5195,
+      "step": 6616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.440811774287775e-05,
+      "loss": 0.4926,
+      "step": 6617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.430160585845252e-05,
+      "loss": 0.5029,
+      "step": 6618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.419516414286792e-05,
+      "loss": 0.6014,
+      "step": 6619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.408879261366886e-05,
+      "loss": 0.4848,
+      "step": 6620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.398249128838892e-05,
+      "loss": 0.4655,
+      "step": 6621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.387626018455002e-05,
+      "loss": 0.5977,
+      "step": 6622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.377009931966227e-05,
+      "loss": 0.5119,
+      "step": 6623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.366400871122452e-05,
+      "loss": 0.3945,
+      "step": 6624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.355798837672394e-05,
+      "loss": 0.5017,
+      "step": 6625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.345203833363618e-05,
+      "loss": 0.6035,
+      "step": 6626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.334615859942506e-05,
+      "loss": 0.5325,
+      "step": 6627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.32403491915431e-05,
+      "loss": 0.5553,
+      "step": 6628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.313461012743117e-05,
+      "loss": 0.5815,
+      "step": 6629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.302894142451826e-05,
+      "loss": 0.4546,
+      "step": 6630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.292334310022214e-05,
+      "loss": 0.572,
+      "step": 6631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.28178151719488e-05,
+      "loss": 0.4552,
+      "step": 6632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.271235765709261e-05,
+      "loss": 0.5688,
+      "step": 6633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.260697057303644e-05,
+      "loss": 0.4784,
+      "step": 6634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.250165393715141e-05,
+      "loss": 0.048,
+      "step": 6635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.239640776679729e-05,
+      "loss": 0.4767,
+      "step": 6636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.229123207932176e-05,
+      "loss": 0.4789,
+      "step": 6637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.218612689206133e-05,
+      "loss": 0.4852,
+      "step": 6638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.208109222234072e-05,
+      "loss": 0.4487,
+      "step": 6639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.197612808747311e-05,
+      "loss": 0.5569,
+      "step": 6640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.187123450475985e-05,
+      "loss": 0.5072,
+      "step": 6641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.176641149149077e-05,
+      "loss": 0.5087,
+      "step": 6642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.166165906494416e-05,
+      "loss": 0.4688,
+      "step": 6643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.15569772423867e-05,
+      "loss": 0.4451,
+      "step": 6644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.145236604107308e-05,
+      "loss": 0.5155,
+      "step": 6645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.134782547824681e-05,
+      "loss": 0.5489,
+      "step": 6646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.124335557113942e-05,
+      "loss": 0.4698,
+      "step": 6647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.113895633697104e-05,
+      "loss": 0.5271,
+      "step": 6648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.103462779295e-05,
+      "loss": 0.4624,
+      "step": 6649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0930369956273e-05,
+      "loss": 0.4457,
+      "step": 6650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.082618284412518e-05,
+      "loss": 0.4183,
+      "step": 6651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.072206647367974e-05,
+      "loss": 0.7712,
+      "step": 6652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.061802086209857e-05,
+      "loss": 0.5283,
+      "step": 6653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.051404602653166e-05,
+      "loss": 0.4586,
+      "step": 6654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.041014198411755e-05,
+      "loss": 0.4611,
+      "step": 6655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.030630875198286e-05,
+      "loss": 0.474,
+      "step": 6656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.020254634724272e-05,
+      "loss": 0.4213,
+      "step": 6657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.009885478700063e-05,
+      "loss": 0.4647,
+      "step": 6658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.999523408834807e-05,
+      "loss": 0.4797,
+      "step": 6659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.989168426836518e-05,
+      "loss": 0.4948,
+      "step": 6660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.978820534412033e-05,
+      "loss": 0.4249,
+      "step": 6661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.968479733267019e-05,
+      "loss": 0.4884,
+      "step": 6662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.958146025105971e-05,
+      "loss": 0.4568,
+      "step": 6663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.947819411632222e-05,
+      "loss": 0.4338,
+      "step": 6664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.937499894547933e-05,
+      "loss": 0.4705,
+      "step": 6665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.927187475554097e-05,
+      "loss": 0.0478,
+      "step": 6666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.916882156350517e-05,
+      "loss": 0.4641,
+      "step": 6667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.906583938635852e-05,
+      "loss": 0.4235,
+      "step": 6668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.896292824107581e-05,
+      "loss": 0.4974,
+      "step": 6669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.886008814462008e-05,
+      "loss": 0.6305,
+      "step": 6670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.875731911394278e-05,
+      "loss": 0.4825,
+      "step": 6671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.865462116598353e-05,
+      "loss": 0.5303,
+      "step": 6672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.855199431767034e-05,
+      "loss": 0.4642,
+      "step": 6673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.844943858591929e-05,
+      "loss": 0.0482,
+      "step": 6674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.834695398763491e-05,
+      "loss": 0.4952,
+      "step": 6675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.824454053970996e-05,
+      "loss": 0.4385,
+      "step": 6676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.814219825902557e-05,
+      "loss": 0.5563,
+      "step": 6677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.803992716245094e-05,
+      "loss": 0.4695,
+      "step": 6678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.793772726684389e-05,
+      "loss": 0.5298,
+      "step": 6679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.783559858904991e-05,
+      "loss": 0.4757,
+      "step": 6680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.77335411459033e-05,
+      "loss": 0.4356,
+      "step": 6681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.76315549542264e-05,
+      "loss": 0.5125,
+      "step": 6682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.75296400308299e-05,
+      "loss": 0.4956,
+      "step": 6683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.742779639251257e-05,
+      "loss": 0.4541,
+      "step": 6684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.732602405606148e-05,
+      "loss": 0.5863,
+      "step": 6685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.72243230382521e-05,
+      "loss": 0.392,
+      "step": 6686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.712269335584797e-05,
+      "loss": 0.4395,
+      "step": 6687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.702113502560114e-05,
+      "loss": 0.5188,
+      "step": 6688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.691964806425143e-05,
+      "loss": 0.533,
+      "step": 6689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.681823248852731e-05,
+      "loss": 0.5118,
+      "step": 6690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.671688831514527e-05,
+      "loss": 0.4779,
+      "step": 6691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.66156155608102e-05,
+      "loss": 0.434,
+      "step": 6692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.651441424221505e-05,
+      "loss": 0.6368,
+      "step": 6693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.641328437604106e-05,
+      "loss": 0.4546,
+      "step": 6694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.631222597895786e-05,
+      "loss": 0.4086,
+      "step": 6695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.621123906762288e-05,
+      "loss": 0.5503,
+      "step": 6696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.611032365868213e-05,
+      "loss": 0.4786,
+      "step": 6697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.600947976876981e-05,
+      "loss": 0.4663,
+      "step": 6698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.590870741450811e-05,
+      "loss": 0.3881,
+      "step": 6699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.580800661250764e-05,
+      "loss": 0.5135,
+      "step": 6700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.570737737936716e-05,
+      "loss": 0.4614,
+      "step": 6701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.56068197316737e-05,
+      "loss": 0.5616,
+      "step": 6702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.550633368600223e-05,
+      "loss": 0.4753,
+      "step": 6703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.540591925891609e-05,
+      "loss": 0.4191,
+      "step": 6704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5305576466967e-05,
+      "loss": 0.4589,
+      "step": 6705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.520530532669449e-05,
+      "loss": 0.3854,
+      "step": 6706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.510510585462664e-05,
+      "loss": 0.4052,
+      "step": 6707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.500497806727951e-05,
+      "loss": 0.4187,
+      "step": 6708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.490492198115738e-05,
+      "loss": 0.4951,
+      "step": 6709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.480493761275286e-05,
+      "loss": 0.0477,
+      "step": 6710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.47050249785463e-05,
+      "loss": 0.5116,
+      "step": 6711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.46051840950067e-05,
+      "loss": 0.4817,
+      "step": 6712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.450541497859114e-05,
+      "loss": 0.4815,
+      "step": 6713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.440571764574466e-05,
+      "loss": 0.4856,
+      "step": 6714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.430609211290062e-05,
+      "loss": 0.5412,
+      "step": 6715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.420653839648061e-05,
+      "loss": 0.542,
+      "step": 6716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.410705651289434e-05,
+      "loss": 0.5496,
+      "step": 6717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.40076464785394e-05,
+      "loss": 0.5458,
+      "step": 6718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.390830830980194e-05,
+      "loss": 0.4481,
+      "step": 6719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.380904202305604e-05,
+      "loss": 0.6284,
+      "step": 6720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.370984763466403e-05,
+      "loss": 0.5204,
+      "step": 6721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.36107251609765e-05,
+      "loss": 0.5001,
+      "step": 6722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.351167461833168e-05,
+      "loss": 0.496,
+      "step": 6723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.341269602305655e-05,
+      "loss": 0.4186,
+      "step": 6724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.331378939146593e-05,
+      "loss": 0.479,
+      "step": 6725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.321495473986289e-05,
+      "loss": 0.699,
+      "step": 6726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.311619208453834e-05,
+      "loss": 0.6002,
+      "step": 6727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.301750144177177e-05,
+      "loss": 0.4877,
+      "step": 6728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.291888282783054e-05,
+      "loss": 0.5768,
+      "step": 6729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.28203362589701e-05,
+      "loss": 0.465,
+      "step": 6730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.272186175143436e-05,
+      "loss": 0.5717,
+      "step": 6731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.262345932145475e-05,
+      "loss": 0.4612,
+      "step": 6732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.252512898525137e-05,
+      "loss": 0.4767,
+      "step": 6733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.242687075903214e-05,
+      "loss": 0.4146,
+      "step": 6734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.232868465899327e-05,
+      "loss": 0.423,
+      "step": 6735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.223057070131899e-05,
+      "loss": 0.485,
+      "step": 6736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.213252890218163e-05,
+      "loss": 0.5751,
+      "step": 6737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.20345592777416e-05,
+      "loss": 0.4779,
+      "step": 6738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.193666184414764e-05,
+      "loss": 0.6147,
+      "step": 6739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.183883661753614e-05,
+      "loss": 0.4416,
+      "step": 6740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.174108361403203e-05,
+      "loss": 0.5695,
+      "step": 6741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.164340284974806e-05,
+      "loss": 0.494,
+      "step": 6742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.154579434078528e-05,
+      "loss": 0.484,
+      "step": 6743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.144825810323267e-05,
+      "loss": 0.4951,
+      "step": 6744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.135079415316735e-05,
+      "loss": 0.5251,
+      "step": 6745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.125340250665463e-05,
+      "loss": 0.5139,
+      "step": 6746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.115608317974758e-05,
+      "loss": 0.4591,
+      "step": 6747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.105883618848774e-05,
+      "loss": 0.5291,
+      "step": 6748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0961661548904424e-05,
+      "loss": 0.4827,
+      "step": 6749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.086455927701528e-05,
+      "loss": 0.5278,
+      "step": 6750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.076752938882585e-05,
+      "loss": 0.4576,
+      "step": 6751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.067057190032976e-05,
+      "loss": 0.5133,
+      "step": 6752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.057368682750886e-05,
+      "loss": 0.4625,
+      "step": 6753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.047687418633274e-05,
+      "loss": 0.5682,
+      "step": 6754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.038013399275938e-05,
+      "loss": 0.4772,
+      "step": 6755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.028346626273468e-05,
+      "loss": 0.5531,
+      "step": 6756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.01868710121925e-05,
+      "loss": 0.4916,
+      "step": 6757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0090348257055005e-05,
+      "loss": 0.5323,
+      "step": 6758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.999389801323218e-05,
+      "loss": 0.4139,
+      "step": 6759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.98975202966221e-05,
+      "loss": 0.5167,
+      "step": 6760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.980121512311115e-05,
+      "loss": 0.4622,
+      "step": 6761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.97049825085732e-05,
+      "loss": 0.5442,
+      "step": 6762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.960882246887073e-05,
+      "loss": 0.4637,
+      "step": 6763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.951273501985388e-05,
+      "loss": 0.457,
+      "step": 6764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.941672017736111e-05,
+      "loss": 0.4381,
+      "step": 6765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.932077795721863e-05,
+      "loss": 0.55,
+      "step": 6766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922490837524086e-05,
+      "loss": 0.4351,
+      "step": 6767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.912911144723021e-05,
+      "loss": 0.6355,
+      "step": 6768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903338718897705e-05,
+      "loss": 0.4656,
+      "step": 6769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.893773561626003e-05,
+      "loss": 0.5048,
+      "step": 6770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8842156744845385e-05,
+      "loss": 0.5323,
+      "step": 6771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.874665059048767e-05,
+      "loss": 0.5362,
+      "step": 6772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.86512171689294e-05,
+      "loss": 0.4724,
+      "step": 6773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.855585649590112e-05,
+      "loss": 0.5354,
+      "step": 6774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.84605685871214e-05,
+      "loss": 0.4979,
+      "step": 6775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.836535345829652e-05,
+      "loss": 0.5117,
+      "step": 6776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8270211125121207e-05,
+      "loss": 0.4089,
+      "step": 6777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.817514160327797e-05,
+      "loss": 0.5242,
+      "step": 6778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.808014490843727e-05,
+      "loss": 0.5093,
+      "step": 6779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.79852210562577e-05,
+      "loss": 0.5434,
+      "step": 6780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7890370062385733e-05,
+      "loss": 0.4573,
+      "step": 6781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.779559194245587e-05,
+      "loss": 0.589,
+      "step": 6782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7700886712090783e-05,
+      "loss": 0.499,
+      "step": 6783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.760625438690065e-05,
+      "loss": 0.4762,
+      "step": 6784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.75116949824841e-05,
+      "loss": 0.5474,
+      "step": 6785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7417208514427556e-05,
+      "loss": 0.3874,
+      "step": 6786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.732279499830539e-05,
+      "loss": 0.498,
+      "step": 6787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.722845444968011e-05,
+      "loss": 0.5656,
+      "step": 6788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.713418688410199e-05,
+      "loss": 0.4406,
+      "step": 6789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.703999231710949e-05,
+      "loss": 0.4674,
+      "step": 6790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6945870764228745e-05,
+      "loss": 0.5455,
+      "step": 6791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6851822240974074e-05,
+      "loss": 0.5029,
+      "step": 6792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6757846762847786e-05,
+      "loss": 0.0477,
+      "step": 6793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.666394434534e-05,
+      "loss": 0.5338,
+      "step": 6794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.657011500392889e-05,
+      "loss": 0.5651,
+      "step": 6795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.647635875408058e-05,
+      "loss": 0.5298,
+      "step": 6796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6382675611249214e-05,
+      "loss": 0.4813,
+      "step": 6797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.62890655908766e-05,
+      "loss": 0.5043,
+      "step": 6798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.619552870839279e-05,
+      "loss": 0.0478,
+      "step": 6799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.610206497921566e-05,
+      "loss": 0.5356,
+      "step": 6800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.600867441875113e-05,
+      "loss": 0.4729,
+      "step": 6801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5915357042392866e-05,
+      "loss": 0.4647,
+      "step": 6802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.582211286552269e-05,
+      "loss": 0.4182,
+      "step": 6803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.572894190351019e-05,
+      "loss": 0.4368,
+      "step": 6804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.563584417171303e-05,
+      "loss": 0.4576,
+      "step": 6805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.554281968547659e-05,
+      "loss": 0.5249,
+      "step": 6806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.544986846013439e-05,
+      "loss": 0.4583,
+      "step": 6807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.535699051100773e-05,
+      "loss": 0.5625,
+      "step": 6808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.526418585340609e-05,
+      "loss": 0.416,
+      "step": 6809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5171454502626396e-05,
+      "loss": 0.395,
+      "step": 6810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.507879647395392e-05,
+      "loss": 0.4958,
+      "step": 6811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4986211782661664e-05,
+      "loss": 0.5284,
+      "step": 6812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4893700444010706e-05,
+      "loss": 0.5363,
+      "step": 6813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.480126247324968e-05,
+      "loss": 0.3976,
+      "step": 6814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.470889788561545e-05,
+      "loss": 0.0476,
+      "step": 6815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.461660669633267e-05,
+      "loss": 0.4644,
+      "step": 6816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.452438892061396e-05,
+      "loss": 0.5009,
+      "step": 6817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4432244573659695e-05,
+      "loss": 0.5143,
+      "step": 6818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.434017367065841e-05,
+      "loss": 0.4641,
+      "step": 6819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.424817622678618e-05,
+      "loss": 0.5364,
+      "step": 6820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4156252257207217e-05,
+      "loss": 0.4695,
+      "step": 6821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.40644017770735e-05,
+      "loss": 0.5796,
+      "step": 6822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.39726248015251e-05,
+      "loss": 0.4293,
+      "step": 6823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.388092134568967e-05,
+      "loss": 0.4258,
+      "step": 6824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3789291424683027e-05,
+      "loss": 0.4827,
+      "step": 6825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.369773505360864e-05,
+      "loss": 0.6006,
+      "step": 6826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.360625224755811e-05,
+      "loss": 0.4274,
+      "step": 6827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3514843021610584e-05,
+      "loss": 0.5195,
+      "step": 6828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3423507390833294e-05,
+      "loss": 0.6292,
+      "step": 6829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.333224537028131e-05,
+      "loss": 0.5,
+      "step": 6830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.324105697499759e-05,
+      "loss": 0.5312,
+      "step": 6831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.314994222001296e-05,
+      "loss": 0.4928,
+      "step": 6832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.305890112034595e-05,
+      "loss": 0.4446,
+      "step": 6833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.296793369100328e-05,
+      "loss": 0.4211,
+      "step": 6834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.287703994697907e-05,
+      "loss": 0.5095,
+      "step": 6835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.278621990325572e-05,
+      "loss": 0.4916,
+      "step": 6836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.26954735748032e-05,
+      "loss": 0.5148,
+      "step": 6837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2604800976579435e-05,
+      "loss": 0.4833,
+      "step": 6838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.251420212353031e-05,
+      "loss": 0.5533,
+      "step": 6839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.242367703058937e-05,
+      "loss": 0.5175,
+      "step": 6840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.233322571267818e-05,
+      "loss": 0.463,
+      "step": 6841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.224284818470581e-05,
+      "loss": 0.4716,
+      "step": 6842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.215254446156953e-05,
+      "loss": 0.604,
+      "step": 6843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.20623145581543e-05,
+      "loss": 0.3817,
+      "step": 6844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.19721584893329e-05,
+      "loss": 0.4311,
+      "step": 6845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.188207626996599e-05,
+      "loss": 0.5752,
+      "step": 6846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.179206791490204e-05,
+      "loss": 0.4651,
+      "step": 6847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.170213343897745e-05,
+      "loss": 0.5536,
+      "step": 6848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.161227285701603e-05,
+      "loss": 0.47,
+      "step": 6849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1522486183829875e-05,
+      "loss": 0.6018,
+      "step": 6850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.143277343421876e-05,
+      "loss": 0.5018,
+      "step": 6851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.134313462297024e-05,
+      "loss": 0.5085,
+      "step": 6852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1253569764859566e-05,
+      "loss": 0.5745,
+      "step": 6853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.116407887465002e-05,
+      "loss": 0.5001,
+      "step": 6854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1074661967092536e-05,
+      "loss": 0.5576,
+      "step": 6855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.098531905692605e-05,
+      "loss": 0.467,
+      "step": 6856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.089605015887699e-05,
+      "loss": 0.4609,
+      "step": 6857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.080685528765977e-05,
+      "loss": 0.4224,
+      "step": 6858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.071773445797662e-05,
+      "loss": 0.4452,
+      "step": 6859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0628687684517563e-05,
+      "loss": 0.4938,
+      "step": 6860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.053971498196036e-05,
+      "loss": 0.4364,
+      "step": 6861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0450816364970565e-05,
+      "loss": 0.4268,
+      "step": 6862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.036199184820167e-05,
+      "loss": 0.5348,
+      "step": 6863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.027324144629458e-05,
+      "loss": 0.4952,
+      "step": 6864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.018456517387837e-05,
+      "loss": 0.517,
+      "step": 6865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.009596304556974e-05,
+      "loss": 0.0474,
+      "step": 6866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.000743507597322e-05,
+      "loss": 0.6229,
+      "step": 6867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.991898127968098e-05,
+      "loss": 0.4966,
+      "step": 6868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.983060167127312e-05,
+      "loss": 0.5417,
+      "step": 6869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.974229626531762e-05,
+      "loss": 0.5493,
+      "step": 6870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.965406507636972e-05,
+      "loss": 0.4849,
+      "step": 6871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.956590811897299e-05,
+      "loss": 0.5142,
+      "step": 6872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9477825407658496e-05,
+      "loss": 0.4966,
+      "step": 6873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938981695694511e-05,
+      "loss": 0.5365,
+      "step": 6874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.930188278133946e-05,
+      "loss": 0.4354,
+      "step": 6875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9214022895336e-05,
+      "loss": 0.5256,
+      "step": 6876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.912623731341676e-05,
+      "loss": 0.5866,
+      "step": 6877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.903852605005183e-05,
+      "loss": 0.53,
+      "step": 6878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.895088911969864e-05,
+      "loss": 0.472,
+      "step": 6879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8863326536802646e-05,
+      "loss": 0.5421,
+      "step": 6880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.877583831579707e-05,
+      "loss": 0.4653,
+      "step": 6881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.868842447110267e-05,
+      "loss": 0.5057,
+      "step": 6882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860108501712823e-05,
+      "loss": 0.5233,
+      "step": 6883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.851381996826998e-05,
+      "loss": 0.4739,
+      "step": 6884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8426629338912174e-05,
+      "loss": 0.4343,
+      "step": 6885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.833951314342644e-05,
+      "loss": 0.4562,
+      "step": 6886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.825247139617245e-05,
+      "loss": 0.4453,
+      "step": 6887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.816550411149745e-05,
+      "loss": 0.0477,
+      "step": 6888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8078611303736475e-05,
+      "loss": 0.493,
+      "step": 6889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7991792987212294e-05,
+      "loss": 0.6029,
+      "step": 6890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.790504917623545e-05,
+      "loss": 0.4437,
+      "step": 6891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.781837988510396e-05,
+      "loss": 0.5126,
+      "step": 6892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773178512810372e-05,
+      "loss": 0.5424,
+      "step": 6893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.764526491950844e-05,
+      "loss": 0.4525,
+      "step": 6894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.755881927357952e-05,
+      "loss": 0.0473,
+      "step": 6895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.747244820456581e-05,
+      "loss": 0.457,
+      "step": 6896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738615172670407e-05,
+      "loss": 0.5786,
+      "step": 6897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7299929854218826e-05,
+      "loss": 0.5002,
+      "step": 6898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.721378260132225e-05,
+      "loss": 0.452,
+      "step": 6899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7127709982214165e-05,
+      "loss": 0.5438,
+      "step": 6900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.704171201108204e-05,
+      "loss": 0.5457,
+      "step": 6901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6955788702101154e-05,
+      "loss": 0.4661,
+      "step": 6902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6869940069434516e-05,
+      "loss": 0.6616,
+      "step": 6903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.678416612723263e-05,
+      "loss": 0.5955,
+      "step": 6904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6698466889633914e-05,
+      "loss": 0.4792,
+      "step": 6905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.661284237076435e-05,
+      "loss": 0.5438,
+      "step": 6906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.65272925847377e-05,
+      "loss": 0.4895,
+      "step": 6907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.644181754565513e-05,
+      "loss": 0.5223,
+      "step": 6908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.635641726760581e-05,
+      "loss": 0.3837,
+      "step": 6909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.627109176466643e-05,
+      "loss": 0.4818,
+      "step": 6910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6185841050901455e-05,
+      "loss": 0.4732,
+      "step": 6911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6100665140362875e-05,
+      "loss": 0.5521,
+      "step": 6912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.601556404709045e-05,
+      "loss": 0.4071,
+      "step": 6913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.593053778511169e-05,
+      "loss": 0.0477,
+      "step": 6914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5845586368441546e-05,
+      "loss": 0.4365,
+      "step": 6915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576070981108271e-05,
+      "loss": 0.4614,
+      "step": 6916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567590812702571e-05,
+      "loss": 0.5422,
+      "step": 6917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.559118133024853e-05,
+      "loss": 0.4451,
+      "step": 6918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.550652943471695e-05,
+      "loss": 0.4994,
+      "step": 6919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.542195245438424e-05,
+      "loss": 0.4647,
+      "step": 6920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5337450403191424e-05,
+      "loss": 0.4873,
+      "step": 6921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.525302329506736e-05,
+      "loss": 0.4852,
+      "step": 6922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5168671143928144e-05,
+      "loss": 0.5964,
+      "step": 6923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.508439396367775e-05,
+      "loss": 0.4712,
+      "step": 6924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5000191768207856e-05,
+      "loss": 0.5172,
+      "step": 6925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.491606457139763e-05,
+      "loss": 0.5994,
+      "step": 6926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.483201238711404e-05,
+      "loss": 0.5087,
+      "step": 6927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.47480352292115e-05,
+      "loss": 0.6488,
+      "step": 6928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.466413311153239e-05,
+      "loss": 0.5438,
+      "step": 6929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.458030604790614e-05,
+      "loss": 0.4526,
+      "step": 6930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.449655405215036e-05,
+      "loss": 0.5357,
+      "step": 6931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.441287713807007e-05,
+      "loss": 0.5778,
+      "step": 6932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.432927531945791e-05,
+      "loss": 0.5129,
+      "step": 6933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4245748610094115e-05,
+      "loss": 0.4247,
+      "step": 6934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4162297023746746e-05,
+      "loss": 0.5601,
+      "step": 6935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.407892057417112e-05,
+      "loss": 0.4518,
+      "step": 6936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.399561927511042e-05,
+      "loss": 0.4875,
+      "step": 6937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.391239314029544e-05,
+      "loss": 0.5927,
+      "step": 6938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.382924218344459e-05,
+      "loss": 0.49,
+      "step": 6939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.374616641826362e-05,
+      "loss": 0.451,
+      "step": 6940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.366316585844632e-05,
+      "loss": 0.5381,
+      "step": 6941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3580240517673675e-05,
+      "loss": 0.4984,
+      "step": 6942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3497390409614644e-05,
+      "loss": 0.4886,
+      "step": 6943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3414615547925526e-05,
+      "loss": 0.0473,
+      "step": 6944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3331915946250245e-05,
+      "loss": 0.4974,
+      "step": 6945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.324929161822039e-05,
+      "loss": 0.5568,
+      "step": 6946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.316674257745512e-05,
+      "loss": 0.4919,
+      "step": 6947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3084268837561224e-05,
+      "loss": 0.4823,
+      "step": 6948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.300187041213294e-05,
+      "loss": 0.5117,
+      "step": 6949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.291954731475228e-05,
+      "loss": 0.4724,
+      "step": 6950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.28372995589888e-05,
+      "loss": 0.5161,
+      "step": 6951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2755127158399465e-05,
+      "loss": 0.5638,
+      "step": 6952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.267303012652895e-05,
+      "loss": 0.6024,
+      "step": 6953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.25910084769095e-05,
+      "loss": 0.5714,
+      "step": 6954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2509062223061e-05,
+      "loss": 0.5915,
+      "step": 6955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.242719137849077e-05,
+      "loss": 0.502,
+      "step": 6956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2345395956693786e-05,
+      "loss": 0.6011,
+      "step": 6957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2263675971152715e-05,
+      "loss": 0.6058,
+      "step": 6958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.218203143533739e-05,
+      "loss": 0.5181,
+      "step": 6959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.210046236270565e-05,
+      "loss": 0.5204,
+      "step": 6960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.201896876670258e-05,
+      "loss": 0.4789,
+      "step": 6961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1937550660761105e-05,
+      "loss": 0.601,
+      "step": 6962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.185620805830142e-05,
+      "loss": 0.4847,
+      "step": 6963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.177494097273155e-05,
+      "loss": 0.4943,
+      "step": 6964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.16937494174468e-05,
+      "loss": 0.4625,
+      "step": 6965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.161263340583038e-05,
+      "loss": 0.4781,
+      "step": 6966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.153159295125253e-05,
+      "loss": 0.468,
+      "step": 6967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.145062806707151e-05,
+      "loss": 0.49,
+      "step": 6968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.136973876663286e-05,
+      "loss": 0.5275,
+      "step": 6969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.128892506326981e-05,
+      "loss": 0.5646,
+      "step": 6970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.120818697030309e-05,
+      "loss": 0.5315,
+      "step": 6971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.11275245010409e-05,
+      "loss": 0.4961,
+      "step": 6972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.104693766877915e-05,
+      "loss": 0.3928,
+      "step": 6973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0966426486800933e-05,
+      "loss": 0.4431,
+      "step": 6974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.088599096837714e-05,
+      "loss": 0.5021,
+      "step": 6975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.080563112676622e-05,
+      "loss": 0.4911,
+      "step": 6976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.072534697521407e-05,
+      "loss": 0.5576,
+      "step": 6977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.064513852695417e-05,
+      "loss": 0.5304,
+      "step": 6978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.056500579520728e-05,
+      "loss": 0.4475,
+      "step": 6979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0484948793181934e-05,
+      "loss": 0.5952,
+      "step": 6980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0404967534074134e-05,
+      "loss": 0.5776,
+      "step": 6981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.03250620310675e-05,
+      "loss": 0.4089,
+      "step": 6982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.024523229733279e-05,
+      "loss": 0.4214,
+      "step": 6983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.016547834602868e-05,
+      "loss": 0.6075,
+      "step": 6984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0085800190301166e-05,
+      "loss": 0.4286,
+      "step": 6985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0006197843283744e-05,
+      "loss": 0.4844,
+      "step": 6986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.992667131809757e-05,
+      "loss": 0.4608,
+      "step": 6987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.984722062785101e-05,
+      "loss": 0.4698,
+      "step": 6988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.976784578564024e-05,
+      "loss": 0.5078,
+      "step": 6989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.96885468045487e-05,
+      "loss": 0.5215,
+      "step": 6990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.960932369764747e-05,
+      "loss": 0.5045,
+      "step": 6991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.953017647799506e-05,
+      "loss": 0.4633,
+      "step": 6992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9451105158637534e-05,
+      "loss": 0.5135,
+      "step": 6993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.937210975260836e-05,
+      "loss": 0.4247,
+      "step": 6994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.92931902729286e-05,
+      "loss": 0.5542,
+      "step": 6995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.921434673260665e-05,
+      "loss": 0.5999,
+      "step": 6996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9135579144638446e-05,
+      "loss": 0.4773,
+      "step": 6997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.905688752200748e-05,
+      "loss": 0.5255,
+      "step": 6998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.897827187768466e-05,
+      "loss": 0.5836,
+      "step": 6999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.889973222462839e-05,
+      "loss": 0.5146,
+      "step": 7000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.882126857578455e-05,
+      "loss": 0.6034,
+      "step": 7001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.874288094408657e-05,
+      "loss": 0.4791,
+      "step": 7002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8664569342455045e-05,
+      "loss": 0.5352,
+      "step": 7003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8586333783798445e-05,
+      "loss": 0.54,
+      "step": 7004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.850817428101239e-05,
+      "loss": 0.5624,
+      "step": 7005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.843009084698018e-05,
+      "loss": 0.4969,
+      "step": 7006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.835208349457242e-05,
+      "loss": 0.5409,
+      "step": 7007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.827415223664732e-05,
+      "loss": 0.5065,
+      "step": 7008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8196297086050434e-05,
+      "loss": 0.6348,
+      "step": 7009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.811851805561478e-05,
+      "loss": 0.5365,
+      "step": 7010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.804081515816082e-05,
+      "loss": 0.5594,
+      "step": 7011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.79631884064966e-05,
+      "loss": 0.5748,
+      "step": 7012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.788563781341742e-05,
+      "loss": 0.4263,
+      "step": 7013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.780816339170617e-05,
+      "loss": 0.5748,
+      "step": 7014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.773076515413315e-05,
+      "loss": 0.4825,
+      "step": 7015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7653443113456084e-05,
+      "loss": 0.0471,
+      "step": 7016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.757619728242018e-05,
+      "loss": 0.578,
+      "step": 7017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.74990276737579e-05,
+      "loss": 0.554,
+      "step": 7018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.742193430018942e-05,
+      "loss": 0.4614,
+      "step": 7019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7344917174422124e-05,
+      "loss": 0.4791,
+      "step": 7020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.726797630915107e-05,
+      "loss": 0.5977,
+      "step": 7021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.719111171705841e-05,
+      "loss": 0.5197,
+      "step": 7022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.711432341081406e-05,
+      "loss": 0.5071,
+      "step": 7023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.70376114030751e-05,
+      "loss": 0.5928,
+      "step": 7024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.696097570648627e-05,
+      "loss": 0.4669,
+      "step": 7025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.688441633367951e-05,
+      "loss": 0.542,
+      "step": 7026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.680793329727422e-05,
+      "loss": 0.6603,
+      "step": 7027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6731526609877386e-05,
+      "loss": 0.4948,
+      "step": 7028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6655196284083314e-05,
+      "loss": 0.4817,
+      "step": 7029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.657894233247361e-05,
+      "loss": 0.5443,
+      "step": 7030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6502764767617544e-05,
+      "loss": 0.519,
+      "step": 7031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.642666360207142e-05,
+      "loss": 0.5389,
+      "step": 7032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6350638848379304e-05,
+      "loss": 0.469,
+      "step": 7033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.627469051907251e-05,
+      "loss": 0.5497,
+      "step": 7034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6198818626669797e-05,
+      "loss": 0.5674,
+      "step": 7035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.612302318367722e-05,
+      "loss": 0.4238,
+      "step": 7036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6047304202588394e-05,
+      "loss": 0.4681,
+      "step": 7037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.597166169588423e-05,
+      "loss": 0.5311,
+      "step": 7038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.589609567603313e-05,
+      "loss": 0.5681,
+      "step": 7039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5820606155490655e-05,
+      "loss": 0.4783,
+      "step": 7040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.574519314669999e-05,
+      "loss": 0.5117,
+      "step": 7041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.566985666209166e-05,
+      "loss": 0.5009,
+      "step": 7042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.559459671408349e-05,
+      "loss": 0.5166,
+      "step": 7043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5519413315080854e-05,
+      "loss": 0.5277,
+      "step": 7044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.544430647747632e-05,
+      "loss": 0.5634,
+      "step": 7045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5369276213649995e-05,
+      "loss": 0.0473,
+      "step": 7046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.52943225359692e-05,
+      "loss": 0.5579,
+      "step": 7047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5219445456788734e-05,
+      "loss": 0.4886,
+      "step": 7048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.514464498845077e-05,
+      "loss": 0.4241,
+      "step": 7049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.506992114328483e-05,
+      "loss": 0.4299,
+      "step": 7050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.499527393360791e-05,
+      "loss": 0.5963,
+      "step": 7051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.492070337172415e-05,
+      "loss": 0.5048,
+      "step": 7052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.484620946992534e-05,
+      "loss": 0.4784,
+      "step": 7053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4771792240490316e-05,
+      "loss": 0.4829,
+      "step": 7054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.46974516956855e-05,
+      "loss": 0.5463,
+      "step": 7055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.462318784776458e-05,
+      "loss": 0.4729,
+      "step": 7056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4549000708968716e-05,
+      "loss": 0.5232,
+      "step": 7057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.447489029152634e-05,
+      "loss": 0.5399,
+      "step": 7058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.440085660765319e-05,
+      "loss": 0.5851,
+      "step": 7059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.432689966955249e-05,
+      "loss": 0.5396,
+      "step": 7060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4253019489414564e-05,
+      "loss": 0.5437,
+      "step": 7061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.417921607941737e-05,
+      "loss": 0.5199,
+      "step": 7062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.410548945172609e-05,
+      "loss": 0.45,
+      "step": 7063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.403183961849332e-05,
+      "loss": 0.5455,
+      "step": 7064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3958266591858746e-05,
+      "loss": 0.6049,
+      "step": 7065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.388477038394972e-05,
+      "loss": 0.4841,
+      "step": 7066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3811351006880766e-05,
+      "loss": 0.4969,
+      "step": 7067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.373800847275377e-05,
+      "loss": 0.4417,
+      "step": 7068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.366474279365789e-05,
+      "loss": 0.4718,
+      "step": 7069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3591553981669746e-05,
+      "loss": 0.0477,
+      "step": 7070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.351844204885318e-05,
+      "loss": 0.0474,
+      "step": 7071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.34454070072594e-05,
+      "loss": 0.4321,
+      "step": 7072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.337244886892693e-05,
+      "loss": 0.7078,
+      "step": 7073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3299567645881655e-05,
+      "loss": 0.5551,
+      "step": 7074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3226763350136855e-05,
+      "loss": 0.566,
+      "step": 7075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.315403599369282e-05,
+      "loss": 0.4943,
+      "step": 7076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.308138558853746e-05,
+      "loss": 0.3923,
+      "step": 7077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3008812146645914e-05,
+      "loss": 0.4531,
+      "step": 7078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.293631567998062e-05,
+      "loss": 0.4887,
+      "step": 7079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.28638962004914e-05,
+      "loss": 0.4315,
+      "step": 7080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.279155372011522e-05,
+      "loss": 0.5648,
+      "step": 7081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.271928825077652e-05,
+      "loss": 0.4324,
+      "step": 7082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.264709980438701e-05,
+      "loss": 0.4085,
+      "step": 7083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.257498839284556e-05,
+      "loss": 0.5468,
+      "step": 7084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.25029540280386e-05,
+      "loss": 0.4817,
+      "step": 7085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.243099672183958e-05,
+      "loss": 0.5594,
+      "step": 7086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.235911648610951e-05,
+      "loss": 0.5249,
+      "step": 7087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.228731333269646e-05,
+      "loss": 0.5889,
+      "step": 7088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.221558727343604e-05,
+      "loss": 0.4525,
+      "step": 7089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2143938320151e-05,
+      "loss": 0.527,
+      "step": 7090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.207236648465123e-05,
+      "loss": 0.4465,
+      "step": 7091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.200087177873423e-05,
+      "loss": 0.4889,
+      "step": 7092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.192945421418464e-05,
+      "loss": 0.4692,
+      "step": 7093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.185811380277431e-05,
+      "loss": 0.423,
+      "step": 7094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.178685055626252e-05,
+      "loss": 0.4413,
+      "step": 7095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.171566448639568e-05,
+      "loss": 0.5251,
+      "step": 7096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.16445556049077e-05,
+      "loss": 0.4458,
+      "step": 7097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.157352392351942e-05,
+      "loss": 0.4506,
+      "step": 7098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1502569453939255e-05,
+      "loss": 0.4915,
+      "step": 7099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1431692207862736e-05,
+      "loss": 0.4843,
+      "step": 7100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.13608921969728e-05,
+      "loss": 0.5009,
+      "step": 7101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1290169432939555e-05,
+      "loss": 0.5388,
+      "step": 7102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1219523927420336e-05,
+      "loss": 0.473,
+      "step": 7103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.114895569205994e-05,
+      "loss": 0.4861,
+      "step": 7104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.107846473849013e-05,
+      "loss": 0.5449,
+      "step": 7105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1008051078330156e-05,
+      "loss": 0.4471,
+      "step": 7106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.093771472318652e-05,
+      "loss": 0.5972,
+      "step": 7107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.086745568465288e-05,
+      "loss": 0.5986,
+      "step": 7108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.079727397431015e-05,
+      "loss": 0.5055,
+      "step": 7109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.072716960372657e-05,
+      "loss": 0.6194,
+      "step": 7110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.065714258445756e-05,
+      "loss": 0.5391,
+      "step": 7111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.058719292804601e-05,
+      "loss": 0.5007,
+      "step": 7112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0517320646021696e-05,
+      "loss": 0.4648,
+      "step": 7113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0447525749901895e-05,
+      "loss": 0.4258,
+      "step": 7114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.037780825119102e-05,
+      "loss": 0.5281,
+      "step": 7115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.030816816138082e-05,
+      "loss": 0.569,
+      "step": 7116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.023860549195018e-05,
+      "loss": 0.4841,
+      "step": 7117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0169120254365302e-05,
+      "loss": 0.491,
+      "step": 7118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0099712460079696e-05,
+      "loss": 0.5461,
+      "step": 7119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.003038212053383e-05,
+      "loss": 0.5118,
+      "step": 7120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9961129247155662e-05,
+      "loss": 0.5049,
+      "step": 7121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.989195385136029e-05,
+      "loss": 0.505,
+      "step": 7122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9822855944550088e-05,
+      "loss": 0.735,
+      "step": 7123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9753835538114616e-05,
+      "loss": 0.5364,
+      "step": 7124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9684892643430607e-05,
+      "loss": 0.4746,
+      "step": 7125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.961602727186219e-05,
+      "loss": 0.5854,
+      "step": 7126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9547239434760454e-05,
+      "loss": 0.5121,
+      "step": 7127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.947852914346394e-05,
+      "loss": 0.4805,
+      "step": 7128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9409896409298307e-05,
+      "loss": 0.4629,
+      "step": 7129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.934134124357646e-05,
+      "loss": 0.5198,
+      "step": 7130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9272863657598516e-05,
+      "loss": 0.5398,
+      "step": 7131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.920446366265178e-05,
+      "loss": 0.4598,
+      "step": 7132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.913614127001074e-05,
+      "loss": 0.4723,
+      "step": 7133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.906789649093722e-05,
+      "loss": 0.5398,
+      "step": 7134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8999729336680047e-05,
+      "loss": 0.5722,
+      "step": 7135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8931639818475465e-05,
+      "loss": 0.5106,
+      "step": 7136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.886362794754671e-05,
+      "loss": 0.5603,
+      "step": 7137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.879569373510449e-05,
+      "loss": 0.4268,
+      "step": 7138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.872783719234645e-05,
+      "loss": 0.4662,
+      "step": 7139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8660058330457594e-05,
+      "loss": 0.6024,
+      "step": 7140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8592357160610095e-05,
+      "loss": 0.5979,
+      "step": 7141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8524733693963135e-05,
+      "loss": 0.0476,
+      "step": 7142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.845718794166341e-05,
+      "loss": 0.5249,
+      "step": 7143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8389719914844513e-05,
+      "loss": 0.652,
+      "step": 7144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8322329624627497e-05,
+      "loss": 0.4974,
+      "step": 7145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8255017082120305e-05,
+      "loss": 0.7131,
+      "step": 7146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8187782298418397e-05,
+      "loss": 0.4882,
+      "step": 7147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8120625284604075e-05,
+      "loss": 0.0476,
+      "step": 7148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8053546051747036e-05,
+      "loss": 0.4712,
+      "step": 7149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7986544610904106e-05,
+      "loss": 0.5563,
+      "step": 7150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7919620973119342e-05,
+      "loss": 0.4359,
+      "step": 7151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7852775149423802e-05,
+      "loss": 0.6149,
+      "step": 7152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.778600715083596e-05,
+      "loss": 0.5984,
+      "step": 7153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.771931698836122e-05,
+      "loss": 0.4167,
+      "step": 7154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7652704672992303e-05,
+      "loss": 0.5569,
+      "step": 7155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.758617021570925e-05,
+      "loss": 0.5353,
+      "step": 7156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7519713627478794e-05,
+      "loss": 0.4122,
+      "step": 7157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.745333491925528e-05,
+      "loss": 0.3697,
+      "step": 7158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7387034101980067e-05,
+      "loss": 0.4937,
+      "step": 7159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7320811186581695e-05,
+      "loss": 0.4448,
+      "step": 7160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.725466618397576e-05,
+      "loss": 0.533,
+      "step": 7161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7188599105065103e-05,
+      "loss": 0.4833,
+      "step": 7162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.712260996073984e-05,
+      "loss": 0.4998,
+      "step": 7163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7056698761876997e-05,
+      "loss": 0.4614,
+      "step": 7164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6990865519340823e-05,
+      "loss": 0.494,
+      "step": 7165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6925110243982807e-05,
+      "loss": 0.6734,
+      "step": 7166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6859432946641616e-05,
+      "loss": 0.4335,
+      "step": 7167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6793833638142918e-05,
+      "loss": 0.5505,
+      "step": 7168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6728312329299565e-05,
+      "loss": 0.5427,
+      "step": 7169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6662869030911753e-05,
+      "loss": 0.5465,
+      "step": 7170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.659750375376646e-05,
+      "loss": 0.4781,
+      "step": 7171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.653221650863802e-05,
+      "loss": 0.5312,
+      "step": 7172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6467007306287983e-05,
+      "loss": 0.4429,
+      "step": 7173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.640187615746481e-05,
+      "loss": 0.561,
+      "step": 7174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6336823072904303e-05,
+      "loss": 0.5325,
+      "step": 7175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6271848063329275e-05,
+      "loss": 0.4852,
+      "step": 7176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6206951139449708e-05,
+      "loss": 0.541,
+      "step": 7177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.614213231196283e-05,
+      "loss": 0.5155,
+      "step": 7178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6077391591552646e-05,
+      "loss": 0.0475,
+      "step": 7179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6012728988890677e-05,
+      "loss": 0.5677,
+      "step": 7180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5948144514635342e-05,
+      "loss": 0.4897,
+      "step": 7181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.588363817943229e-05,
+      "loss": 0.5289,
+      "step": 7182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5819209993914183e-05,
+      "loss": 0.4424,
+      "step": 7183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.575485996870097e-05,
+      "loss": 0.6376,
+      "step": 7184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5690588114399614e-05,
+      "loss": 0.5007,
+      "step": 7185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5626394441604028e-05,
+      "loss": 0.4235,
+      "step": 7186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5562278960895525e-05,
+      "loss": 0.533,
+      "step": 7187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5498241682842425e-05,
+      "loss": 0.4917,
+      "step": 7188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5434282618000072e-05,
+      "loss": 0.0474,
+      "step": 7189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5370401776911077e-05,
+      "loss": 0.4635,
+      "step": 7190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530659917010497e-05,
+      "loss": 0.5217,
+      "step": 7191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.524287480809845e-05,
+      "loss": 0.4687,
+      "step": 7192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5179228701395496e-05,
+      "loss": 0.519,
+      "step": 7193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5115660860487e-05,
+      "loss": 0.4647,
+      "step": 7194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5052171295850856e-05,
+      "loss": 0.4982,
+      "step": 7195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4988760017952305e-05,
+      "loss": 0.6697,
+      "step": 7196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.49254270372436e-05,
+      "loss": 0.5251,
+      "step": 7197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.486217236416399e-05,
+      "loss": 0.556,
+      "step": 7198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.479899600913993e-05,
+      "loss": 0.4938,
+      "step": 7199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4735897982584965e-05,
+      "loss": 0.6381,
+      "step": 7200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4672878294899557e-05,
+      "loss": 0.4574,
+      "step": 7201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4609936956471513e-05,
+      "loss": 0.51,
+      "step": 7202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4547073977675528e-05,
+      "loss": 0.473,
+      "step": 7203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4484289368873425e-05,
+      "loss": 0.6301,
+      "step": 7204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.442158314041426e-05,
+      "loss": 0.4755,
+      "step": 7205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.435895530263388e-05,
+      "loss": 0.5601,
+      "step": 7206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4296405865855575e-05,
+      "loss": 0.5276,
+      "step": 7207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4233934840389327e-05,
+      "loss": 0.5347,
+      "step": 7208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4171542236532452e-05,
+      "loss": 0.4784,
+      "step": 7209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4109228064569277e-05,
+      "loss": 0.5704,
+      "step": 7210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4046992334771145e-05,
+      "loss": 0.3981,
+      "step": 7211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.398483505739657e-05,
+      "loss": 0.5342,
+      "step": 7212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3922756242691023e-05,
+      "loss": 0.5736,
+      "step": 7213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.386075590088721e-05,
+      "loss": 0.5436,
+      "step": 7214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.379883404220462e-05,
+      "loss": 0.4602,
+      "step": 7215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.373699067685009e-05,
+      "loss": 0.5464,
+      "step": 7216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3675225815017353e-05,
+      "loss": 0.4824,
+      "step": 7217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3613539466887268e-05,
+      "loss": 0.5249,
+      "step": 7218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.35519316426277e-05,
+      "loss": 0.4849,
+      "step": 7219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3490402352393692e-05,
+      "loss": 0.5184,
+      "step": 7220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3428951606327242e-05,
+      "loss": 0.4653,
+      "step": 7221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3367579414557415e-05,
+      "loss": 0.4571,
+      "step": 7222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3306285787200288e-05,
+      "loss": 0.5405,
+      "step": 7223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3245070734359e-05,
+      "loss": 0.5479,
+      "step": 7224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3183934266123806e-05,
+      "loss": 0.506,
+      "step": 7225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.312287639257199e-05,
+      "loss": 0.5688,
+      "step": 7226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3061897123767884e-05,
+      "loss": 0.541,
+      "step": 7227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.300099646976278e-05,
+      "loss": 0.5088,
+      "step": 7228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.294017444059515e-05,
+      "loss": 0.4906,
+      "step": 7229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2879431046290365e-05,
+      "loss": 0.5636,
+      "step": 7230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.281876629686086e-05,
+      "loss": 0.5009,
+      "step": 7231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.275818020230619e-05,
+      "loss": 0.0473,
+      "step": 7232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2697672772612976e-05,
+      "loss": 0.5168,
+      "step": 7233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.263724401775469e-05,
+      "loss": 0.4785,
+      "step": 7234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.257689394769191e-05,
+      "loss": 0.4204,
+      "step": 7235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2516622572372415e-05,
+      "loss": 0.5363,
+      "step": 7236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.24564299017308e-05,
+      "loss": 0.3792,
+      "step": 7237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.239631594568875e-05,
+      "loss": 0.5697,
+      "step": 7238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2336280714155e-05,
+      "loss": 0.4844,
+      "step": 7239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2276324217025245e-05,
+      "loss": 0.625,
+      "step": 7240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2216446464182304e-05,
+      "loss": 0.5667,
+      "step": 7241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2156647465496005e-05,
+      "loss": 0.4453,
+      "step": 7242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2096927230823128e-05,
+      "loss": 0.463,
+      "step": 7243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.203728577000741e-05,
+      "loss": 0.5846,
+      "step": 7244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1977723092879766e-05,
+      "loss": 0.4741,
+      "step": 7245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1918239209258063e-05,
+      "loss": 0.5636,
+      "step": 7246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.185883412894707e-05,
+      "loss": 0.6113,
+      "step": 7247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1799507861738787e-05,
+      "loss": 0.5315,
+      "step": 7248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.174026041741206e-05,
+      "loss": 0.4728,
+      "step": 7249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1681091805732746e-05,
+      "loss": 0.533,
+      "step": 7250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1622002036453814e-05,
+      "loss": 0.5353,
+      "step": 7251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1562991119315035e-05,
+      "loss": 0.682,
+      "step": 7252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1504059064043403e-05,
+      "loss": 0.4698,
+      "step": 7253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1445205880352814e-05,
+      "loss": 0.5389,
+      "step": 7254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1386431577944177e-05,
+      "loss": 0.5096,
+      "step": 7255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1327736166505408e-05,
+      "loss": 0.4839,
+      "step": 7256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1269119655711323e-05,
+      "loss": 0.4439,
+      "step": 7257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.121058205522397e-05,
+      "loss": 0.5776,
+      "step": 7258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1152123374692132e-05,
+      "loss": 0.5652,
+      "step": 7259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.109374362375166e-05,
+      "loss": 0.5737,
+      "step": 7260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1035442812025462e-05,
+      "loss": 0.5359,
+      "step": 7261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.097722094912341e-05,
+      "loss": 0.5223,
+      "step": 7262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.091907804464227e-05,
+      "loss": 0.4915,
+      "step": 7263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0861014108165986e-05,
+      "loss": 0.4897,
+      "step": 7264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0803029149265406e-05,
+      "loss": 0.54,
+      "step": 7265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0745123177498104e-05,
+      "loss": 0.5365,
+      "step": 7266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0687296202409002e-05,
+      "loss": 0.493,
+      "step": 7267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.062954823352986e-05,
+      "loss": 0.4775,
+      "step": 7268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0571879280379345e-05,
+      "loss": 0.542,
+      "step": 7269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0514289352463245e-05,
+      "loss": 0.4603,
+      "step": 7270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.045677845927413e-05,
+      "loss": 0.4107,
+      "step": 7271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0399346610291747e-05,
+      "loss": 0.4984,
+      "step": 7272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0341993814982752e-05,
+      "loss": 0.5873,
+      "step": 7273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0284720082800636e-05,
+      "loss": 0.452,
+      "step": 7274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0227525423185955e-05,
+      "loss": 0.5464,
+      "step": 7275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0170409845566283e-05,
+      "loss": 0.5157,
+      "step": 7276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0113373359356146e-05,
+      "loss": 0.5286,
+      "step": 7277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.005641597395691e-05,
+      "loss": 0.557,
+      "step": 7278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9999537698757064e-05,
+      "loss": 0.4436,
+      "step": 7279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9942738543131944e-05,
+      "loss": 0.453,
+      "step": 7280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9886018516443948e-05,
+      "loss": 0.5887,
+      "step": 7281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9829377628042266e-05,
+      "loss": 0.5016,
+      "step": 7282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9772815887263207e-05,
+      "loss": 0.5629,
+      "step": 7283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.971633330342998e-05,
+      "loss": 0.5836,
+      "step": 7284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.965992988585269e-05,
+      "loss": 0.5768,
+      "step": 7285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9603605643828513e-05,
+      "loss": 0.5077,
+      "step": 7286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9547360586641473e-05,
+      "loss": 0.5859,
+      "step": 7287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9491194723562534e-05,
+      "loss": 0.5559,
+      "step": 7288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9435108063849684e-05,
+      "loss": 0.5798,
+      "step": 7289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9379100616747747e-05,
+      "loss": 0.5114,
+      "step": 7290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9323172391488676e-05,
+      "loss": 0.5479,
+      "step": 7291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.926732339729115e-05,
+      "loss": 0.4644,
+      "step": 7292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9211553643360913e-05,
+      "loss": 0.5995,
+      "step": 7293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9155863138890672e-05,
+      "loss": 0.4789,
+      "step": 7294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9100251893060026e-05,
+      "loss": 0.5176,
+      "step": 7295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9044719915035367e-05,
+      "loss": 0.4863,
+      "step": 7296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.898926721397032e-05,
+      "loss": 0.5013,
+      "step": 7297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.893389379900512e-05,
+      "loss": 0.511,
+      "step": 7298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.887859967926725e-05,
+      "loss": 0.427,
+      "step": 7299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8823384863870864e-05,
+      "loss": 0.5275,
+      "step": 7300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8768249361917234e-05,
+      "loss": 0.5678,
+      "step": 7301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8713193182494424e-05,
+      "loss": 0.4761,
+      "step": 7302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8658216334677402e-05,
+      "loss": 0.54,
+      "step": 7303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.860331882752825e-05,
+      "loss": 0.4581,
+      "step": 7304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8548500670095724e-05,
+      "loss": 0.4301,
+      "step": 7305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8493761871415772e-05,
+      "loss": 0.6289,
+      "step": 7306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8439102440510946e-05,
+      "loss": 0.4353,
+      "step": 7307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8384522386391043e-05,
+      "loss": 0.5504,
+      "step": 7308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.833002171805259e-05,
+      "loss": 0.739,
+      "step": 7309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.827560044447896e-05,
+      "loss": 0.5165,
+      "step": 7310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8221258574640575e-05,
+      "loss": 0.5153,
+      "step": 7311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.816699611749473e-05,
+      "loss": 0.4342,
+      "step": 7312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.81128130819857e-05,
+      "loss": 0.5338,
+      "step": 7313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8058709477044523e-05,
+      "loss": 0.5463,
+      "step": 7314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.800468531158922e-05,
+      "loss": 0.4935,
+      "step": 7315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.795074059452484e-05,
+      "loss": 0.5178,
+      "step": 7316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7896875334743046e-05,
+      "loss": 0.4084,
+      "step": 7317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.784308954112268e-05,
+      "loss": 0.5411,
+      "step": 7318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7789383222529267e-05,
+      "loss": 0.5753,
+      "step": 7319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7735756387815495e-05,
+      "loss": 0.5277,
+      "step": 7320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7682209045820684e-05,
+      "loss": 0.4346,
+      "step": 7321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7628741205371224e-05,
+      "loss": 0.443,
+      "step": 7322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7575352875280283e-05,
+      "loss": 0.4963,
+      "step": 7323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7522044064348042e-05,
+      "loss": 0.5916,
+      "step": 7324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7468814781361476e-05,
+      "loss": 0.4458,
+      "step": 7325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7415665035094453e-05,
+      "loss": 0.0473,
+      "step": 7326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7362594834307855e-05,
+      "loss": 0.449,
+      "step": 7327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7309604187749293e-05,
+      "loss": 0.5256,
+      "step": 7328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.72566931041534e-05,
+      "loss": 0.4305,
+      "step": 7329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.720386159224163e-05,
+      "loss": 0.5203,
+      "step": 7330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7151109660722308e-05,
+      "loss": 0.5972,
+      "step": 7331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7098437318290584e-05,
+      "loss": 0.4836,
+      "step": 7332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.704584457362862e-05,
+      "loss": 0.4902,
+      "step": 7333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.699333143540538e-05,
+      "loss": 0.5385,
+      "step": 7334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6940897912276765e-05,
+      "loss": 0.5931,
+      "step": 7335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.688854401288542e-05,
+      "loss": 0.4973,
+      "step": 7336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.683626974586111e-05,
+      "loss": 0.4196,
+      "step": 7337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.678407511982022e-05,
+      "loss": 0.5892,
+      "step": 7338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6731960143366143e-05,
+      "loss": 0.5093,
+      "step": 7339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6679924825089066e-05,
+      "loss": 0.5005,
+      "step": 7340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6627969173566127e-05,
+      "loss": 0.6311,
+      "step": 7341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.657609319736125e-05,
+      "loss": 0.431,
+      "step": 7342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6524296905025325e-05,
+      "loss": 0.5018,
+      "step": 7343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.647258030509602e-05,
+      "loss": 0.4846,
+      "step": 7344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6420943406097954e-05,
+      "loss": 0.4664,
+      "step": 7345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6369386216542548e-05,
+      "loss": 0.4846,
+      "step": 7346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6317908744928e-05,
+      "loss": 0.5861,
+      "step": 7347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266510999739525e-05,
+      "loss": 0.5388,
+      "step": 7348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215192989449125e-05,
+      "loss": 0.4785,
+      "step": 7349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.616395472251564e-05,
+      "loss": 0.5008,
+      "step": 7350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6112796207384818e-05,
+      "loss": 0.4497,
+      "step": 7351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6061717452489245e-05,
+      "loss": 0.5232,
+      "step": 7352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.601071846624841e-05,
+      "loss": 0.4655,
+      "step": 7353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5959799257068475e-05,
+      "loss": 0.4462,
+      "step": 7354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.590895983334256e-05,
+      "loss": 0.6841,
+      "step": 7355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5858200203450744e-05,
+      "loss": 0.5138,
+      "step": 7356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5807520375759776e-05,
+      "loss": 0.5303,
+      "step": 7357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.575692035862336e-05,
+      "loss": 0.4596,
+      "step": 7358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5706400160382107e-05,
+      "loss": 0.5341,
+      "step": 7359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5655959789363182e-05,
+      "loss": 0.6113,
+      "step": 7360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5605599253880886e-05,
+      "loss": 0.6218,
+      "step": 7361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.55553185622363e-05,
+      "loss": 0.4509,
+      "step": 7362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5505117722717288e-05,
+      "loss": 0.4261,
+      "step": 7363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5454996743598514e-05,
+      "loss": 0.4929,
+      "step": 7364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.540495563314154e-05,
+      "loss": 0.5157,
+      "step": 7365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.535499439959481e-05,
+      "loss": 0.0477,
+      "step": 7366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5305113051193474e-05,
+      "loss": 0.0476,
+      "step": 7367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5255311596159726e-05,
+      "loss": 0.5303,
+      "step": 7368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5205590042702332e-05,
+      "loss": 0.4971,
+      "step": 7369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5155948399017016e-05,
+      "loss": 0.4944,
+      "step": 7370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5106386673286344e-05,
+      "loss": 0.4999,
+      "step": 7371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5056904873679722e-05,
+      "loss": 0.4817,
+      "step": 7372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5007503008353296e-05,
+      "loss": 0.0479,
+      "step": 7373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4958181085450107e-05,
+      "loss": 0.6525,
+      "step": 7374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4908939113100095e-05,
+      "loss": 0.5256,
+      "step": 7375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4859777099419769e-05,
+      "loss": 0.4471,
+      "step": 7376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4810695052512646e-05,
+      "loss": 0.4855,
+      "step": 7377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4761692980469144e-05,
+      "loss": 0.4789,
+      "step": 7378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4712770891366246e-05,
+      "loss": 0.4797,
+      "step": 7379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4663928793268e-05,
+      "loss": 0.3944,
+      "step": 7380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4615166694225135e-05,
+      "loss": 0.458,
+      "step": 7381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4566484602275221e-05,
+      "loss": 0.4886,
+      "step": 7382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4517882525442616e-05,
+      "loss": 0.4791,
+      "step": 7383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4469360471738469e-05,
+      "loss": 0.6072,
+      "step": 7384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4420918449160881e-05,
+      "loss": 0.5259,
+      "step": 7385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4372556465694576e-05,
+      "loss": 0.4976,
+      "step": 7386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4324274529311233e-05,
+      "loss": 0.6414,
+      "step": 7387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4276072647969207e-05,
+      "loss": 0.4382,
+      "step": 7388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4227950829613811e-05,
+      "loss": 0.5936,
+      "step": 7389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4179909082177033e-05,
+      "loss": 0.4551,
+      "step": 7390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4131947413577705e-05,
+      "loss": 0.5255,
+      "step": 7391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4084065831721394e-05,
+      "loss": 0.6429,
+      "step": 7392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4036264344500615e-05,
+      "loss": 0.5874,
+      "step": 7393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3988542959794625e-05,
+      "loss": 0.5242,
+      "step": 7394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3940901685469298e-05,
+      "loss": 0.5834,
+      "step": 7395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3893340529377629e-05,
+      "loss": 0.5349,
+      "step": 7396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.384585949935918e-05,
+      "loss": 0.5491,
+      "step": 7397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.379845860324025e-05,
+      "loss": 0.4983,
+      "step": 7398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3751137848834138e-05,
+      "loss": 0.5331,
+      "step": 7399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3703897243940833e-05,
+      "loss": 0.5687,
+      "step": 7400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3656736796347102e-05,
+      "loss": 0.5497,
+      "step": 7401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3609656513826563e-05,
+      "loss": 0.5791,
+      "step": 7402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3562656404139395e-05,
+      "loss": 0.5211,
+      "step": 7403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3515736475032847e-05,
+      "loss": 0.3898,
+      "step": 7404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3468896734240898e-05,
+      "loss": 0.5083,
+      "step": 7405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3422137189484207e-05,
+      "loss": 0.4465,
+      "step": 7406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3375457848470162e-05,
+      "loss": 0.5054,
+      "step": 7407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.332885871889311e-05,
+      "loss": 0.5138,
+      "step": 7408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3282339808434074e-05,
+      "loss": 0.4611,
+      "step": 7409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3235901124760919e-05,
+      "loss": 0.5815,
+      "step": 7410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3189542675528133e-05,
+      "loss": 0.5345,
+      "step": 7411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.314326446837727e-05,
+      "loss": 0.5906,
+      "step": 7412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3097066510936283e-05,
+      "loss": 0.4426,
+      "step": 7413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3050948810820141e-05,
+      "loss": 0.5446,
+      "step": 7414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.300491137563059e-05,
+      "loss": 0.0476,
+      "step": 7415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2958954212956009e-05,
+      "loss": 0.5959,
+      "step": 7416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.291307733037167e-05,
+      "loss": 0.5386,
+      "step": 7417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2867280735439524e-05,
+      "loss": 0.5729,
+      "step": 7418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2821564435708421e-05,
+      "loss": 0.559,
+      "step": 7419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.277592843871378e-05,
+      "loss": 0.5118,
+      "step": 7420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.273037275197797e-05,
+      "loss": 0.5966,
+      "step": 7421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.268489738300993e-05,
+      "loss": 0.5466,
+      "step": 7422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2639502339305553e-05,
+      "loss": 0.4965,
+      "step": 7423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2594187628347409e-05,
+      "loss": 0.4425,
+      "step": 7424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.25489532576048e-05,
+      "loss": 0.5485,
+      "step": 7425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2503799234533819e-05,
+      "loss": 0.6732,
+      "step": 7426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2458725566577289e-05,
+      "loss": 0.4894,
+      "step": 7427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2413732261164824e-05,
+      "loss": 0.4774,
+      "step": 7428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2368819325712821e-05,
+      "loss": 0.5505,
+      "step": 7429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2323986767624251e-05,
+      "loss": 0.6165,
+      "step": 7430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.227923459428909e-05,
+      "loss": 0.5862,
+      "step": 7431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2234562813083883e-05,
+      "loss": 0.5679,
+      "step": 7432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2189971431372016e-05,
+      "loss": 0.5097,
+      "step": 7433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2145460456503609e-05,
+      "loss": 0.5322,
+      "step": 7434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2101029895815407e-05,
+      "loss": 0.5145,
+      "step": 7435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2056679756631106e-05,
+      "loss": 0.506,
+      "step": 7436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2012410046260968e-05,
+      "loss": 0.4989,
+      "step": 7437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1968220772002103e-05,
+      "loss": 0.546,
+      "step": 7438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1924111941138294e-05,
+      "loss": 0.4954,
+      "step": 7439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.188008356094017e-05,
+      "loss": 0.4711,
+      "step": 7440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1836135638665035e-05,
+      "loss": 0.5267,
+      "step": 7441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1792268181556875e-05,
+      "loss": 0.4282,
+      "step": 7442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1748481196846406e-05,
+      "loss": 0.5544,
+      "step": 7443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1704774691751241e-05,
+      "loss": 0.5493,
+      "step": 7444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1661148673475619e-05,
+      "loss": 0.5421,
+      "step": 7445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1617603149210454e-05,
+      "loss": 0.5204,
+      "step": 7446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1574138126133505e-05,
+      "loss": 0.4557,
+      "step": 7447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.153075361140915e-05,
+      "loss": 0.6217,
+      "step": 7448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1487449612188616e-05,
+      "loss": 0.6171,
+      "step": 7449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1444226135609859e-05,
+      "loss": 0.5281,
+      "step": 7450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1401083188797346e-05,
+      "loss": 0.6041,
+      "step": 7451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1358020778862499e-05,
+      "loss": 0.5327,
+      "step": 7452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.131503891290342e-05,
+      "loss": 0.5627,
+      "step": 7453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1272137598004884e-05,
+      "loss": 0.58,
+      "step": 7454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1229316841238457e-05,
+      "loss": 0.3986,
+      "step": 7455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1186576649662327e-05,
+      "loss": 0.4526,
+      "step": 7456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.114391703032147e-05,
+      "loss": 0.5604,
+      "step": 7457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1101337990247595e-05,
+      "loss": 0.5175,
+      "step": 7458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1058839536459086e-05,
+      "loss": 0.4596,
+      "step": 7459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1016421675961009e-05,
+      "loss": 0.649,
+      "step": 7460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0974084415745267e-05,
+      "loss": 0.5029,
+      "step": 7461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0931827762790392e-05,
+      "loss": 0.5602,
+      "step": 7462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.08896517240617e-05,
+      "loss": 0.4446,
+      "step": 7463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0847556306511019e-05,
+      "loss": 0.478,
+      "step": 7464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0805541517077133e-05,
+      "loss": 0.509,
+      "step": 7465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0763607362685446e-05,
+      "loss": 0.4393,
+      "step": 7466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0721753850247984e-05,
+      "loss": 0.4551,
+      "step": 7467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0679980986663674e-05,
+      "loss": 0.4396,
+      "step": 7468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0638288778817894e-05,
+      "loss": 0.5406,
+      "step": 7469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.059667723358304e-05,
+      "loss": 0.4453,
+      "step": 7470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0555146357817846e-05,
+      "loss": 0.4329,
+      "step": 7471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0513696158368057e-05,
+      "loss": 0.4703,
+      "step": 7472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0472326642065988e-05,
+      "loss": 0.5875,
+      "step": 7473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0431037815730682e-05,
+      "loss": 0.4683,
+      "step": 7474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0389829686167807e-05,
+      "loss": 0.5243,
+      "step": 7475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0348702260169873e-05,
+      "loss": 0.5587,
+      "step": 7476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.030765554451596e-05,
+      "loss": 0.0474,
+      "step": 7477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.026668954597193e-05,
+      "loss": 0.4999,
+      "step": 7478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0225804271290218e-05,
+      "loss": 0.4501,
+      "step": 7479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0184999727210154e-05,
+      "loss": 0.0475,
+      "step": 7480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0144275920457524e-05,
+      "loss": 0.5304,
+      "step": 7481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0103632857745016e-05,
+      "loss": 0.4963,
+      "step": 7482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.006307054577188e-05,
+      "loss": 0.4907,
+      "step": 7483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0022588991224157e-05,
+      "loss": 0.5502,
+      "step": 7484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.982188200774455e-06,
+      "loss": 0.4458,
+      "step": 7485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.941868181082115e-06,
+      "loss": 0.4948,
+      "step": 7486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.901628938793206e-06,
+      "loss": 0.499,
+      "step": 7487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.861470480540424e-06,
+      "loss": 0.6323,
+      "step": 7488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.821392812943253e-06,
+      "loss": 0.5701,
+      "step": 7489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.781395942607685e-06,
+      "loss": 0.4441,
+      "step": 7490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.741479876126614e-06,
+      "loss": 0.3784,
+      "step": 7491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.701644620079386e-06,
+      "loss": 0.6035,
+      "step": 7492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.661890181032195e-06,
+      "loss": 0.5161,
+      "step": 7493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.622216565537856e-06,
+      "loss": 0.5367,
+      "step": 7494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.582623780135801e-06,
+      "loss": 0.6031,
+      "step": 7495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.543111831352259e-06,
+      "loss": 0.4555,
+      "step": 7496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.50368072570007e-06,
+      "loss": 0.4154,
+      "step": 7497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.464330469678762e-06,
+      "loss": 0.5438,
+      "step": 7498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.425061069774533e-06,
+      "loss": 0.0478,
+      "step": 7499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.385872532460204e-06,
+      "loss": 0.5813,
+      "step": 7500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.346764864195334e-06,
+      "loss": 0.4587,
+      "step": 7501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.307738071426153e-06,
+      "loss": 0.5677,
+      "step": 7502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.268792160585515e-06,
+      "loss": 0.4026,
+      "step": 7503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.229927138092898e-06,
+      "loss": 0.5375,
+      "step": 7504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.191143010354619e-06,
+      "loss": 0.5624,
+      "step": 7505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.15243978376351e-06,
+      "loss": 0.5353,
+      "step": 7506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.113817464699192e-06,
+      "loss": 0.5536,
+      "step": 7507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.075276059527738e-06,
+      "loss": 0.4439,
+      "step": 7508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.036815574602009e-06,
+      "loss": 0.4402,
+      "step": 7509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.998436016261657e-06,
+      "loss": 0.5674,
+      "step": 7510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.96013739083279e-06,
+      "loss": 0.4603,
+      "step": 7511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.921919704628301e-06,
+      "loss": 0.6084,
+      "step": 7512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.883782963947706e-06,
+      "loss": 0.4868,
+      "step": 7513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.845727175077146e-06,
+      "loss": 0.6306,
+      "step": 7514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.807752344289377e-06,
+      "loss": 0.4448,
+      "step": 7515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.769858477844005e-06,
+      "loss": 0.5464,
+      "step": 7516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.732045581987036e-06,
+      "loss": 0.5104,
+      "step": 7517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.69431366295137e-06,
+      "loss": 0.5391,
+      "step": 7518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.656662726956366e-06,
+      "loss": 0.4713,
+      "step": 7519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.61909278020817e-06,
+      "loss": 0.475,
+      "step": 7520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.581603828899497e-06,
+      "loss": 0.5817,
+      "step": 7521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.544195879209737e-06,
+      "loss": 0.4489,
+      "step": 7522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.506868937304901e-06,
+      "loss": 0.5476,
+      "step": 7523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.469623009337734e-06,
+      "loss": 0.5503,
+      "step": 7524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.432458101447494e-06,
+      "loss": 0.4429,
+      "step": 7525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.395374219760221e-06,
+      "loss": 0.6356,
+      "step": 7526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.358371370388473e-06,
+      "loss": 0.5798,
+      "step": 7527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.321449559431648e-06,
+      "loss": 0.58,
+      "step": 7528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.28460879297549e-06,
+      "loss": 0.5662,
+      "step": 7529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.24784907709264e-06,
+      "loss": 0.5651,
+      "step": 7530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.21117041784225e-06,
+      "loss": 0.549,
+      "step": 7531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.174572821270154e-06,
+      "loss": 0.4811,
+      "step": 7532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.138056293408858e-06,
+      "loss": 0.671,
+      "step": 7533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.10162084027738e-06,
+      "loss": 0.6262,
+      "step": 7534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.065266467881528e-06,
+      "loss": 0.5673,
+      "step": 7535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.028993182213673e-06,
+      "loss": 0.5413,
+      "step": 7536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.992800989252758e-06,
+      "loss": 0.5179,
+      "step": 7537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.956689894964508e-06,
+      "loss": 0.0479,
+      "step": 7538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.920659905301163e-06,
+      "loss": 0.6379,
+      "step": 7539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.884711026201585e-06,
+      "loss": 0.4768,
+      "step": 7540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.848843263591421e-06,
+      "loss": 0.5747,
+      "step": 7541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.81305662338272e-06,
+      "loss": 0.5114,
+      "step": 7542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.777351111474373e-06,
+      "loss": 0.4329,
+      "step": 7543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.74172673375173e-06,
+      "loss": 0.4506,
+      "step": 7544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.706183496086871e-06,
+      "loss": 0.439,
+      "step": 7545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.670721404338443e-06,
+      "loss": 0.4668,
+      "step": 7546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.63534046435177e-06,
+      "loss": 0.5427,
+      "step": 7547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.600040681958742e-06,
+      "loss": 0.5852,
+      "step": 7548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.564822062977983e-06,
+      "loss": 0.5026,
+      "step": 7549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.529684613214515e-06,
+      "loss": 0.5242,
+      "step": 7550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.494628338460319e-06,
+      "loss": 0.588,
+      "step": 7551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4596532444936586e-06,
+      "loss": 0.5677,
+      "step": 7552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4247593370795345e-06,
+      "loss": 0.6387,
+      "step": 7553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.389946621969679e-06,
+      "loss": 0.7043,
+      "step": 7554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.355215104902335e-06,
+      "loss": 0.5358,
+      "step": 7555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.320564791602313e-06,
+      "loss": 0.5316,
+      "step": 7556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.28599568778121e-06,
+      "loss": 0.394,
+      "step": 7557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.25150779913708e-06,
+      "loss": 0.5415,
+      "step": 7558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.2171011313545975e-06,
+      "loss": 0.501,
+      "step": 7559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.18277569010517e-06,
+      "loss": 0.5402,
+      "step": 7560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.148531481046661e-06,
+      "loss": 0.4275,
+      "step": 7561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.114368509823666e-06,
+      "loss": 0.4677,
+      "step": 7562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.080286782067347e-06,
+      "loss": 0.4784,
+      "step": 7563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.046286303395433e-06,
+      "loss": 0.0475,
+      "step": 7564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.012367079412385e-06,
+      "loss": 0.4868,
+      "step": 7565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.97852911570912e-06,
+      "loss": 0.6702,
+      "step": 7566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.944772417863232e-06,
+      "loss": 0.5582,
+      "step": 7567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.911096991438881e-06,
+      "loss": 0.53,
+      "step": 7568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.87750284198696e-06,
+      "loss": 0.5176,
+      "step": 7569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.843989975044762e-06,
+      "loss": 0.6164,
+      "step": 7570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.810558396136424e-06,
+      "loss": 0.4761,
+      "step": 7571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.7772081107724255e-06,
+      "loss": 0.5406,
+      "step": 7572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.743939124450038e-06,
+      "loss": 0.4941,
+      "step": 7573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.7107514426529845e-06,
+      "loss": 0.5801,
+      "step": 7574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.6776450708517785e-06,
+      "loss": 0.5544,
+      "step": 7575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.644620014503389e-06,
+      "loss": 0.551,
+      "step": 7576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.611676279051293e-06,
+      "loss": 0.4351,
+      "step": 7577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.5788138699258145e-06,
+      "loss": 0.6226,
+      "step": 7578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.5460327925437325e-06,
+      "loss": 0.499,
+      "step": 7579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.51333305230839e-06,
+      "loss": 0.5115,
+      "step": 7580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.480714654609754e-06,
+      "loss": 0.4911,
+      "step": 7581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.448177604824412e-06,
+      "loss": 0.4514,
+      "step": 7582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4157219083154635e-06,
+      "loss": 0.4855,
+      "step": 7583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.383347570432741e-06,
+      "loss": 0.5433,
+      "step": 7584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.351054596512529e-06,
+      "loss": 0.4507,
+      "step": 7585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.318842991877738e-06,
+      "loss": 0.5854,
+      "step": 7586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.286712761837954e-06,
+      "loss": 0.5999,
+      "step": 7587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.254663911689218e-06,
+      "loss": 0.5828,
+      "step": 7588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.222696446714193e-06,
+      "loss": 0.5327,
+      "step": 7589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1908103721822205e-06,
+      "loss": 0.4979,
+      "step": 7590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.159005693349095e-06,
+      "loss": 0.4409,
+      "step": 7591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.127282415457236e-06,
+      "loss": 0.4761,
+      "step": 7592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.095640543735736e-06,
+      "loss": 0.59,
+      "step": 7593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.064080083400203e-06,
+      "loss": 0.4749,
+      "step": 7594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.032601039652807e-06,
+      "loss": 0.6135,
+      "step": 7595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.001203417682177e-06,
+      "loss": 0.0478,
+      "step": 7596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.969887222663839e-06,
+      "loss": 0.5995,
+      "step": 7597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.938652459759608e-06,
+      "loss": 0.631,
+      "step": 7598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9074991341180326e-06,
+      "loss": 0.548,
+      "step": 7599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8764272508741724e-06,
+      "loss": 0.5192,
+      "step": 7600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8454368151496515e-06,
+      "loss": 0.5427,
+      "step": 7601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.814527832052719e-06,
+      "loss": 0.436,
+      "step": 7602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.783700306678186e-06,
+      "loss": 0.5526,
+      "step": 7603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.752954244107378e-06,
+      "loss": 0.6538,
+      "step": 7604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7222896494082964e-06,
+      "loss": 0.5887,
+      "step": 7605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.691706527635399e-06,
+      "loss": 0.3965,
+      "step": 7606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.661204883829762e-06,
+      "loss": 0.5897,
+      "step": 7607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.630784723019145e-06,
+      "loss": 0.5374,
+      "step": 7608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6004460502177e-06,
+      "loss": 0.6295,
+      "step": 7609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.570188870426207e-06,
+      "loss": 0.4886,
+      "step": 7610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.540013188632065e-06,
+      "loss": 0.6086,
+      "step": 7611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.509919009809128e-06,
+      "loss": 0.5153,
+      "step": 7612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4799063389179836e-06,
+      "loss": 0.4633,
+      "step": 7613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.449975180905675e-06,
+      "loss": 0.4447,
+      "step": 7614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.420125540705756e-06,
+      "loss": 0.5769,
+      "step": 7615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.390357423238457e-06,
+      "loss": 0.5375,
+      "step": 7616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.360670833410519e-06,
+      "loss": 0.6647,
+      "step": 7617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3310657761152516e-06,
+      "loss": 0.5808,
+      "step": 7618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3015422562325274e-06,
+      "loss": 0.4778,
+      "step": 7619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.272100278628788e-06,
+      "loss": 0.5687,
+      "step": 7620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.242739848156985e-06,
+      "loss": 0.5985,
+      "step": 7621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.213460969656691e-06,
+      "loss": 0.4753,
+      "step": 7622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.184263647953991e-06,
+      "loss": 0.4958,
+      "step": 7623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.155147887861589e-06,
+      "loss": 0.4933,
+      "step": 7624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.126113694178647e-06,
+      "loss": 0.4901,
+      "step": 7625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.097161071690948e-06,
+      "loss": 0.532,
+      "step": 7626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0682900251708406e-06,
+      "loss": 0.6028,
+      "step": 7627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.039500559377186e-06,
+      "loss": 0.5652,
+      "step": 7628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.010792679055465e-06,
+      "loss": 0.4747,
+      "step": 7629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.982166388937615e-06,
+      "loss": 0.4772,
+      "step": 7630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.953621693742249e-06,
+      "loss": 0.0475,
+      "step": 7631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9251585981743285e-06,
+      "loss": 0.4998,
+      "step": 7632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.896777106925543e-06,
+      "loss": 0.5056,
+      "step": 7633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.868477224674095e-06,
+      "loss": 0.5669,
+      "step": 7634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840258956084753e-06,
+      "loss": 0.5116,
+      "step": 7635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.812122305808742e-06,
+      "loss": 0.5306,
+      "step": 7636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7840672784839054e-06,
+      "loss": 0.6213,
+      "step": 7637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.756093878734657e-06,
+      "loss": 0.6122,
+      "step": 7638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.728202111171808e-06,
+      "loss": 0.5802,
+      "step": 7639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.70039198039296e-06,
+      "loss": 0.5065,
+      "step": 7640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.672663490981999e-06,
+      "loss": 0.3914,
+      "step": 7641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.645016647509603e-06,
+      "loss": 0.4086,
+      "step": 7642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.617451454532739e-06,
+      "loss": 0.6154,
+      "step": 7643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.589967916595106e-06,
+      "loss": 0.0477,
+      "step": 7644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.562566038226857e-06,
+      "loss": 0.5553,
+      "step": 7645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.535245823944767e-06,
+      "loss": 0.4767,
+      "step": 7646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.508007278252013e-06,
+      "loss": 0.4869,
+      "step": 7647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.480850405638393e-06,
+      "loss": 0.4717,
+      "step": 7648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4537752105802685e-06,
+      "loss": 0.5063,
+      "step": 7649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.426781697540461e-06,
+      "loss": 0.556,
+      "step": 7650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.399869870968465e-06,
+      "loss": 0.485,
+      "step": 7651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.373039735300122e-06,
+      "loss": 0.5737,
+      "step": 7652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3462912949579496e-06,
+      "loss": 0.5021,
+      "step": 7653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.319624554350976e-06,
+      "loss": 0.5381,
+      "step": 7654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.293039517874686e-06,
+      "loss": 0.5812,
+      "step": 7655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.26653618991113e-06,
+      "loss": 0.436,
+      "step": 7656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.240114574829035e-06,
+      "loss": 0.5581,
+      "step": 7657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.213774676983472e-06,
+      "loss": 0.5175,
+      "step": 7658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.187516500716027e-06,
+      "loss": 0.6621,
+      "step": 7659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1613400503550115e-06,
+      "loss": 0.5179,
+      "step": 7660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.135245330215087e-06,
+      "loss": 0.4761,
+      "step": 7661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.109232344597535e-06,
+      "loss": 0.501,
+      "step": 7662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.083301097790093e-06,
+      "loss": 0.5122,
+      "step": 7663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.057451594067119e-06,
+      "loss": 0.562,
+      "step": 7664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.031683837689426e-06,
+      "loss": 0.5001,
+      "step": 7665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.00599783290434e-06,
+      "loss": 0.4365,
+      "step": 7666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.980393583945807e-06,
+      "loss": 0.5281,
+      "step": 7667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954871095034229e-06,
+      "loss": 0.4429,
+      "step": 7668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.929430370376464e-06,
+      "loss": 0.5496,
+      "step": 7669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.904071414165988e-06,
+      "loss": 0.5204,
+      "step": 7670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.87879423058285e-06,
+      "loss": 0.5145,
+      "step": 7671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.853598823793436e-06,
+      "loss": 0.4926,
+      "step": 7672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.828485197950871e-06,
+      "loss": 0.4912,
+      "step": 7673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8034533571946196e-06,
+      "loss": 0.6182,
+      "step": 7674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.778503305650771e-06,
+      "loss": 0.5221,
+      "step": 7675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.753635047431869e-06,
+      "loss": 0.6266,
+      "step": 7676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7288485866370235e-06,
+      "loss": 0.4503,
+      "step": 7677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7041439273517995e-06,
+      "loss": 0.5037,
+      "step": 7678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6795210736484395e-06,
+      "loss": 0.5393,
+      "step": 7679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.654980029585475e-06,
+      "loss": 0.558,
+      "step": 7680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6305207992081147e-06,
+      "loss": 0.564,
+      "step": 7681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.606143386548022e-06,
+      "loss": 0.4969,
+      "step": 7682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5818477956233163e-06,
+      "loss": 0.0478,
+      "step": 7683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5576340304387944e-06,
+      "loss": 0.6107,
+      "step": 7684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5335020949856523e-06,
+      "loss": 0.5154,
+      "step": 7685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5094519932415414e-06,
+      "loss": 0.4509,
+      "step": 7686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4854837291707352e-06,
+      "loss": 0.4355,
+      "step": 7687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4615973067239627e-06,
+      "loss": 0.4923,
+      "step": 7688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4377927298385183e-06,
+      "loss": 0.6141,
+      "step": 7689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4140700024382077e-06,
+      "loss": 0.5288,
+      "step": 7690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3904291284331256e-06,
+      "loss": 0.5229,
+      "step": 7691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3668701117202105e-06,
+      "loss": 0.4642,
+      "step": 7692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3433929561827447e-06,
+      "loss": 0.5232,
+      "step": 7693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.319997665690411e-06,
+      "loss": 0.4523,
+      "step": 7694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.296684244099568e-06,
+      "loss": 0.6161,
+      "step": 7695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2734526952530876e-06,
+      "loss": 0.5129,
+      "step": 7696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2503030229801833e-06,
+      "loss": 0.5826,
+      "step": 7697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2272352310966924e-06,
+      "loss": 0.0477,
+      "step": 7698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.204249323404962e-06,
+      "loss": 0.4906,
+      "step": 7699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1813453036937946e-06,
+      "loss": 0.5529,
+      "step": 7700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.158523175738559e-06,
+      "loss": 0.3964,
+      "step": 7701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1357829433009688e-06,
+      "loss": 0.5775,
+      "step": 7702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1131246101294144e-06,
+      "loss": 0.4332,
+      "step": 7703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0905481799587408e-06,
+      "loss": 0.4451,
+      "step": 7704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.06805365651025e-06,
+      "loss": 0.5503,
+      "step": 7705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.045641043491698e-06,
+      "loss": 0.506,
+      "step": 7706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0233103445975184e-06,
+      "loss": 0.5131,
+      "step": 7707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.001061563508489e-06,
+      "loss": 0.6552,
+      "step": 7708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9788947038919544e-06,
+      "loss": 0.4823,
+      "step": 7709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9568097694016583e-06,
+      "loss": 0.5544,
+      "step": 7710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9348067636779664e-06,
+      "loss": 0.5624,
+      "step": 7711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.912885690347644e-06,
+      "loss": 0.5237,
+      "step": 7712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.891046553024079e-06,
+      "loss": 0.4182,
+      "step": 7713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8692893553069476e-06,
+      "loss": 0.5978,
+      "step": 7714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.847614100782603e-06,
+      "loss": 0.4791,
+      "step": 7715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8260207930238534e-06,
+      "loss": 0.4585,
+      "step": 7716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.804509435589908e-06,
+      "loss": 0.6189,
+      "step": 7717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7830800320265414e-06,
+      "loss": 0.5748,
+      "step": 7718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7617325858660947e-06,
+      "loss": 0.5308,
+      "step": 7719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.740467100627253e-06,
+      "loss": 0.5765,
+      "step": 7720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7192835798152127e-06,
+      "loss": 0.5663,
+      "step": 7721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6981820269217917e-06,
+      "loss": 0.5931,
+      "step": 7722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6771624454250966e-06,
+      "loss": 0.4829,
+      "step": 7723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.656224838789967e-06,
+      "loss": 0.5958,
+      "step": 7724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6353692104674755e-06,
+      "loss": 0.0475,
+      "step": 7725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6145955638954278e-06,
+      "loss": 0.5753,
+      "step": 7726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.593903902497863e-06,
+      "loss": 0.579,
+      "step": 7727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5732942296854967e-06,
+      "loss": 0.0477,
+      "step": 7728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.552766548855501e-06,
+      "loss": 0.661,
+      "step": 7729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5323208633914464e-06,
+      "loss": 0.584,
+      "step": 7730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.511957176663415e-06,
+      "loss": 0.4963,
+      "step": 7731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4916754920281116e-06,
+      "loss": 0.5585,
+      "step": 7732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4714758128285277e-06,
+      "loss": 0.491,
+      "step": 7733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.451358142394222e-06,
+      "loss": 0.5922,
+      "step": 7734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.431322484041265e-06,
+      "loss": 0.502,
+      "step": 7735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4113688410721255e-06,
+      "loss": 0.5589,
+      "step": 7736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3914972167758953e-06,
+      "loss": 0.5259,
+      "step": 7737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.371707614428009e-06,
+      "loss": 0.4514,
+      "step": 7738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3520000372904693e-06,
+      "loss": 0.4934,
+      "step": 7739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3323744886116773e-06,
+      "loss": 0.5502,
+      "step": 7740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.312830971626545e-06,
+      "loss": 0.746,
+      "step": 7741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.293369489556496e-06,
+      "loss": 0.6011,
+      "step": 7742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2739900456094643e-06,
+      "loss": 0.5055,
+      "step": 7743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.254692642979728e-06,
+      "loss": 0.5859,
+      "step": 7744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2354772848481864e-06,
+      "loss": 0.571,
+      "step": 7745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2163439743820846e-06,
+      "loss": 0.4457,
+      "step": 7746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.197292714735233e-06,
+      "loss": 0.5922,
+      "step": 7747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.178323509047897e-06,
+      "loss": 0.4532,
+      "step": 7748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.159436360446909e-06,
+      "loss": 0.4952,
+      "step": 7749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1406312720453347e-06,
+      "loss": 0.5062,
+      "step": 7750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1219082469428608e-06,
+      "loss": 0.5042,
+      "step": 7751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1032672882257963e-06,
+      "loss": 0.4928,
+      "step": 7752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0847083989666284e-06,
+      "loss": 0.5787,
+      "step": 7753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0662315822245203e-06,
+      "loss": 0.6583,
+      "step": 7754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047836841045092e-06,
+      "loss": 0.5721,
+      "step": 7755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.029524178460307e-06,
+      "loss": 0.4817,
+      "step": 7756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.011293597488695e-06,
+      "loss": 0.6191,
+      "step": 7757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.993145101135296e-06,
+      "loss": 0.4613,
+      "step": 7758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.975078692391552e-06,
+      "loss": 0.4095,
+      "step": 7759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9570943742353574e-06,
+      "loss": 0.4344,
+      "step": 7760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9391921496311194e-06,
+      "loss": 0.5253,
+      "step": 7761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9213720215296992e-06,
+      "loss": 0.5691,
+      "step": 7762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9036339928685253e-06,
+      "loss": 0.521,
+      "step": 7763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.885978066571259e-06,
+      "loss": 0.4917,
+      "step": 7764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.868404245548183e-06,
+      "loss": 0.514,
+      "step": 7765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8509125326960918e-06,
+      "loss": 0.5812,
+      "step": 7766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8335029308981788e-06,
+      "loss": 0.5229,
+      "step": 7767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8161754430240928e-06,
+      "loss": 0.5727,
+      "step": 7768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7989300719299939e-06,
+      "loss": 0.4387,
+      "step": 7769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7817668204583858e-06,
+      "loss": 0.4661,
+      "step": 7770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7646856914384502e-06,
+      "loss": 0.5323,
+      "step": 7771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7476866876856013e-06,
+      "loss": 0.6218,
+      "step": 7772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7307698120018755e-06,
+      "loss": 0.5381,
+      "step": 7773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7139350671757092e-06,
+      "loss": 0.4756,
+      "step": 7774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6971824559819937e-06,
+      "loss": 0.5327,
+      "step": 7775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6805119811821311e-06,
+      "loss": 0.5291,
+      "step": 7776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6639236455239792e-06,
+      "loss": 0.5518,
+      "step": 7777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6474174517417395e-06,
+      "loss": 0.4697,
+      "step": 7778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6309934025562911e-06,
+      "loss": 0.0476,
+      "step": 7779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6146515006747465e-06,
+      "loss": 0.5811,
+      "step": 7780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.598391748790784e-06,
+      "loss": 0.5057,
+      "step": 7781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5822141495845932e-06,
+      "loss": 0.6151,
+      "step": 7782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5661187057227077e-06,
+      "loss": 0.584,
+      "step": 7783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5501054198582276e-06,
+      "loss": 0.3854,
+      "step": 7784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.534174294630597e-06,
+      "loss": 0.4679,
+      "step": 7785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.518325332665882e-06,
+      "loss": 0.6237,
+      "step": 7786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5025585365763817e-06,
+      "loss": 0.4276,
+      "step": 7787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4868739089610727e-06,
+      "loss": 0.3993,
+      "step": 7788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4712714524052206e-06,
+      "loss": 0.6298,
+      "step": 7789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4557511694806568e-06,
+      "loss": 0.5386,
+      "step": 7790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4403130627455575e-06,
+      "loss": 0.5382,
+      "step": 7791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4249571347447199e-06,
+      "loss": 0.5945,
+      "step": 7792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4096833880092309e-06,
+      "loss": 0.5078,
+      "step": 7793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3944918250567428e-06,
+      "loss": 0.4732,
+      "step": 7794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3793824483912531e-06,
+      "loss": 0.6105,
+      "step": 7795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3643552605032694e-06,
+      "loss": 0.571,
+      "step": 7796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3494102638698658e-06,
+      "loss": 0.5862,
+      "step": 7797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3345474609543496e-06,
+      "loss": 0.6293,
+      "step": 7798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3197668542066499e-06,
+      "loss": 0.4729,
+      "step": 7799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3050684460630958e-06,
+      "loss": 0.5336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2904522389464157e-06,
+      "loss": 0.5599,
+      "step": 7801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.275918235265794e-06,
+      "loss": 0.4423,
+      "step": 7802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2614664374170358e-06,
+      "loss": 0.5481,
+      "step": 7803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.247096847782181e-06,
+      "loss": 0.0478,
+      "step": 7804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2328094687297787e-06,
+      "loss": 0.5897,
+      "step": 7805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2186043026148896e-06,
+      "loss": 0.4556,
+      "step": 7806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2044813517789744e-06,
+      "loss": 0.6562,
+      "step": 7807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1904406185499482e-06,
+      "loss": 0.6118,
+      "step": 7808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1764821052421827e-06,
+      "loss": 0.559,
+      "step": 7809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1626058141565033e-06,
+      "loss": 0.5691,
+      "step": 7810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1488117475801363e-06,
+      "loss": 0.5483,
+      "step": 7811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1350999077868186e-06,
+      "loss": 0.6143,
+      "step": 7812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1214702970367419e-06,
+      "loss": 0.4749,
+      "step": 7813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1079229175764428e-06,
+      "loss": 0.0477,
+      "step": 7814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0944577716389126e-06,
+      "loss": 0.4448,
+      "step": 7815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0810748614437649e-06,
+      "loss": 0.4868,
+      "step": 7816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0677741891969017e-06,
+      "loss": 0.531,
+      "step": 7817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0545557570906805e-06,
+      "loss": 0.6083,
+      "step": 7818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.041419567303914e-06,
+      "loss": 0.4865,
+      "step": 7819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.02836562200187e-06,
+      "loss": 0.6169,
+      "step": 7820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0153939233363275e-06,
+      "loss": 0.5186,
+      "step": 7821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0025044734453536e-06,
+      "loss": 0.5806,
+      "step": 7822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.896972744535826e-07,
+      "loss": 0.5525,
+      "step": 7823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.769723284721033e-07,
+      "loss": 0.6251,
+      "step": 7824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.643296375982935e-07,
+      "loss": 0.5927,
+      "step": 7825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.51769203916153e-07,
+      "loss": 0.6241,
+      "step": 7826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.392910294960811e-07,
+      "loss": 0.4334,
+      "step": 7827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.268951163947659e-07,
+      "loss": 0.4764,
+      "step": 7828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.145814666555174e-07,
+      "loss": 0.4572,
+      "step": 7829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.023500823080455e-07,
+      "loss": 0.7252,
+      "step": 7830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.902009653684595e-07,
+      "loss": 0.5682,
+      "step": 7831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.781341178393243e-07,
+      "loss": 0.5514,
+      "step": 7832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.661495417096044e-07,
+      "loss": 0.4839,
+      "step": 7833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.542472389548305e-07,
+      "loss": 0.4756,
+      "step": 7834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.424272115368226e-07,
+      "loss": 0.4519,
+      "step": 7835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.306894614039107e-07,
+      "loss": 0.6108,
+      "step": 7836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.190339904908805e-07,
+      "loss": 0.597,
+      "step": 7837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.074608007188621e-07,
+      "loss": 0.5741,
+      "step": 7838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.959698939956072e-07,
+      "loss": 0.5217,
+      "step": 7839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.845612722151007e-07,
+      "loss": 0.4463,
+      "step": 7840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.73234937257894e-07,
+      "loss": 0.6842,
+      "step": 7841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.619908909908824e-07,
+      "loss": 0.53,
+      "step": 7842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.508291352674724e-07,
+      "loss": 0.5125,
+      "step": 7843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.397496719275254e-07,
+      "loss": 0.5366,
+      "step": 7844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.287525027972475e-07,
+      "loss": 0.0475,
+      "step": 7845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.178376296893552e-07,
+      "loss": 0.4686,
+      "step": 7846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.070050544029094e-07,
+      "loss": 0.4774,
+      "step": 7847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.962547787235374e-07,
+      "loss": 0.4916,
+      "step": 7848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.855868044232105e-07,
+      "loss": 0.427,
+      "step": 7849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.750011332603556e-07,
+      "loss": 0.7803,
+      "step": 7850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.644977669797991e-07,
+      "loss": 0.5032,
+      "step": 7851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.540767073129339e-07,
+      "loss": 0.4697,
+      "step": 7852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.437379559773304e-07,
+      "loss": 0.5334,
+      "step": 7853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.334815146772921e-07,
+      "loss": 0.5,
+      "step": 7854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.233073851033555e-07,
+      "loss": 0.6113,
+      "step": 7855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.132155689325125e-07,
+      "loss": 0.4866,
+      "step": 7856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.032060678283213e-07,
+      "loss": 0.491,
+      "step": 7857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.932788834405733e-07,
+      "loss": 0.4611,
+      "step": 7858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.834340174055707e-07,
+      "loss": 0.4009,
+      "step": 7859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.736714713461821e-07,
+      "loss": 0.4484,
+      "step": 7860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.639912468714537e-07,
+      "loss": 0.4761,
+      "step": 7861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.543933455771089e-07,
+      "loss": 0.458,
+      "step": 7862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.448777690451601e-07,
+      "loss": 0.4619,
+      "step": 7863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.354445188440749e-07,
+      "loss": 0.5344,
+      "step": 7864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.260935965287761e-07,
+      "loss": 0.5562,
+      "step": 7865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.168250036405309e-07,
+      "loss": 0.0478,
+      "step": 7866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.076387417071726e-07,
+      "loss": 0.4437,
+      "step": 7867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.985348122428235e-07,
+      "loss": 0.5638,
+      "step": 7868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.895132167482274e-07,
+      "loss": 0.4995,
+      "step": 7869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.805739567103062e-07,
+      "loss": 0.4465,
+      "step": 7870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.717170336026033e-07,
+      "loss": 0.4673,
+      "step": 7871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6294244888500647e-07,
+      "loss": 0.6117,
+      "step": 7872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5425020400385874e-07,
+      "loss": 0.5306,
+      "step": 7873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4564030039195845e-07,
+      "loss": 0.5198,
+      "step": 7874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.371127394683927e-07,
+      "loss": 0.5348,
+      "step": 7875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.286675226388703e-07,
+      "loss": 0.0475,
+      "step": 7876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2030465129544447e-07,
+      "loss": 0.6111,
+      "step": 7877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.120241268165126e-07,
+      "loss": 0.4967,
+      "step": 7878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0382595056703833e-07,
+      "loss": 0.4869,
+      "step": 7879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.957101238983296e-07,
+      "loss": 0.5543,
+      "step": 7880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8767664814809424e-07,
+      "loss": 0.5054,
+      "step": 7881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7972552464055066e-07,
+      "loss": 0.0476,
+      "step": 7882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.718567546862617e-07,
+      "loss": 0.6113,
+      "step": 7883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6407033958230086e-07,
+      "loss": 0.5037,
+      "step": 7884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.563662806121415e-07,
+      "loss": 0.5294,
+      "step": 7885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.487445790456012e-07,
+      "loss": 0.4808,
+      "step": 7886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.412052361390083e-07,
+      "loss": 0.4591,
+      "step": 7887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3374825313509106e-07,
+      "loss": 0.483,
+      "step": 7888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.263736312629773e-07,
+      "loss": 0.5454,
+      "step": 7889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.190813717382501e-07,
+      "loss": 0.4285,
+      "step": 7890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1187147576294815e-07,
+      "loss": 0.5436,
+      "step": 7891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0474394452545406e-07,
+      "loss": 0.5551,
+      "step": 7892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.976987792006058e-07,
+      "loss": 0.5409,
+      "step": 7893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.907359809497523e-07,
+      "loss": 0.533,
+      "step": 7894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8385555092053096e-07,
+      "loss": 0.646,
+      "step": 7895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7705749024703464e-07,
+      "loss": 0.5385,
+      "step": 7896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7034180004981146e-07,
+      "loss": 0.4281,
+      "step": 7897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.637084814358648e-07,
+      "loss": 0.0477,
+      "step": 7898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5715753549854227e-07,
+      "loss": 0.4658,
+      "step": 7899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.506889633176468e-07,
+      "loss": 0.5637,
+      "step": 7900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.443027659594921e-07,
+      "loss": 0.5946,
+      "step": 7901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3799894447662506e-07,
+      "loss": 0.4062,
+      "step": 7902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3177749990815898e-07,
+      "loss": 0.5251,
+      "step": 7903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2563843327960688e-07,
+      "loss": 0.5648,
+      "step": 7904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1958174560282595e-07,
+      "loss": 0.4984,
+      "step": 7905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1360743787623982e-07,
+      "loss": 0.444,
+      "step": 7906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0771551108450525e-07,
+      "loss": 0.4882,
+      "step": 7907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.019059661989009e-07,
+      "loss": 0.4939,
+      "step": 7908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9617880417699408e-07,
+      "loss": 0.488,
+      "step": 7909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9053402596275193e-07,
+      "loss": 0.504,
+      "step": 7910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.849716324867079e-07,
+      "loss": 0.5244,
+      "step": 7911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7949162466568414e-07,
+      "loss": 0.4855,
+      "step": 7912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.740940034029026e-07,
+      "loss": 0.5443,
+      "step": 7913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6877876958815153e-07,
+      "loss": 0.6101,
+      "step": 7914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.635459240975079e-07,
+      "loss": 0.4833,
+      "step": 7915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.583954677935595e-07,
+      "loss": 0.4698,
+      "step": 7916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5332740152518288e-07,
+      "loss": 0.5421,
+      "step": 7917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4834172612787633e-07,
+      "loss": 0.5222,
+      "step": 7918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4343844242331594e-07,
+      "loss": 0.5052,
+      "step": 7919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.386175512197996e-07,
+      "loss": 0.6548,
+      "step": 7920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3387905331191386e-07,
+      "loss": 0.5387,
+      "step": 7921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.292229494808117e-07,
+      "loss": 0.5176,
+      "step": 7922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2464924049387928e-07,
+      "loss": 0.6306,
+      "step": 7923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2015792710501349e-07,
+      "loss": 0.5841,
+      "step": 7924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1574901005456661e-07,
+      "loss": 0.5014,
+      "step": 7925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1142249006929061e-07,
+      "loss": 0.4728,
+      "step": 7926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0717836786228174e-07,
+      "loss": 0.5928,
+      "step": 7927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0301664413309153e-07,
+      "loss": 0.501,
+      "step": 7928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.893731956778229e-08,
+      "loss": 0.5613,
+      "step": 7929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.494039483870509e-08,
+      "loss": 0.0475,
+      "step": 7930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.102587060472178e-08,
+      "loss": 0.0476,
+      "step": 7931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.719374751098296e-08,
+      "loss": 0.5878,
+      "step": 7932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.344402618926106e-08,
+      "loss": 0.7438,
+      "step": 7933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.97767072575617e-08,
+      "loss": 0.5001,
+      "step": 7934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.619179132040132e-08,
+      "loss": 0.6372,
+      "step": 7935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.268927896869615e-08,
+      "loss": 0.5803,
+      "step": 7936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.926917077976214e-08,
+      "loss": 0.6897,
+      "step": 7937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.593146731737054e-08,
+      "loss": 0.6082,
+      "step": 7938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.267616913158136e-08,
+      "loss": 0.5822,
+      "step": 7939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.9503276759076406e-08,
+      "loss": 0.4899,
+      "step": 7940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6412790722826236e-08,
+      "loss": 0.613,
+      "step": 7941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3404711532256675e-08,
+      "loss": 0.5228,
+      "step": 7942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.047903968319334e-08,
+      "loss": 0.5957,
+      "step": 7943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.763577565780608e-08,
+      "loss": 0.4762,
+      "step": 7944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.487491992483106e-08,
+      "loss": 0.5176,
+      "step": 7945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2196472939348694e-08,
+      "loss": 0.479,
+      "step": 7946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9600435142839176e-08,
+      "loss": 0.5626,
+      "step": 7947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7086806963237964e-08,
+      "loss": 0.5175,
+      "step": 7948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.465558881476927e-08,
+      "loss": 0.5612,
+      "step": 7949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2306781098279114e-08,
+      "loss": 0.5146,
+      "step": 7950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.004038420095778e-08,
+      "loss": 0.3992,
+      "step": 7951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7856398496228785e-08,
+      "loss": 0.5352,
+      "step": 7952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5754824344248472e-08,
+      "loss": 0.6089,
+      "step": 7953
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3735662091350918e-08,
+      "loss": 0.5491,
+      "step": 7954
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.179891207032547e-08,
+      "loss": 0.5558,
+      "step": 7955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9944574600472275e-08,
+      "loss": 0.4954,
+      "step": 7956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.817264998738022e-08,
+      "loss": 0.5411,
+      "step": 7957
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.648313852320449e-08,
+      "loss": 0.5078,
+      "step": 7958
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4876040486333509e-08,
+      "loss": 0.6061,
+      "step": 7959
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3351356141777515e-08,
+      "loss": 0.5858,
+      "step": 7960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.190908574077998e-08,
+      "loss": 0.5497,
+      "step": 7961
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.054922952109516e-08,
+      "loss": 0.4708,
+      "step": 7962
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.271787706877089e-09,
+      "loss": 0.5343,
+      "step": 7963
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.076760508624048e-09,
+      "loss": 0.5408,
+      "step": 7964
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.9641481234006264e-09,
+      "loss": 0.4531,
+      "step": 7965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.933950734615668e-09,
+      "loss": 0.5758,
+      "step": 7966
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.98616851196676e-09,
+      "loss": 0.5009,
+      "step": 7967
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.120801611828817e-09,
+      "loss": 0.6803,
+      "step": 7968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3378501766989645e-09,
+      "loss": 0.5134,
+      "step": 7969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6373143356961393e-09,
+      "loss": 0.567,
+      "step": 7970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0191942042835366e-09,
+      "loss": 0.5302,
+      "step": 7971
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4834898843796297e-09,
+      "loss": 0.741,
+      "step": 7972
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.030201464247149e-09,
+      "loss": 0.4818,
+      "step": 7973
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.593290186596157e-10,
+      "loss": 0.6036,
+      "step": 7974
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.708726086237846e-10,
+      "loss": 0.4231,
+      "step": 7975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6483228182373466e-10,
+      "loss": 0.5962,
+      "step": 7976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.120807217677935e-11,
+      "loss": 0.2431,
+      "step": 7977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.1664,
+      "step": 7978
+    },
+    {
+      "epoch": 1.0,
+      "step": 7978,
+      "total_flos": 2.3543594708946125e+17,
+      "train_loss": 0.34777136033087575,
+      "train_runtime": 39013.2843,
+      "train_samples_per_second": 26.174,
+      "train_steps_per_second": 0.204
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7978,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.3543594708946125e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}