diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,37528 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.319148936170213e-08,
+      "loss": 2.1896,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0638297872340426e-07,
+      "loss": 1.9309,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5957446808510638e-07,
+      "loss": 2.4176,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-07,
+      "loss": 2.1765,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6595744680851066e-07,
+      "loss": 2.3283,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1914893617021275e-07,
+      "loss": 2.4454,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.723404255319149e-07,
+      "loss": 2.0505,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 2.1727,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.787234042553192e-07,
+      "loss": 1.9568,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.319148936170213e-07,
+      "loss": 2.3516,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.851063829787235e-07,
+      "loss": 2.0263,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.382978723404255e-07,
+      "loss": 2.1272,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.914893617021278e-07,
+      "loss": 1.9594,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.446808510638298e-07,
+      "loss": 2.2739,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.97872340425532e-07,
+      "loss": 1.9003,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 2.1828,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.042553191489363e-07,
+      "loss": 2.2004,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.574468085106384e-07,
+      "loss": 1.9988,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0106382978723404e-06,
+      "loss": 2.007,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0638297872340427e-06,
+      "loss": 2.4169,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1170212765957447e-06,
+      "loss": 1.9398,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.170212765957447e-06,
+      "loss": 2.2848,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.223404255319149e-06,
+      "loss": 2.519,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.276595744680851e-06,
+      "loss": 2.3129,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3297872340425533e-06,
+      "loss": 2.3449,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3829787234042555e-06,
+      "loss": 1.5922,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4361702127659578e-06,
+      "loss": 2.3029,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4893617021276596e-06,
+      "loss": 2.1347,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5425531914893618e-06,
+      "loss": 1.697,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.595744680851064e-06,
+      "loss": 2.0585,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.648936170212766e-06,
+      "loss": 2.0525,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7021276595744682e-06,
+      "loss": 1.8226,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7553191489361704e-06,
+      "loss": 1.8556,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8085106382978727e-06,
+      "loss": 2.2701,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8617021276595745e-06,
+      "loss": 2.0765,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9148936170212767e-06,
+      "loss": 1.8789,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.968085106382979e-06,
+      "loss": 2.1437,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.021276595744681e-06,
+      "loss": 1.6856,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.074468085106383e-06,
+      "loss": 2.1305,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 1.8153,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1808510638297876e-06,
+      "loss": 1.388,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2340425531914894e-06,
+      "loss": 1.7756,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2872340425531916e-06,
+      "loss": 1.5824,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.340425531914894e-06,
+      "loss": 1.9909,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.393617021276596e-06,
+      "loss": 1.6726,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.446808510638298e-06,
+      "loss": 1.5734,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 1.7113,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 1.8244,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6063829787234047e-06,
+      "loss": 1.6331,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6595744680851065e-06,
+      "loss": 1.444,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7127659574468084e-06,
+      "loss": 1.8838,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.765957446808511e-06,
+      "loss": 1.7091,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.819148936170213e-06,
+      "loss": 1.4452,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8723404255319155e-06,
+      "loss": 1.6932,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9255319148936174e-06,
+      "loss": 1.7282,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.978723404255319e-06,
+      "loss": 1.5656,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.031914893617022e-06,
+      "loss": 1.6156,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0851063829787237e-06,
+      "loss": 1.5273,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1382978723404255e-06,
+      "loss": 1.5808,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.191489361702128e-06,
+      "loss": 1.5903,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.24468085106383e-06,
+      "loss": 1.3772,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.297872340425532e-06,
+      "loss": 1.6868,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3510638297872345e-06,
+      "loss": 1.7514,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4042553191489363e-06,
+      "loss": 1.4622,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.457446808510639e-06,
+      "loss": 1.5769,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.510638297872341e-06,
+      "loss": 1.8337,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5638297872340426e-06,
+      "loss": 1.4944,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6170212765957453e-06,
+      "loss": 1.666,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.670212765957447e-06,
+      "loss": 1.4785,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.723404255319149e-06,
+      "loss": 1.8736,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7765957446808516e-06,
+      "loss": 1.5672,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8297872340425535e-06,
+      "loss": 1.7157,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.882978723404256e-06,
+      "loss": 1.5203,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.936170212765958e-06,
+      "loss": 1.4192,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.98936170212766e-06,
+      "loss": 1.9732,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.042553191489362e-06,
+      "loss": 1.1789,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.095744680851064e-06,
+      "loss": 1.42,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.148936170212766e-06,
+      "loss": 1.3331,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.202127659574468e-06,
+      "loss": 1.3787,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 1.3971,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.308510638297873e-06,
+      "loss": 2.0409,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.361702127659575e-06,
+      "loss": 1.7326,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.414893617021277e-06,
+      "loss": 1.677,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.468085106382979e-06,
+      "loss": 1.7518,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.521276595744681e-06,
+      "loss": 1.9823,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.574468085106383e-06,
+      "loss": 1.2114,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6276595744680855e-06,
+      "loss": 1.7614,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680851063829788e-06,
+      "loss": 1.9017,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.73404255319149e-06,
+      "loss": 1.4975,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.787234042553192e-06,
+      "loss": 1.6107,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.840425531914894e-06,
+      "loss": 1.9093,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.893617021276596e-06,
+      "loss": 1.6945,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946808510638298e-06,
+      "loss": 1.628,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 1.4039,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.053191489361703e-06,
+      "loss": 1.4977,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.106382978723404e-06,
+      "loss": 1.6653,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.159574468085107e-06,
+      "loss": 1.4582,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.212765957446809e-06,
+      "loss": 1.5482,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.265957446808511e-06,
+      "loss": 1.6407,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.319148936170213e-06,
+      "loss": 1.4532,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.372340425531915e-06,
+      "loss": 1.2351,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.425531914893617e-06,
+      "loss": 1.6556,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.47872340425532e-06,
+      "loss": 1.4857,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.531914893617022e-06,
+      "loss": 1.856,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5851063829787235e-06,
+      "loss": 1.6542,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.638297872340426e-06,
+      "loss": 1.7204,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.691489361702128e-06,
+      "loss": 1.383,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.744680851063831e-06,
+      "loss": 1.5543,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7978723404255325e-06,
+      "loss": 1.4207,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.851063829787235e-06,
+      "loss": 1.721,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.904255319148937e-06,
+      "loss": 1.6115,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.957446808510638e-06,
+      "loss": 1.3576,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.010638297872341e-06,
+      "loss": 1.5515,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.063829787234044e-06,
+      "loss": 1.7913,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.117021276595745e-06,
+      "loss": 1.4406,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.170212765957447e-06,
+      "loss": 1.9162,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.22340425531915e-06,
+      "loss": 1.6749,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.276595744680851e-06,
+      "loss": 1.7574,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.329787234042554e-06,
+      "loss": 1.5126,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.382978723404256e-06,
+      "loss": 1.3786,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.436170212765958e-06,
+      "loss": 1.4143,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.48936170212766e-06,
+      "loss": 1.6035,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.542553191489362e-06,
+      "loss": 1.884,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.595744680851064e-06,
+      "loss": 1.5439,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.648936170212767e-06,
+      "loss": 1.725,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.702127659574469e-06,
+      "loss": 1.4444,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.75531914893617e-06,
+      "loss": 1.5264,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 1.5251,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.861702127659575e-06,
+      "loss": 1.4353,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.914893617021278e-06,
+      "loss": 1.8459,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.968085106382979e-06,
+      "loss": 1.6563,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.021276595744682e-06,
+      "loss": 1.6444,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.074468085106384e-06,
+      "loss": 1.4885,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.127659574468085e-06,
+      "loss": 1.804,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.1808510638297875e-06,
+      "loss": 1.5577,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.234042553191491e-06,
+      "loss": 1.3226,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.287234042553192e-06,
+      "loss": 1.6614,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.340425531914894e-06,
+      "loss": 1.3228,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.3936170212765965e-06,
+      "loss": 1.5251,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 1.4712,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.1726,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.553191489361703e-06,
+      "loss": 1.4232,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.606382978723405e-06,
+      "loss": 1.5233,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.659574468085107e-06,
+      "loss": 1.6001,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.71276595744681e-06,
+      "loss": 1.8434,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.765957446808511e-06,
+      "loss": 1.8326,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.819148936170213e-06,
+      "loss": 1.6971,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.872340425531916e-06,
+      "loss": 1.3086,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.925531914893617e-06,
+      "loss": 1.9487,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.97872340425532e-06,
+      "loss": 1.5659,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.031914893617022e-06,
+      "loss": 1.65,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.085106382978723e-06,
+      "loss": 1.4959,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.138297872340426e-06,
+      "loss": 1.5489,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.191489361702128e-06,
+      "loss": 1.4272,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.24468085106383e-06,
+      "loss": 1.6186,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.297872340425532e-06,
+      "loss": 1.3346,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.351063829787235e-06,
+      "loss": 1.8076,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.404255319148937e-06,
+      "loss": 1.6066,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.457446808510638e-06,
+      "loss": 1.2922,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 1.3977,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.563829787234044e-06,
+      "loss": 1.4688,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.617021276595746e-06,
+      "loss": 1.9206,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.670212765957447e-06,
+      "loss": 1.5086,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.72340425531915e-06,
+      "loss": 1.3743,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.776595744680852e-06,
+      "loss": 1.4272,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.829787234042555e-06,
+      "loss": 1.1906,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.882978723404256e-06,
+      "loss": 1.362,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 1.3809,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.98936170212766e-06,
+      "loss": 1.6461,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.042553191489362e-06,
+      "loss": 1.7285,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.095744680851063e-06,
+      "loss": 1.686,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.148936170212767e-06,
+      "loss": 1.3889,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.20212765957447e-06,
+      "loss": 1.5363,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.255319148936171e-06,
+      "loss": 1.5149,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.308510638297872e-06,
+      "loss": 1.5468,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.361702127659576e-06,
+      "loss": 1.4624,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.414893617021279e-06,
+      "loss": 1.7409,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.46808510638298e-06,
+      "loss": 1.3748,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.521276595744681e-06,
+      "loss": 1.3813,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.574468085106385e-06,
+      "loss": 1.7416,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.627659574468086e-06,
+      "loss": 1.4399,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.680851063829787e-06,
+      "loss": 1.4148,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.73404255319149e-06,
+      "loss": 1.6278,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.787234042553192e-06,
+      "loss": 1.8733,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.840425531914895e-06,
+      "loss": 1.3448,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.893617021276596e-06,
+      "loss": 1.7913,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.946808510638298e-06,
+      "loss": 1.9398,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.4912,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999328558964e-06,
+      "loss": 1.3713,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997314236036e-06,
+      "loss": 1.5282,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999993957031758e-06,
+      "loss": 1.2198,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999989256947029e-06,
+      "loss": 1.4694,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999983213983114e-06,
+      "loss": 1.5586,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999975828141635e-06,
+      "loss": 1.5201,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999967099424577e-06,
+      "loss": 1.5961,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999957027834282e-06,
+      "loss": 0.9732,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999945613373456e-06,
+      "loss": 1.5479,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999932856045164e-06,
+      "loss": 1.492,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999918755852834e-06,
+      "loss": 1.5921,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99990331280025e-06,
+      "loss": 1.6722,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999886526891565e-06,
+      "loss": 1.547,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999868398131282e-06,
+      "loss": 1.6617,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999848926524272e-06,
+      "loss": 1.349,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999828112075764e-06,
+      "loss": 1.3197,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999805954791351e-06,
+      "loss": 1.7618,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99978245467698e-06,
+      "loss": 1.4408,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999757611738964e-06,
+      "loss": 1.5943,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999731425983975e-06,
+      "loss": 1.7701,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999703897419048e-06,
+      "loss": 1.68,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999675026051576e-06,
+      "loss": 1.4017,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99964481188931e-06,
+      "loss": 1.5816,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999613254940368e-06,
+      "loss": 1.3998,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999580355213223e-06,
+      "loss": 1.6706,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999546112716715e-06,
+      "loss": 1.6646,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999510527460036e-06,
+      "loss": 1.4653,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999473599452746e-06,
+      "loss": 1.4256,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999435328704763e-06,
+      "loss": 1.6169,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999395715226365e-06,
+      "loss": 1.4038,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99935475902819e-06,
+      "loss": 1.3837,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999312460121242e-06,
+      "loss": 1.5356,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999268818516877e-06,
+      "loss": 1.1881,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999223834226817e-06,
+      "loss": 1.474,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999177507263146e-06,
+      "loss": 1.7325,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999129837638303e-06,
+      "loss": 1.5647,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999080825365095e-06,
+      "loss": 1.4327,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999030470456684e-06,
+      "loss": 1.9449,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998978772926592e-06,
+      "loss": 1.7652,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998925732788706e-06,
+      "loss": 1.7041,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99887135005727e-06,
+      "loss": 1.4576,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99881562474689e-06,
+      "loss": 1.4376,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998758556872534e-06,
+      "loss": 1.2531,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998700146449528e-06,
+      "loss": 1.249,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998640393493559e-06,
+      "loss": 1.3469,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998579298020676e-06,
+      "loss": 1.5237,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99851686004729e-06,
+      "loss": 1.9444,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998453079590167e-06,
+      "loss": 1.7197,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998387956666438e-06,
+      "loss": 1.6433,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998321491293592e-06,
+      "loss": 1.5725,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998253683489483e-06,
+      "loss": 1.47,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998184533272321e-06,
+      "loss": 1.6687,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998114040660678e-06,
+      "loss": 1.5919,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998042205673489e-06,
+      "loss": 1.8468,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997969028330042e-06,
+      "loss": 1.6057,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997894508649995e-06,
+      "loss": 1.5285,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997818646653362e-06,
+      "loss": 1.4983,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997741442360515e-06,
+      "loss": 1.481,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997662895792192e-06,
+      "loss": 1.3957,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99758300696949e-06,
+      "loss": 1.5755,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99750177591386e-06,
+      "loss": 1.4039,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997419202647124e-06,
+      "loss": 1.6041,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997335287191456e-06,
+      "loss": 1.2834,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997250029569395e-06,
+      "loss": 1.5501,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997163429803838e-06,
+      "loss": 1.2403,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997075487918047e-06,
+      "loss": 1.4028,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996986203935636e-06,
+      "loss": 1.4676,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99689557788059e-06,
+      "loss": 1.5747,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996803609777244e-06,
+      "loss": 1.6396,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996710299650302e-06,
+      "loss": 1.3831,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996615647524824e-06,
+      "loss": 1.3077,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996519653426229e-06,
+      "loss": 1.7116,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996422317380303e-06,
+      "loss": 1.5535,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996323639413185e-06,
+      "loss": 1.4254,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996223619551377e-06,
+      "loss": 1.2982,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996122257821746e-06,
+      "loss": 1.2358,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996019554251512e-06,
+      "loss": 1.4809,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99591550886826e-06,
+      "loss": 1.7739,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995810121699933e-06,
+      "loss": 1.3946,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995703392774836e-06,
+      "loss": 1.6772,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995595322121633e-06,
+      "loss": 1.6999,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995485909769354e-06,
+      "loss": 1.6408,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995375155747378e-06,
+      "loss": 1.6096,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995263060085456e-06,
+      "loss": 1.4973,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99514962281369e-06,
+      "loss": 1.5657,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99503484396255e-06,
+      "loss": 1.5196,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994918723562861e-06,
+      "loss": 1.099,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99480126164581e-06,
+      "loss": 1.8283,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994682458242946e-06,
+      "loss": 1.5038,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994562313386177e-06,
+      "loss": 1.4471,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99444082710777e-06,
+      "loss": 1.5965,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994317999440351e-06,
+      "loss": 1.4012,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994193830416915e-06,
+      "loss": 1.2818,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994068320070805e-06,
+      "loss": 1.5002,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99394146843573e-06,
+      "loss": 1.3926,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993813275545764e-06,
+      "loss": 1.4686,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993683741435334e-06,
+      "loss": 1.2606,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99355286613923e-06,
+      "loss": 1.8048,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9934206496926e-06,
+      "loss": 1.6504,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993287092130956e-06,
+      "loss": 1.342,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99315219349017e-06,
+      "loss": 1.328,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993015953806472e-06,
+      "loss": 1.3721,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99287837311645e-06,
+      "loss": 1.723,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992739451457058e-06,
+      "loss": 1.5773,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992599188865605e-06,
+      "loss": 1.2987,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992457585379764e-06,
+      "loss": 1.2716,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992314641037564e-06,
+      "loss": 1.4362,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992170355877398e-06,
+      "loss": 1.125,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992024729938019e-06,
+      "loss": 1.366,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991877763258538e-06,
+      "loss": 1.3694,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991729455878423e-06,
+      "loss": 1.1086,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991579807837511e-06,
+      "loss": 1.5673,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99142881917599e-06,
+      "loss": 1.5841,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991276489934416e-06,
+      "loss": 1.2815,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991122820153695e-06,
+      "loss": 1.3141,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990967809875107e-06,
+      "loss": 1.5112,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99081145914028e-06,
+      "loss": 1.3797,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990653767991203e-06,
+      "loss": 1.4235,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990494736470234e-06,
+      "loss": 1.441,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99033436462008e-06,
+      "loss": 1.5358,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990172652483818e-06,
+      "loss": 1.6389,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990009600104875e-06,
+      "loss": 1.4867,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989845207527047e-06,
+      "loss": 1.4453,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989679474794484e-06,
+      "loss": 1.4224,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9895124019517e-06,
+      "loss": 1.5219,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989343989043563e-06,
+      "loss": 1.5226,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98917423611531e-06,
+      "loss": 1.3996,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989003143212526e-06,
+      "loss": 1.7116,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988830710381168e-06,
+      "loss": 1.5962,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988656937667544e-06,
+      "loss": 1.4301,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988481825118329e-06,
+      "loss": 1.2844,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98830537278055e-06,
+      "loss": 1.2822,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988127580701601e-06,
+      "loss": 1.2941,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987948448929232e-06,
+      "loss": 1.8164,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987767977511552e-06,
+      "loss": 1.4486,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987586166497032e-06,
+      "loss": 1.7736,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987403015934504e-06,
+      "loss": 1.5821,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987218525873155e-06,
+      "loss": 1.3606,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987032696362537e-06,
+      "loss": 1.8109,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98684552745256e-06,
+      "loss": 1.3597,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986657019193492e-06,
+      "loss": 1.4029,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98646717163596e-06,
+      "loss": 1.4068,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986275984830955e-06,
+      "loss": 1.7307,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986083458829824e-06,
+      "loss": 1.7385,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985889593684277e-06,
+      "loss": 1.4739,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985694389446378e-06,
+      "loss": 1.2394,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985497846168557e-06,
+      "loss": 1.8157,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9852999639036e-06,
+      "loss": 1.7795,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985100742704655e-06,
+      "loss": 1.4454,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984900182625226e-06,
+      "loss": 1.6279,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98469828371918e-06,
+      "loss": 1.5573,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98449504604074e-06,
+      "loss": 1.4068,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984290469644495e-06,
+      "loss": 1.5415,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984084554585387e-06,
+      "loss": 1.5272,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98387730091872e-06,
+      "loss": 1.5873,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983668708700156e-06,
+      "loss": 1.4013,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983458777985722e-06,
+      "loss": 1.5021,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983247508831795e-06,
+      "loss": 1.4241,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983034901295122e-06,
+      "loss": 1.3738,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9828209554328e-06,
+      "loss": 1.4111,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982605671302293e-06,
+      "loss": 1.3151,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982389048961421e-06,
+      "loss": 1.5362,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982171088468363e-06,
+      "loss": 1.4127,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981951789881657e-06,
+      "loss": 1.411,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981731153260201e-06,
+      "loss": 1.6715,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981509178663256e-06,
+      "loss": 1.4303,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981285866150436e-06,
+      "loss": 1.3457,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98106121578172e-06,
+      "loss": 1.4499,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980835227617443e-06,
+      "loss": 1.4801,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980607901718297e-06,
+      "loss": 1.6223,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980379238145342e-06,
+      "loss": 1.5137,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980149236959986e-06,
+      "loss": 1.6528,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979917898224006e-06,
+      "loss": 1.4088,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979685221999532e-06,
+      "loss": 1.7329,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979451208349055e-06,
+      "loss": 1.4511,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97921585733543e-06,
+      "loss": 1.572,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97897916902186e-06,
+      "loss": 1.5003,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97874114347192e-06,
+      "loss": 1.6044,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978501780749531e-06,
+      "loss": 1.3401,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978261080918988e-06,
+      "loss": 1.2613,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978019044044933e-06,
+      "loss": 1.3753,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977775670192373e-06,
+      "loss": 1.2039,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977530959426668e-06,
+      "loss": 1.5257,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977284911813549e-06,
+      "loss": 1.608,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977037527419094e-06,
+      "loss": 1.5291,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976788806309742e-06,
+      "loss": 1.1668,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976538748552302e-06,
+      "loss": 1.3103,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976287354213924e-06,
+      "loss": 1.4686,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976034623362134e-06,
+      "loss": 1.4348,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975780556064806e-06,
+      "loss": 1.5085,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975525152390178e-06,
+      "loss": 1.6828,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975268412406842e-06,
+      "loss": 1.1885,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975010336183757e-06,
+      "loss": 1.1915,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974750923790234e-06,
+      "loss": 1.3604,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974490175295945e-06,
+      "loss": 1.4678,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97422809077092e-06,
+      "loss": 1.3739,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97396467028555e-06,
+      "loss": 1.3973,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973699913910584e-06,
+      "loss": 1.3801,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973433821717128e-06,
+      "loss": 1.295,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97316639377665e-06,
+      "loss": 1.6365,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972897630160972e-06,
+      "loss": 1.4981,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97262753094228e-06,
+      "loss": 1.5842,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972356096193115e-06,
+      "loss": 1.408,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972083325986377e-06,
+      "loss": 1.6998,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971809220395328e-06,
+      "loss": 1.4993,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971533779493586e-06,
+      "loss": 1.679,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971257003355128e-06,
+      "loss": 1.6154,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970978892054286e-06,
+      "loss": 1.4443,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970699445665758e-06,
+      "loss": 1.4167,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970418664264596e-06,
+      "loss": 1.2477,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97013654792621e-06,
+      "loss": 1.3197,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969853096726372e-06,
+      "loss": 1.5197,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969568310741209e-06,
+      "loss": 1.7977,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969282190047207e-06,
+      "loss": 1.2499,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96899473472121e-06,
+      "loss": 1.3344,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968705944840428e-06,
+      "loss": 1.4901,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968415820482415e-06,
+      "loss": 1.3438,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968124361725098e-06,
+      "loss": 1.3519,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967831568646753e-06,
+      "loss": 1.2019,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967537441326018e-06,
+      "loss": 1.4332,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967241979841886e-06,
+      "loss": 1.3151,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966945184273716e-06,
+      "loss": 1.2625,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966647054701217e-06,
+      "loss": 1.129,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966347591204459e-06,
+      "loss": 1.528,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966046793863873e-06,
+      "loss": 1.0014,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965744662760246e-06,
+      "loss": 1.3746,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96544119797472e-06,
+      "loss": 1.1214,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965136399588803e-06,
+      "loss": 1.3894,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964830267684355e-06,
+      "loss": 1.2926,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964522802343593e-06,
+      "loss": 1.6624,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9642140036491e-06,
+      "loss": 1.9797,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963903871683806e-06,
+      "loss": 1.1235,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96359240653101e-06,
+      "loss": 1.5704,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963279608274364e-06,
+      "loss": 1.4426,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962965476997877e-06,
+      "loss": 1.2086,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962650012785917e-06,
+      "loss": 1.4125,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962333215723208e-06,
+      "loss": 1.4297,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962015085894838e-06,
+      "loss": 1.4376,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96169562338625e-06,
+      "loss": 1.549,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961374828283239e-06,
+      "loss": 1.4556,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961052700671967e-06,
+      "loss": 1.2842,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960729240638947e-06,
+      "loss": 1.1429,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960404448271057e-06,
+      "loss": 1.2951,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960078323655524e-06,
+      "loss": 1.3032,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959750866879941e-06,
+      "loss": 1.4037,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959422078032253e-06,
+      "loss": 1.487,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959091957200766e-06,
+      "loss": 1.4816,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958760504474144e-06,
+      "loss": 1.3341,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958427719941401e-06,
+      "loss": 1.3862,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958093603691923e-06,
+      "loss": 1.2039,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957758155815443e-06,
+      "loss": 1.4852,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957421376402053e-06,
+      "loss": 1.5191,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957083265542202e-06,
+      "loss": 1.5045,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956743823326704e-06,
+      "loss": 1.5927,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956403049846723e-06,
+      "loss": 1.6955,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956060945193781e-06,
+      "loss": 1.2474,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955717509459762e-06,
+      "loss": 1.3685,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955372742736903e-06,
+      "loss": 1.6259,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9550266451178e-06,
+      "loss": 1.853,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954679216695406e-06,
+      "loss": 1.1664,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954330457563035e-06,
+      "loss": 1.1887,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953980367814354e-06,
+      "loss": 1.273,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953628947543387e-06,
+      "loss": 1.608,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953276196844519e-06,
+      "loss": 1.5946,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952922115812492e-06,
+      "loss": 1.4631,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9525667045424e-06,
+      "loss": 1.4968,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952209963129701e-06,
+      "loss": 1.5125,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951851891670206e-06,
+      "loss": 1.5501,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951492490260086e-06,
+      "loss": 1.1984,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951131758995866e-06,
+      "loss": 1.3524,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950769697974432e-06,
+      "loss": 1.3307,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950406307293023e-06,
+      "loss": 1.4976,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950041587049237e-06,
+      "loss": 1.1579,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949675537341031e-06,
+      "loss": 1.6808,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949308158266717e-06,
+      "loss": 1.4666,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948939449924964e-06,
+      "loss": 1.0555,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948569412414797e-06,
+      "loss": 1.2992,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948198045835601e-06,
+      "loss": 1.8393,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947825350287118e-06,
+      "loss": 1.3852,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94745132586944e-06,
+      "loss": 1.4272,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947075972683024e-06,
+      "loss": 1.331,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946699290828683e-06,
+      "loss": 1.4121,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946321280407583e-06,
+      "loss": 1.3512,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94594194152125e-06,
+      "loss": 1.3191,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94556127427156e-06,
+      "loss": 1.4252,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945179278760759e-06,
+      "loss": 1.5408,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944795955091439e-06,
+      "loss": 1.6082,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94441130336655e-06,
+      "loss": 1.3889,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9440253236894e-06,
+      "loss": 1.1079,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943638016163658e-06,
+      "loss": 1.454,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943249380893343e-06,
+      "loss": 1.2481,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942859417982833e-06,
+      "loss": 1.6574,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942468127536864e-06,
+      "loss": 1.5899,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942075509660527e-06,
+      "loss": 1.2073,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94168156445927e-06,
+      "loss": 1.4613,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941286292038894e-06,
+      "loss": 1.4469,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940889692505565e-06,
+      "loss": 1.4774,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940491765965798e-06,
+      "loss": 1.2944,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940092512526468e-06,
+      "loss": 1.1466,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939691932294804e-06,
+      "loss": 1.4165,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93929002537839e-06,
+      "loss": 1.4424,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938886791885172e-06,
+      "loss": 1.293,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938482231923449e-06,
+      "loss": 1.399,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938076345601875e-06,
+      "loss": 1.5376,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937669133029462e-06,
+      "loss": 1.6729,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937260594315578e-06,
+      "loss": 1.4765,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936850729569945e-06,
+      "loss": 1.2098,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936439538902644e-06,
+      "loss": 1.2435,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936027022424113e-06,
+      "loss": 1.5674,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935613180245143e-06,
+      "loss": 1.4736,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93519801247688e-06,
+      "loss": 1.5937,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934781519230832e-06,
+      "loss": 1.3407,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934363700618857e-06,
+      "loss": 1.3736,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933944556753173e-06,
+      "loss": 1.3176,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933524087746348e-06,
+      "loss": 1.2956,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933102293711314e-06,
+      "loss": 1.3339,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932679174761355e-06,
+      "loss": 1.6091,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932254731010108e-06,
+      "loss": 1.0927,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931828962571572e-06,
+      "loss": 1.5565,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931401869560096e-06,
+      "loss": 1.097,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930973452090386e-06,
+      "loss": 1.2162,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93054371027751e-06,
+      "loss": 1.201,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93011264423688e-06,
+      "loss": 1.3439,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929680254084273e-06,
+      "loss": 1.3727,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92924653993582e-06,
+      "loss": 1.2471,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928811501908006e-06,
+      "loss": 1.536,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92837514011767e-06,
+      "loss": 1.663,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92793745468201e-06,
+      "loss": 1.3617,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92749844571858e-06,
+      "loss": 1.2323,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927058113345282e-06,
+      "loss": 1.4637,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926616457680385e-06,
+      "loss": 1.6275,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926173478842502e-06,
+      "loss": 1.4189,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925729176950612e-06,
+      "loss": 1.199,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925283552124039e-06,
+      "loss": 1.7542,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92483660448247e-06,
+      "loss": 1.4686,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924388334145943e-06,
+      "loss": 1.2127,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923938741234855e-06,
+      "loss": 1.3057,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923487825869955e-06,
+      "loss": 1.3092,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923035588172347e-06,
+      "loss": 1.3102,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922582028263495e-06,
+      "loss": 1.4788,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922127146265209e-06,
+      "loss": 1.5596,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921670942299664e-06,
+      "loss": 1.3318,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921213416489383e-06,
+      "loss": 1.3529,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92075456895725e-06,
+      "loss": 1.3048,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920294399826498e-06,
+      "loss": 1.2903,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919832909220717e-06,
+      "loss": 1.5349,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919370097263855e-06,
+      "loss": 1.3725,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91890596408021e-06,
+      "loss": 1.5024,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918440509794437e-06,
+      "loss": 1.5802,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917973734531549e-06,
+      "loss": 1.7153,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91750563841691e-06,
+      "loss": 1.2669,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917036221576235e-06,
+      "loss": 1.4627,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916565484135605e-06,
+      "loss": 1.2337,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916093426221445e-06,
+      "loss": 1.0792,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915620047960539e-06,
+      "loss": 1.2695,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915145349480027e-06,
+      "loss": 1.6202,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914669330907399e-06,
+      "loss": 1.4767,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914191992370504e-06,
+      "loss": 1.3045,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913713333997546e-06,
+      "loss": 1.3801,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913233355917075e-06,
+      "loss": 1.1416,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91275205825801e-06,
+      "loss": 1.3209,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91226944114961e-06,
+      "loss": 1.4139,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911785504721498e-06,
+      "loss": 1.1667,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911300249103646e-06,
+      "loss": 1.3982,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910813674426381e-06,
+      "loss": 1.358,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910325780820391e-06,
+      "loss": 1.3127,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909836568416706e-06,
+      "loss": 1.4266,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90934603734672e-06,
+      "loss": 1.7113,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90885418774218e-06,
+      "loss": 1.4992,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908361019735181e-06,
+      "loss": 1.5164,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90786653345818e-06,
+      "loss": 1.411,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907370729043984e-06,
+      "loss": 1.3032,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906873606625752e-06,
+      "loss": 1.3416,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906375166336998e-06,
+      "loss": 1.5196,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905875408311596e-06,
+      "loss": 1.5682,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905374332683768e-06,
+      "loss": 1.6055,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90487193958809e-06,
+      "loss": 1.5216,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904368229159494e-06,
+      "loss": 1.3627,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90386320153326e-06,
+      "loss": 1.4637,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903356856845035e-06,
+      "loss": 1.6588,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902849195230803e-06,
+      "loss": 1.533,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902340216826915e-06,
+      "loss": 1.3449,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90182992177007e-06,
+      "loss": 1.1256,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90131831019732e-06,
+      "loss": 1.2258,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900805382246071e-06,
+      "loss": 1.5867,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900291138054086e-06,
+      "loss": 1.2774,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899775577759475e-06,
+      "loss": 1.3394,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899258701500712e-06,
+      "loss": 1.3614,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89874050941661e-06,
+      "loss": 1.3137,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89822100164635e-06,
+      "loss": 1.2486,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897700178329453e-06,
+      "loss": 1.3825,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897178039605803e-06,
+      "loss": 1.4924,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896654585615634e-06,
+      "loss": 1.428,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896129816499535e-06,
+      "loss": 1.1454,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895603732398446e-06,
+      "loss": 1.2483,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89507633345366e-06,
+      "loss": 1.4751,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894547619806822e-06,
+      "loss": 1.3512,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894017591599934e-06,
+      "loss": 1.4646,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89348624897535e-06,
+      "loss": 1.4405,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892953592075776e-06,
+      "loss": 1.5638,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89241962104427e-06,
+      "loss": 1.4103,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891884336024242e-06,
+      "loss": 1.2002,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89134773715946e-06,
+      "loss": 1.6254,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890809824594041e-06,
+      "loss": 1.5474,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890270598472457e-06,
+      "loss": 1.5207,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889730058939529e-06,
+      "loss": 1.5346,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889188206140434e-06,
+      "loss": 1.6959,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8886450402207e-06,
+      "loss": 1.2914,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888100561326211e-06,
+      "loss": 1.1877,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8875547696032e-06,
+      "loss": 1.1379,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887007665198255e-06,
+      "loss": 1.2095,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88645924825831e-06,
+      "loss": 1.5232,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885909518930664e-06,
+      "loss": 1.1347,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885358477362956e-06,
+      "loss": 1.3899,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884806123703186e-06,
+      "loss": 1.3147,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8842524580997e-06,
+      "loss": 1.5167,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883697480701204e-06,
+      "loss": 1.56,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883141191656748e-06,
+      "loss": 1.1152,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882583591115741e-06,
+      "loss": 1.3236,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88202467922794e-06,
+      "loss": 1.342,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881464456143454e-06,
+      "loss": 1.1526,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880902922012747e-06,
+      "loss": 1.5005,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880340076986633e-06,
+      "loss": 1.1779,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879775921216284e-06,
+      "loss": 0.9276,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87921045485321e-06,
+      "loss": 1.4626,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87864367804929e-06,
+      "loss": 1.6473,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878075590956742e-06,
+      "loss": 1.4326,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877506193728144e-06,
+      "loss": 1.5053,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876935486516419e-06,
+      "loss": 1.2745,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876363469474848e-06,
+      "loss": 1.3828,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875790142757062e-06,
+      "loss": 1.2108,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87521550651704e-06,
+      "loss": 1.4388,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874639560909118e-06,
+      "loss": 1.4144,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874062306087983e-06,
+      "loss": 1.4929,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873483742208668e-06,
+      "loss": 1.1709,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872903869426564e-06,
+      "loss": 1.3737,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87232268789741e-06,
+      "loss": 1.4857,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8717401977773e-06,
+      "loss": 1.2631,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871156399222677e-06,
+      "loss": 1.3772,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870571292390331e-06,
+      "loss": 1.5194,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869984877437413e-06,
+      "loss": 1.536,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869397154521418e-06,
+      "loss": 1.2307,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868808123800194e-06,
+      "loss": 1.2386,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868217785431942e-06,
+      "loss": 1.1981,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86762613957521e-06,
+      "loss": 1.5868,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867033186388906e-06,
+      "loss": 1.6891,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866438926032278e-06,
+      "loss": 1.2959,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865843358664933e-06,
+      "loss": 1.3535,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865246484446824e-06,
+      "loss": 1.5338,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86464830353826e-06,
+      "loss": 1.2346,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864048816099897e-06,
+      "loss": 1.3101,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863448022292742e-06,
+      "loss": 1.3471,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862845922278156e-06,
+      "loss": 1.1808,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86224251621785e-06,
+      "loss": 1.3104,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861637804273882e-06,
+      "loss": 1.655,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861031786608663e-06,
+      "loss": 1.3851,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860424463384956e-06,
+      "loss": 1.4014,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859815834765875e-06,
+      "loss": 1.3629,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859205900914884e-06,
+      "loss": 1.5925,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858594661995792e-06,
+      "loss": 1.3427,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857982118172768e-06,
+      "loss": 1.115,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857368269610325e-06,
+      "loss": 1.3676,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856753116473327e-06,
+      "loss": 1.1032,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856136658926993e-06,
+      "loss": 1.0848,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855518897136886e-06,
+      "loss": 1.2614,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854899831268926e-06,
+      "loss": 1.0964,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854279461489373e-06,
+      "loss": 1.4199,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85365778796485e-06,
+      "loss": 1.6904,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853034810862321e-06,
+      "loss": 1.0996,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852410530349102e-06,
+      "loss": 1.2217,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851784946592862e-06,
+      "loss": 1.5686,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851158059761617e-06,
+      "loss": 1.4227,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850529870023735e-06,
+      "loss": 1.3543,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849900377547933e-06,
+      "loss": 1.5282,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849269582503276e-06,
+      "loss": 1.2078,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848637485059183e-06,
+      "loss": 1.5535,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84800408538542e-06,
+      "loss": 1.3411,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8473693836521e-06,
+      "loss": 1.4368,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846733380029692e-06,
+      "loss": 1.3839,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846096074689012e-06,
+      "loss": 1.2865,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845457467801223e-06,
+      "loss": 1.3105,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844817559537841e-06,
+      "loss": 1.2671,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84417635007073e-06,
+      "loss": 1.1474,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843533839572105e-06,
+      "loss": 1.2852,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84289002821453e-06,
+      "loss": 1.513,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842244916170913e-06,
+      "loss": 1.4728,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841598503614518e-06,
+      "loss": 1.2797,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840950790718959e-06,
+      "loss": 1.3388,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840301777658192e-06,
+      "loss": 1.4121,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83965146460653e-06,
+      "loss": 1.441,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838999851738631e-06,
+      "loss": 1.2339,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838346939229501e-06,
+      "loss": 1.7537,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837692727254498e-06,
+      "loss": 0.8909,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83703721598933e-06,
+      "loss": 1.3381,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836380405610048e-06,
+      "loss": 1.4357,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835722296293058e-06,
+      "loss": 0.8773,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835062888215114e-06,
+      "loss": 1.394,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834402181553314e-06,
+      "loss": 1.5966,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833740176485111e-06,
+      "loss": 1.132,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833076873188303e-06,
+      "loss": 1.2458,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832412271841037e-06,
+      "loss": 1.3804,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831746372621811e-06,
+      "loss": 1.1943,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831079175709467e-06,
+      "loss": 1.2403,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830410681283203e-06,
+      "loss": 1.6172,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829740889522555e-06,
+      "loss": 1.4651,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829069800607418e-06,
+      "loss": 1.4585,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828397414718026e-06,
+      "loss": 1.2079,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827723732034972e-06,
+      "loss": 1.1462,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827048752739189e-06,
+      "loss": 1.5896,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826372477011956e-06,
+      "loss": 1.6571,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82569490503491e-06,
+      "loss": 1.1476,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825016036990029e-06,
+      "loss": 1.2906,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824335873059643e-06,
+      "loss": 1.0769,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823654413426424e-06,
+      "loss": 1.1892,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8229716582734e-06,
+      "loss": 1.0307,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822287607783938e-06,
+      "loss": 1.2409,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821602262141762e-06,
+      "loss": 1.5215,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820915621530939e-06,
+      "loss": 1.4316,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820227686135885e-06,
+      "loss": 1.4289,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81953845614136e-06,
+      "loss": 1.4269,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81884793173248e-06,
+      "loss": 1.3163,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818156113094699e-06,
+      "loss": 1.4602,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817463000413826e-06,
+      "loss": 1.5109,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816768593876012e-06,
+      "loss": 1.1951,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81607289366776e-06,
+      "loss": 1.3978,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81537589997592e-06,
+      "loss": 1.3826,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814677612987684e-06,
+      "loss": 1.4101,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8139780328906e-06,
+      "loss": 1.1738,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813277159872556e-06,
+      "loss": 1.4629,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812574994121791e-06,
+      "loss": 1.0788,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81187153582689e-06,
+      "loss": 1.3241,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811166785176785e-06,
+      "loss": 1.2576,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810460742360755e-06,
+      "loss": 1.6996,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809753407568427e-06,
+      "loss": 1.0662,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809044780989776e-06,
+      "loss": 1.2694,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80833486281512e-06,
+      "loss": 1.493,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807623653235126e-06,
+      "loss": 1.4193,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80691115244081e-06,
+      "loss": 1.3991,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806197360623531e-06,
+      "loss": 1.6239,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805482277974999e-06,
+      "loss": 1.4553,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804765904687269e-06,
+      "loss": 1.2632,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804048240952736e-06,
+      "loss": 1.4599,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803329286964153e-06,
+      "loss": 1.379,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802609042914614e-06,
+      "loss": 1.3685,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801887508997558e-06,
+      "loss": 1.4263,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80116468540677e-06,
+      "loss": 1.1986,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800440572336388e-06,
+      "loss": 1.1928,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79971516998089e-06,
+      "loss": 1.1758,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798988478535098e-06,
+      "loss": 1.4501,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79826049819419e-06,
+      "loss": 1.3393,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797531229153682e-06,
+      "loss": 1.2705,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796800671609436e-06,
+      "loss": 1.1947,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796068825757667e-06,
+      "loss": 1.1858,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795335691794929e-06,
+      "loss": 1.542,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794601269918125e-06,
+      "loss": 1.4903,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793865560324503e-06,
+      "loss": 1.0759,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793128563211656e-06,
+      "loss": 1.2909,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792390278777527e-06,
+      "loss": 1.1684,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7916507072204e-06,
+      "loss": 1.3953,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790909848738907e-06,
+      "loss": 1.1058,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790167703532025e-06,
+      "loss": 1.1717,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789424271799075e-06,
+      "loss": 1.3171,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788679553739728e-06,
+      "loss": 1.3132,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787933549553996e-06,
+      "loss": 1.3098,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787186259442241e-06,
+      "loss": 1.2877,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786437683605161e-06,
+      "loss": 1.3015,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785687822243813e-06,
+      "loss": 1.4078,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78493667555959e-06,
+      "loss": 1.3331,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784184243754229e-06,
+      "loss": 1.4328,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783430527029818e-06,
+      "loss": 1.302,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782675525588787e-06,
+      "loss": 1.4115,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781919239633912e-06,
+      "loss": 1.4842,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781161669368316e-06,
+      "loss": 0.9139,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780402814995458e-06,
+      "loss": 1.6129,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779642676719155e-06,
+      "loss": 1.1472,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77888125474356e-06,
+      "loss": 1.1613,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77811854927317e-06,
+      "loss": 1.4811,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777354560512835e-06,
+      "loss": 1.1954,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77658928866774e-06,
+      "loss": 1.2636,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77582273394342e-06,
+      "loss": 1.2436,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775054896545755e-06,
+      "loss": 1.1796,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774285776680967e-06,
+      "loss": 1.2955,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773515374555623e-06,
+      "loss": 1.5743,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772743690376636e-06,
+      "loss": 1.4254,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77197072435126e-06,
+      "loss": 1.5106,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7711964766871e-06,
+      "loss": 1.2803,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770420947592096e-06,
+      "loss": 1.3417,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76964413727454e-06,
+      "loss": 1.2892,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76886604594306e-06,
+      "loss": 1.2073,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768086673806638e-06,
+      "loss": 1.4911,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767306021074593e-06,
+      "loss": 1.3457,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766524087956592e-06,
+      "loss": 1.1539,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765740874662642e-06,
+      "loss": 1.5323,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764956381403095e-06,
+      "loss": 1.2905,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764170608388648e-06,
+      "loss": 1.3764,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76338355583034e-06,
+      "loss": 1.1693,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762595223939557e-06,
+      "loss": 1.2442,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761805612928025e-06,
+      "loss": 1.0826,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761014723007815e-06,
+      "loss": 1.1723,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760222554391343e-06,
+      "loss": 1.3977,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759429107291366e-06,
+      "loss": 1.4179,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758634381920982e-06,
+      "loss": 1.2587,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757838378493637e-06,
+      "loss": 1.4001,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757041097223123e-06,
+      "loss": 1.0965,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756242538323566e-06,
+      "loss": 1.1987,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755442702009443e-06,
+      "loss": 1.4834,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75464158849557e-06,
+      "loss": 1.3645,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753839197997105e-06,
+      "loss": 1.4365,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753035530729556e-06,
+      "loss": 1.3258,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752230586908767e-06,
+      "loss": 1.1162,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751424366750923e-06,
+      "loss": 0.9814,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75061687047256e-06,
+      "loss": 0.9906,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749808098290553e-06,
+      "loss": 1.1008,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748998050422117e-06,
+      "loss": 1.2073,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748186727084812e-06,
+      "loss": 1.4005,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747374128496541e-06,
+      "loss": 1.1883,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746560254875547e-06,
+      "loss": 1.2689,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745745106440422e-06,
+      "loss": 0.9337,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74492868341009e-06,
+      "loss": 1.425,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744110986003826e-06,
+      "loss": 1.3157,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743292014441243e-06,
+      "loss": 1.56,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742471768942299e-06,
+      "loss": 1.2055,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741650249727292e-06,
+      "loss": 1.2568,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740827457016863e-06,
+      "loss": 1.4929,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740003391031992e-06,
+      "loss": 1.4963,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739178051994008e-06,
+      "loss": 1.5084,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738351440124573e-06,
+      "loss": 1.3077,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7375235556457e-06,
+      "loss": 1.3051,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736694398779735e-06,
+      "loss": 1.4365,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735863969749373e-06,
+      "loss": 1.5571,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735032268777647e-06,
+      "loss": 1.2147,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734199296087932e-06,
+      "loss": 1.3653,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733365051903944e-06,
+      "loss": 1.2961,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732529536449741e-06,
+      "loss": 1.6206,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731692749949726e-06,
+      "loss": 1.3374,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730854692628637e-06,
+      "loss": 1.3286,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730015364711558e-06,
+      "loss": 1.5754,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729174766423912e-06,
+      "loss": 1.5181,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728332897991463e-06,
+      "loss": 1.2298,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72748975964032e-06,
+      "loss": 1.4478,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726645351596926e-06,
+      "loss": 1.4133,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725799674088072e-06,
+      "loss": 1.2026,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724952727340886e-06,
+      "loss": 1.0579,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724104511582838e-06,
+      "loss": 1.376,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72325502704174e-06,
+      "loss": 1.2047,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72240427394574e-06,
+      "loss": 1.1017,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721552252523334e-06,
+      "loss": 1.1444,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720698963003351e-06,
+      "loss": 1.3332,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719844405614971e-06,
+      "loss": 0.9747,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7189885805877e-06,
+      "loss": 1.1229,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718131488151399e-06,
+      "loss": 1.3769,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717273128536259e-06,
+      "loss": 1.3591,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716413501972815e-06,
+      "loss": 1.5749,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715552608691944e-06,
+      "loss": 1.6509,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714690448924863e-06,
+      "loss": 1.1343,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713827022903124e-06,
+      "loss": 1.3183,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712962330858626e-06,
+      "loss": 1.2631,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712096373023603e-06,
+      "loss": 1.2628,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711229149630634e-06,
+      "loss": 1.0759,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710360660912629e-06,
+      "loss": 1.1151,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709490907102849e-06,
+      "loss": 1.1915,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708619888434887e-06,
+      "loss": 1.1002,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707747605142676e-06,
+      "loss": 1.3675,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706874057460497e-06,
+      "loss": 1.237,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705999245622958e-06,
+      "loss": 1.3548,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705123169865016e-06,
+      "loss": 1.196,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704245830421963e-06,
+      "loss": 1.0277,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703367227529432e-06,
+      "loss": 1.2188,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702487361423395e-06,
+      "loss": 1.3898,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701606232340165e-06,
+      "loss": 1.2838,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70072384051639e-06,
+      "loss": 1.6066,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699840186189061e-06,
+      "loss": 1.2274,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698955269595505e-06,
+      "loss": 1.1908,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698069090973391e-06,
+      "loss": 1.5003,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697181650560727e-06,
+      "loss": 1.3296,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696292948595857e-06,
+      "loss": 1.2583,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695402985317467e-06,
+      "loss": 1.7118,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694511760964578e-06,
+      "loss": 1.2411,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693619275776552e-06,
+      "loss": 1.2944,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69272552999309e-06,
+      "loss": 0.9773,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691830523854231e-06,
+      "loss": 1.3152,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690934257600353e-06,
+      "loss": 1.5913,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690036731472172e-06,
+      "loss": 1.2352,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689137945710742e-06,
+      "loss": 1.2238,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688237900557455e-06,
+      "loss": 1.4078,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687336596254045e-06,
+      "loss": 1.1235,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686434033042575e-06,
+      "loss": 1.4056,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685530211165459e-06,
+      "loss": 1.3575,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684625130865437e-06,
+      "loss": 1.3987,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683718792385595e-06,
+      "loss": 1.2309,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682811195969355e-06,
+      "loss": 1.3685,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681902341860471e-06,
+      "loss": 1.4436,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680992230303046e-06,
+      "loss": 1.121,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680080861541511e-06,
+      "loss": 1.5306,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67916823582064e-06,
+      "loss": 1.515,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678254353385538e-06,
+      "loss": 1.6698,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677339214481661e-06,
+      "loss": 1.6648,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676422819354785e-06,
+      "loss": 1.2196,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675505168251038e-06,
+      "loss": 1.4809,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674586261416874e-06,
+      "loss": 1.2851,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673666099099095e-06,
+      "loss": 1.5359,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672744681544834e-06,
+      "loss": 1.5104,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67182200900156e-06,
+      "loss": 1.2787,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670898081717079e-06,
+      "loss": 1.464,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669972899939542e-06,
+      "loss": 1.2509,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669046463917427e-06,
+      "loss": 1.33,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668118773899554e-06,
+      "loss": 1.1816,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667189830135078e-06,
+      "loss": 1.2856,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666259632873494e-06,
+      "loss": 1.2222,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665328182364627e-06,
+      "loss": 1.0874,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664395478858645e-06,
+      "loss": 1.2952,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663461522606049e-06,
+      "loss": 1.1251,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662526313857678e-06,
+      "loss": 1.2549,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66158985286471e-06,
+      "loss": 1.2292,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66065213987865e-06,
+      "loss": 1.1644,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659713175151352e-06,
+      "loss": 1.0509,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658772958934997e-06,
+      "loss": 1.3394,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657831491482103e-06,
+      "loss": 1.3643,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65688877304553e-06,
+      "loss": 1.1758,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655944803878467e-06,
+      "loss": 1.3101,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654999584234444e-06,
+      "loss": 1.4593,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654053114367321e-06,
+      "loss": 1.0217,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653105394531302e-06,
+      "loss": 1.4107,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65215642498092e-06,
+      "loss": 1.49,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651206205971045e-06,
+      "loss": 1.0879,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650254737756883e-06,
+      "loss": 1.6694,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64930202059398e-06,
+      "loss": 1.1877,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648348054738208e-06,
+      "loss": 1.4756,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647392840445784e-06,
+      "loss": 1.205,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646436377973253e-06,
+      "loss": 1.1989,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645478667577502e-06,
+      "loss": 1.3692,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644519709515746e-06,
+      "loss": 1.3388,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64355950404554e-06,
+      "loss": 1.1513,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642598051424772e-06,
+      "loss": 1.1743,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641635351911664e-06,
+      "loss": 1.1209,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640671405764777e-06,
+      "loss": 1.3222,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639706213243004e-06,
+      "loss": 1.6152,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638739774605572e-06,
+      "loss": 1.5928,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637772090112043e-06,
+      "loss": 1.565,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636803160022314e-06,
+      "loss": 1.1641,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63583298459662e-06,
+      "loss": 1.153,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634861564095525e-06,
+      "loss": 1.5205,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633888898779928e-06,
+      "loss": 1.0695,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632914988911066e-06,
+      "loss": 1.5127,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631939834750509e-06,
+      "loss": 1.3462,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63096343656016e-06,
+      "loss": 1.2084,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629985794602253e-06,
+      "loss": 1.2783,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629006909139363e-06,
+      "loss": 1.4602,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628026780434397e-06,
+      "loss": 1.093,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62704540875059e-06,
+      "loss": 1.3133,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626062794351518e-06,
+      "loss": 0.9842,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625078937501089e-06,
+      "loss": 1.1941,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624093838463541e-06,
+      "loss": 1.3387,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62310749750345e-06,
+      "loss": 1.1564,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622119914885725e-06,
+      "loss": 1.2774,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621131090875603e-06,
+      "loss": 1.1899,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620141025738665e-06,
+      "loss": 1.1442,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619149719740817e-06,
+      "loss": 1.2379,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618157173148298e-06,
+      "loss": 1.1774,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617163386227683e-06,
+      "loss": 0.7932,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61616835924588e-06,
+      "loss": 1.38,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615172092470134e-06,
+      "loss": 1.6384,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614174586168012e-06,
+      "loss": 1.2054,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613175840607428e-06,
+      "loss": 1.3326,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612175856056614e-06,
+      "loss": 1.3049,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611174632784147e-06,
+      "loss": 1.4493,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610172171058929e-06,
+      "loss": 1.5361,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609168471150202e-06,
+      "loss": 1.1026,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608163533327532e-06,
+      "loss": 1.446,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607157357860823e-06,
+      "loss": 1.025,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60614994502031e-06,
+      "loss": 1.4628,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605141295076561e-06,
+      "loss": 1.2442,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604131408300474e-06,
+      "loss": 1.2699,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603120284963284e-06,
+      "loss": 1.0057,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60210792533655e-06,
+      "loss": 1.3286,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601094329692173e-06,
+      "loss": 1.2834,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600079498302375e-06,
+      "loss": 1.3377,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599063431439721e-06,
+      "loss": 0.872,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598046129377101e-06,
+      "loss": 1.3907,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597027592387739e-06,
+      "loss": 1.5915,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.596007820745188e-06,
+      "loss": 1.3931,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594986814723335e-06,
+      "loss": 1.0603,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593964574596399e-06,
+      "loss": 1.0345,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59294110063893e-06,
+      "loss": 1.4771,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59191639312581e-06,
+      "loss": 1.3601,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590890452332249e-06,
+      "loss": 1.394,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58986327853379e-06,
+      "loss": 1.3055,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588834872006308e-06,
+      "loss": 1.1906,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587805233026011e-06,
+      "loss": 1.4674,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586774361869436e-06,
+      "loss": 1.0642,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585742258813447e-06,
+      "loss": 1.1432,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584708924135245e-06,
+      "loss": 1.5874,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583674358112357e-06,
+      "loss": 1.7677,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582638561022646e-06,
+      "loss": 1.3357,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581601533144302e-06,
+      "loss": 0.9126,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580563274755848e-06,
+      "loss": 1.213,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57952378613613e-06,
+      "loss": 1.4081,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578483067564335e-06,
+      "loss": 1.2606,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577441119319974e-06,
+      "loss": 1.1321,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576397941682891e-06,
+      "loss": 1.4603,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575353534933256e-06,
+      "loss": 1.3029,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574307899351574e-06,
+      "loss": 1.33,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57326103521868e-06,
+      "loss": 1.1633,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572212942815734e-06,
+      "loss": 1.3663,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571163622424226e-06,
+      "loss": 1.1861,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570113074325986e-06,
+      "loss": 1.1678,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569061298803162e-06,
+      "loss": 1.1567,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568008296138238e-06,
+      "loss": 1.4457,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566954066614023e-06,
+      "loss": 1.2629,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565898610513661e-06,
+      "loss": 1.4722,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56484192812062e-06,
+      "loss": 1.1897,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563784019718704e-06,
+      "loss": 1.1652,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562724885592038e-06,
+      "loss": 1.0623,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561664526025082e-06,
+      "loss": 1.3181,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560602941302624e-06,
+      "loss": 1.1362,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55954013170978e-06,
+      "loss": 1.1123,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558476097531996e-06,
+      "loss": 1.4409,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557410839055047e-06,
+      "loss": 1.3976,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556344356565035e-06,
+      "loss": 1.2873,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555276650348393e-06,
+      "loss": 1.4798,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554207720691881e-06,
+      "loss": 1.1439,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55313756788259e-06,
+      "loss": 1.6671,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552066192207935e-06,
+      "loss": 1.1992,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550993593955665e-06,
+      "loss": 1.3504,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549919773413853e-06,
+      "loss": 1.1578,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548844730870903e-06,
+      "loss": 1.2239,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547768466615544e-06,
+      "loss": 1.5155,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546690980936836e-06,
+      "loss": 1.3697,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545612274124168e-06,
+      "loss": 1.2181,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544532346467254e-06,
+      "loss": 1.6087,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543451198256136e-06,
+      "loss": 1.4129,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542368829781186e-06,
+      "loss": 1.6747,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541285241333104e-06,
+      "loss": 1.1504,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540200433202913e-06,
+      "loss": 1.3535,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539114405681971e-06,
+      "loss": 1.3703,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538027159061955e-06,
+      "loss": 1.5081,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536938693634878e-06,
+      "loss": 1.4345,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535849009693072e-06,
+      "loss": 1.2326,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534758107529204e-06,
+      "loss": 1.08,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533665987436262e-06,
+      "loss": 1.4879,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532572649707565e-06,
+      "loss": 1.2993,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531478094636758e-06,
+      "loss": 1.3417,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530382322517812e-06,
+      "loss": 1.0734,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529285333645027e-06,
+      "loss": 1.1257,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528187128313026e-06,
+      "loss": 0.9747,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527087706816762e-06,
+      "loss": 1.1355,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525987069451514e-06,
+      "loss": 1.4101,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524885216512887e-06,
+      "loss": 1.2795,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523782148296812e-06,
+      "loss": 1.2544,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522677865099548e-06,
+      "loss": 1.2622,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52157236721768e-06,
+      "loss": 1.1673,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520465654948119e-06,
+      "loss": 1.3691,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519357728588098e-06,
+      "loss": 1.1522,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518248588435185e-06,
+      "loss": 1.2213,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517138234787267e-06,
+      "loss": 1.1695,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516026667942557e-06,
+      "loss": 1.501,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514913888199595e-06,
+      "loss": 1.4277,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513799895857252e-06,
+      "loss": 1.1666,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512684691214717e-06,
+      "loss": 1.0667,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511568274571508e-06,
+      "loss": 1.3624,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51045064622747e-06,
+      "loss": 1.0942,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509331806482767e-06,
+      "loss": 1.3124,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508211755637897e-06,
+      "loss": 1.0448,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507090493993677e-06,
+      "loss": 1.1354,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505968021851255e-06,
+      "loss": 1.0197,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504844339512096e-06,
+      "loss": 1.2894,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503719447277996e-06,
+      "loss": 1.2112,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502593345451078e-06,
+      "loss": 1.4039,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501466034333782e-06,
+      "loss": 1.1173,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500337514228878e-06,
+      "loss": 1.1097,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499207785439462e-06,
+      "loss": 1.0549,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49807684826895e-06,
+      "loss": 1.3345,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496944703021087e-06,
+      "loss": 1.4,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495811349999941e-06,
+      "loss": 1.4147,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4946767895099e-06,
+      "loss": 1.3896,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493541021855685e-06,
+      "loss": 1.1852,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492404047342332e-06,
+      "loss": 1.7441,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49126586627521e-06,
+      "loss": 1.3419,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490126478960003e-06,
+      "loss": 1.0037,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488985885702728e-06,
+      "loss": 1.2228,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487844086809718e-06,
+      "loss": 1.1327,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486701082587635e-06,
+      "loss": 1.1852,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485556873343462e-06,
+      "loss": 1.1953,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484411459384508e-06,
+      "loss": 1.1091,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483264841018403e-06,
+      "loss": 1.4416,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482117018553101e-06,
+      "loss": 1.258,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480967992296881e-06,
+      "loss": 1.3338,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479817762558345e-06,
+      "loss": 1.0852,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478666329646417e-06,
+      "loss": 1.2982,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477513693870347e-06,
+      "loss": 1.1114,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4763598555397e-06,
+      "loss": 1.3368,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475204814964374e-06,
+      "loss": 1.2174,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474048572454585e-06,
+      "loss": 1.0065,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472891128320874e-06,
+      "loss": 1.522,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4717324828741e-06,
+      "loss": 1.1306,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470572636425451e-06,
+      "loss": 1.1545,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46941158928643e-06,
+      "loss": 1.3497,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46824934176887e-06,
+      "loss": 1.1777,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467085894184925e-06,
+      "loss": 1.1175,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465921246847067e-06,
+      "loss": 1.1176,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464755400068093e-06,
+      "loss": 1.3183,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463588354161122e-06,
+      "loss": 1.591,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462420109439595e-06,
+      "loss": 1.575,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461250666217277e-06,
+      "loss": 1.2573,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46008002480825e-06,
+      "loss": 0.8862,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458908185526921e-06,
+      "loss": 1.3567,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457735148688018e-06,
+      "loss": 1.1088,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456560914606594e-06,
+      "loss": 1.1718,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45538548359802e-06,
+      "loss": 1.4247,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454208855977986e-06,
+      "loss": 1.4701,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45303103206251e-06,
+      "loss": 1.2146,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451852012167924e-06,
+      "loss": 1.3185,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450671796610887e-06,
+      "loss": 1.2193,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449490385708378e-06,
+      "loss": 1.4591,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448307779777694e-06,
+      "loss": 1.2002,
+      "step": 1103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447123979136457e-06,
+      "loss": 1.3724,
+      "step": 1104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445938984102607e-06,
+      "loss": 0.9376,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444752794994408e-06,
+      "loss": 1.3541,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443565412130438e-06,
+      "loss": 1.352,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4423768358296e-06,
+      "loss": 1.2334,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441187066411123e-06,
+      "loss": 1.0981,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439996104194546e-06,
+      "loss": 1.2706,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438803949499738e-06,
+      "loss": 1.2658,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437610602646878e-06,
+      "loss": 1.1797,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436416063956475e-06,
+      "loss": 1.2369,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43522033374935e-06,
+      "loss": 1.4671,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434023412346653e-06,
+      "loss": 1.4435,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432825300069848e-06,
+      "loss": 1.2069,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431625997240715e-06,
+      "loss": 1.5572,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430425504181361e-06,
+      "loss": 1.3689,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429223821214213e-06,
+      "loss": 1.1736,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428020948662012e-06,
+      "loss": 1.4497,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42681688684782e-06,
+      "loss": 1.0317,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425611636095023e-06,
+      "loss": 1.2437,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42440519672732e-06,
+      "loss": 1.3031,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423197569068733e-06,
+      "loss": 1.2305,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421988753443602e-06,
+      "loss": 1.5507,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420778750176588e-06,
+      "loss": 1.2108,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419567559592669e-06,
+      "loss": 1.2328,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418355182017138e-06,
+      "loss": 1.1177,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417141617775618e-06,
+      "loss": 1.3749,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41592686719404e-06,
+      "loss": 1.0713,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414710930598655e-06,
+      "loss": 1.1963,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413493808316038e-06,
+      "loss": 1.3141,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41227550067308e-06,
+      "loss": 1.2852,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411056007996989e-06,
+      "loss": 1.1642,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40983533061529e-06,
+      "loss": 1.3601,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408613468855829e-06,
+      "loss": 1.2792,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407390423046772e-06,
+      "loss": 1.0079,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406166193516596e-06,
+      "loss": 1.0455,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404940780594105e-06,
+      "loss": 1.1425,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403714184608411e-06,
+      "loss": 1.1914,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402486405888953e-06,
+      "loss": 1.2684,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40125744476548e-06,
+      "loss": 1.33,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400027301568063e-06,
+      "loss": 1.383,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398795976627091e-06,
+      "loss": 0.9595,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397563470273267e-06,
+      "loss": 1.1971,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396329782837614e-06,
+      "loss": 1.4356,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395094914651472e-06,
+      "loss": 1.0431,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393858866046494e-06,
+      "loss": 1.1695,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39262163735466e-06,
+      "loss": 1.4009,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391383228908253e-06,
+      "loss": 1.3663,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390143641039885e-06,
+      "loss": 1.5531,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388902874082482e-06,
+      "loss": 1.5111,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387660928369278e-06,
+      "loss": 1.5936,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386417804233836e-06,
+      "loss": 1.1724,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385173502010028e-06,
+      "loss": 1.1453,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383928022032044e-06,
+      "loss": 1.3252,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382681364634388e-06,
+      "loss": 1.2843,
+      "step": 1157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381433530151887e-06,
+      "loss": 1.2808,
+      "step": 1158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38018451891968e-06,
+      "loss": 1.1023,
+      "step": 1159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37893433127322e-06,
+      "loss": 1.0836,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377682967548276e-06,
+      "loss": 1.3361,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376430428080939e-06,
+      "loss": 1.1723,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375176713207607e-06,
+      "loss": 1.4686,
+      "step": 1163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373921823265004e-06,
+      "loss": 1.1516,
+      "step": 1164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372665758590158e-06,
+      "loss": 1.2283,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371408519520421e-06,
+      "loss": 0.962,
+      "step": 1166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370150106393461e-06,
+      "loss": 0.9777,
+      "step": 1167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36889051954725e-06,
+      "loss": 1.3597,
+      "step": 1168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367629759320093e-06,
+      "loss": 1.2522,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366367826050593e-06,
+      "loss": 1.1577,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36510472007768e-06,
+      "loss": 1.3464,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36384044174059e-06,
+      "loss": 1.2173,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362574991378882e-06,
+      "loss": 0.9899,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361308369332426e-06,
+      "loss": 0.9947,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360040575941406e-06,
+      "loss": 1.3343,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358771611546319e-06,
+      "loss": 1.3045,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357501476487982e-06,
+      "loss": 1.2052,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356230171107524e-06,
+      "loss": 1.0587,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354957695746385e-06,
+      "loss": 1.1738,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353684050746323e-06,
+      "loss": 1.4932,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352409236449408e-06,
+      "loss": 1.4194,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351133253198027e-06,
+      "loss": 1.371,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34985610133488e-06,
+      "loss": 1.2827,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348577781202976e-06,
+      "loss": 1.3561,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347298293145644e-06,
+      "loss": 1.1664,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346017637506523e-06,
+      "loss": 1.3418,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34473581462957e-06,
+      "loss": 1.2407,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343452824859048e-06,
+      "loss": 1.171,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34216866853954e-06,
+      "loss": 1.1968,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340883346015941e-06,
+      "loss": 1.2292,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339596857633458e-06,
+      "loss": 1.2418,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338309203737609e-06,
+      "loss": 1.211,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33702038467423e-06,
+      "loss": 1.1903,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335730400789466e-06,
+      "loss": 1.1546,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334439252429779e-06,
+      "loss": 0.9243,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333146939941938e-06,
+      "loss": 1.3743,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331853463673028e-06,
+      "loss": 1.5302,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330558823970448e-06,
+      "loss": 1.275,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329263021181906e-06,
+      "loss": 1.3147,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327966055655424e-06,
+      "loss": 1.3848,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326667927739337e-06,
+      "loss": 1.3948,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325368637782292e-06,
+      "loss": 1.2933,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324068186133246e-06,
+      "loss": 1.1474,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322766573141473e-06,
+      "loss": 1.0764,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321463799156552e-06,
+      "loss": 1.2144,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320159864528378e-06,
+      "loss": 1.2649,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318854769607157e-06,
+      "loss": 1.1213,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31754851474341e-06,
+      "loss": 1.532,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316241100287962e-06,
+      "loss": 1.3305,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314932526591956e-06,
+      "loss": 1.2488,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313622794006842e-06,
+      "loss": 1.2633,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312311902884388e-06,
+      "loss": 1.5744,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310999853576661e-06,
+      "loss": 1.5772,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309686646436053e-06,
+      "loss": 1.3422,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308372281815258e-06,
+      "loss": 1.1986,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307056760067284e-06,
+      "loss": 1.0519,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305740081545448e-06,
+      "loss": 1.369,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30442224660338e-06,
+      "loss": 1.2473,
+      "step": 1218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.303103255595016e-06,
+      "loss": 1.4961,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301783108874611e-06,
+      "loss": 1.2035,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300461806796724e-06,
+      "loss": 1.5295,
+      "step": 1221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299139349716221e-06,
+      "loss": 1.2891,
+      "step": 1222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29781573798829e-06,
+      "loss": 1.4624,
+      "step": 1223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296490971968416e-06,
+      "loss": 1.1294,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295165052012403e-06,
+      "loss": 1.2816,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293837978476359e-06,
+      "loss": 1.028,
+      "step": 1226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292509751716707e-06,
+      "loss": 1.2792,
+      "step": 1227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291180372090178e-06,
+      "loss": 1.2743,
+      "step": 1228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289849839953811e-06,
+      "loss": 1.0938,
+      "step": 1229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288518155664956e-06,
+      "loss": 1.4585,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28718531958127e-06,
+      "loss": 1.4396,
+      "step": 1231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285851332060722e-06,
+      "loss": 1.3826,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284516193461593e-06,
+      "loss": 1.2327,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283179904142465e-06,
+      "loss": 1.1728,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281842464462236e-06,
+      "loss": 1.076,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280503874780112e-06,
+      "loss": 1.3763,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279164135455606e-06,
+      "loss": 0.8908,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277823246848537e-06,
+      "loss": 0.9467,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27648120931904e-06,
+      "loss": 1.3277,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275138023227555e-06,
+      "loss": 1.1734,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273793688934825e-06,
+      "loss": 1.0785,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272448206801912e-06,
+      "loss": 1.2911,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271101577190178e-06,
+      "loss": 1.3125,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269753800461299e-06,
+      "loss": 1.4577,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268404876977251e-06,
+      "loss": 1.2943,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267054807100327e-06,
+      "loss": 0.9536,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265703591193122e-06,
+      "loss": 1.3083,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264351229618541e-06,
+      "loss": 1.3061,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262997722739798e-06,
+      "loss": 1.2741,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261643070920409e-06,
+      "loss": 1.2992,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260287274524208e-06,
+      "loss": 0.9434,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258930333915325e-06,
+      "loss": 1.3225,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257572249458204e-06,
+      "loss": 1.3153,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256213021517593e-06,
+      "loss": 1.0923,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25485265045855e-06,
+      "loss": 1.3807,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253491136646437e-06,
+      "loss": 1.3761,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252128480446927e-06,
+      "loss": 1.0936,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250764682225997e-06,
+      "loss": 1.2115,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249399742349928e-06,
+      "loss": 1.0173,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248033661185313e-06,
+      "loss": 1.0844,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24666643909905e-06,
+      "loss": 1.1712,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24529807645834e-06,
+      "loss": 1.3976,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243928573630695e-06,
+      "loss": 1.3892,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24255793098393e-06,
+      "loss": 0.9577,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241186148886168e-06,
+      "loss": 1.1918,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23981322770584e-06,
+      "loss": 1.289,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238439167811672e-06,
+      "loss": 1.2578,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237063969572713e-06,
+      "loss": 1.5153,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235687633358305e-06,
+      "loss": 1.0648,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2343101595381e-06,
+      "loss": 1.3853,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232931548482054e-06,
+      "loss": 1.2293,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23155180056043e-06,
+      "loss": 1.5405,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230170916143795e-06,
+      "loss": 1.3077,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228788895603024e-06,
+      "loss": 1.0156,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227405739309294e-06,
+      "loss": 0.9744,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226021447634085e-06,
+      "loss": 0.9991,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224636020949193e-06,
+      "loss": 1.2017,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223249459626704e-06,
+      "loss": 1.1828,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221861764039018e-06,
+      "loss": 1.0655,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220472934558838e-06,
+      "loss": 1.152,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219082971559168e-06,
+      "loss": 1.2845,
+      "step": 1281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217691875413323e-06,
+      "loss": 1.1601,
+      "step": 1282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216299646494915e-06,
+      "loss": 1.2201,
+      "step": 1283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214906285177867e-06,
+      "loss": 1.2962,
+      "step": 1284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213511791836402e-06,
+      "loss": 1.0615,
+      "step": 1285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212116166845048e-06,
+      "loss": 0.989,
+      "step": 1286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210719410578635e-06,
+      "loss": 1.3152,
+      "step": 1287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209321523412303e-06,
+      "loss": 0.9728,
+      "step": 1288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207922505721486e-06,
+      "loss": 1.1791,
+      "step": 1289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206522357881931e-06,
+      "loss": 1.4059,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205121080269686e-06,
+      "loss": 1.55,
+      "step": 1291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203718673261098e-06,
+      "loss": 1.2478,
+      "step": 1292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202315137232818e-06,
+      "loss": 1.0394,
+      "step": 1293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20091047256181e-06,
+      "loss": 1.1653,
+      "step": 1294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199504679625328e-06,
+      "loss": 1.2348,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198097758800938e-06,
+      "loss": 1.2801,
+      "step": 1296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196689710466504e-06,
+      "loss": 1.4256,
+      "step": 1297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195280535000196e-06,
+      "loss": 1.4139,
+      "step": 1298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193870232780481e-06,
+      "loss": 0.9184,
+      "step": 1299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19245880418614e-06,
+      "loss": 1.4133,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191046249596242e-06,
+      "loss": 1.4257,
+      "step": 1301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189632569390172e-06,
+      "loss": 1.5086,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188217763947609e-06,
+      "loss": 1.0503,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186801833648535e-06,
+      "loss": 1.2332,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185384778873236e-06,
+      "loss": 1.0198,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183966600002301e-06,
+      "loss": 1.1528,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182547297416618e-06,
+      "loss": 1.3893,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181126871497378e-06,
+      "loss": 1.2509,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179705322626076e-06,
+      "loss": 1.4848,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178282651184506e-06,
+      "loss": 1.2075,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176858857554759e-06,
+      "loss": 1.1464,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175433942119238e-06,
+      "loss": 1.182,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174007905260643e-06,
+      "loss": 1.3808,
+      "step": 1313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172580747361968e-06,
+      "loss": 1.4114,
+      "step": 1314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171152468806517e-06,
+      "loss": 1.1345,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169723069977892e-06,
+      "loss": 1.2293,
+      "step": 1316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168292551259996e-06,
+      "loss": 1.0407,
+      "step": 1317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166860913037032e-06,
+      "loss": 1.3958,
+      "step": 1318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165428155693504e-06,
+      "loss": 1.534,
+      "step": 1319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163994279614218e-06,
+      "loss": 1.2362,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162559285184278e-06,
+      "loss": 1.2302,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161123172789091e-06,
+      "loss": 1.2993,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159685942814363e-06,
+      "loss": 1.5424,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158247595646098e-06,
+      "loss": 1.1242,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156808131670603e-06,
+      "loss": 1.035,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155367551274485e-06,
+      "loss": 0.8892,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15392585484465e-06,
+      "loss": 0.9189,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152483042768302e-06,
+      "loss": 1.6333,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151039115432946e-06,
+      "loss": 1.2057,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149594073226391e-06,
+      "loss": 0.8735,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148147916536739e-06,
+      "loss": 1.1884,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14670064575239e-06,
+      "loss": 1.0384,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145252261262052e-06,
+      "loss": 1.4011,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143802763454723e-06,
+      "loss": 1.1787,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142352152719706e-06,
+      "loss": 1.2534,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140900429446601e-06,
+      "loss": 1.3767,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139447594025307e-06,
+      "loss": 1.22,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137993646846018e-06,
+      "loss": 0.9935,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136538588299233e-06,
+      "loss": 1.2152,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135082418775746e-06,
+      "loss": 0.9858,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133625138666651e-06,
+      "loss": 1.3393,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132166748363335e-06,
+      "loss": 1.1744,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130707248257493e-06,
+      "loss": 1.1755,
+      "step": 1343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.129246638741108e-06,
+      "loss": 1.1225,
+      "step": 1344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127784920206464e-06,
+      "loss": 1.2142,
+      "step": 1345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126322093046149e-06,
+      "loss": 1.0319,
+      "step": 1346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12485815765304e-06,
+      "loss": 1.5496,
+      "step": 1347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123393114420318e-06,
+      "loss": 1.146,
+      "step": 1348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121926963741457e-06,
+      "loss": 1.2853,
+      "step": 1349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120459706010233e-06,
+      "loss": 1.3469,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118991341620713e-06,
+      "loss": 1.0952,
+      "step": 1351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11752187096727e-06,
+      "loss": 1.2481,
+      "step": 1352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116051294444565e-06,
+      "loss": 1.2007,
+      "step": 1353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114579612447562e-06,
+      "loss": 1.3677,
+      "step": 1354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113106825371519e-06,
+      "loss": 1.3294,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111632933611993e-06,
+      "loss": 1.1755,
+      "step": 1356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110157937564837e-06,
+      "loss": 1.2662,
+      "step": 1357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108681837626199e-06,
+      "loss": 1.276,
+      "step": 1358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107204634192523e-06,
+      "loss": 1.2108,
+      "step": 1359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105726327660556e-06,
+      "loss": 1.0192,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10424691842733e-06,
+      "loss": 1.094,
+      "step": 1361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102766406890185e-06,
+      "loss": 1.5832,
+      "step": 1362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101284793446748e-06,
+      "loss": 1.0387,
+      "step": 1363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099802078494947e-06,
+      "loss": 1.1671,
+      "step": 1364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098318262433004e-06,
+      "loss": 1.1413,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096833345659437e-06,
+      "loss": 1.1883,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095347328573058e-06,
+      "loss": 1.5987,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09386021157298e-06,
+      "loss": 1.4416,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092371995058604e-06,
+      "loss": 1.3248,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09088267942963e-06,
+      "loss": 1.24,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089392265086055e-06,
+      "loss": 1.3002,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087900752428168e-06,
+      "loss": 1.4236,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086408141856555e-06,
+      "loss": 1.3898,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084914433772094e-06,
+      "loss": 1.3414,
+      "step": 1374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083419628575961e-06,
+      "loss": 1.1356,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081923726669626e-06,
+      "loss": 1.0751,
+      "step": 1376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080426728454853e-06,
+      "loss": 1.1837,
+      "step": 1377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0789286343337e-06,
+      "loss": 1.3152,
+      "step": 1378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07742944470852e-06,
+      "loss": 1.4382,
+      "step": 1379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075929159981957e-06,
+      "loss": 1.2516,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074427780556956e-06,
+      "loss": 1.0872,
+      "step": 1381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072925306836751e-06,
+      "loss": 1.3946,
+      "step": 1382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07142173922487e-06,
+      "loss": 1.3148,
+      "step": 1383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06991707812514e-06,
+      "loss": 1.1,
+      "step": 1384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06841132394167e-06,
+      "loss": 1.2617,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066904477078875e-06,
+      "loss": 1.1999,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065396537941457e-06,
+      "loss": 1.077,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063887506934417e-06,
+      "loss": 1.0449,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062377384463039e-06,
+      "loss": 1.5025,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06086617093291e-06,
+      "loss": 0.9665,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059353866749904e-06,
+      "loss": 1.3187,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057840472320192e-06,
+      "loss": 1.2243,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056325988050237e-06,
+      "loss": 1.0537,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054810414346789e-06,
+      "loss": 1.2143,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053293751616901e-06,
+      "loss": 0.9751,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05177600026791e-06,
+      "loss": 1.3257,
+      "step": 1396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050257160707449e-06,
+      "loss": 1.2176,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048737233343442e-06,
+      "loss": 1.4315,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047216218584105e-06,
+      "loss": 0.8722,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045694116837948e-06,
+      "loss": 1.6107,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044170928513773e-06,
+      "loss": 1.2124,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042646654020667e-06,
+      "loss": 1.1857,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04112129376802e-06,
+      "loss": 1.5911,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039594848165507e-06,
+      "loss": 1.2786,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038067317623092e-06,
+      "loss": 1.5649,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036538702551037e-06,
+      "loss": 1.4229,
+      "step": 1406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035009003359889e-06,
+      "loss": 1.5181,
+      "step": 1407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03347822046049e-06,
+      "loss": 0.8909,
+      "step": 1408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031946354263975e-06,
+      "loss": 1.2577,
+      "step": 1409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03041340518176e-06,
+      "loss": 1.0741,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028879373625569e-06,
+      "loss": 1.2553,
+      "step": 1411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027344260007401e-06,
+      "loss": 1.2347,
+      "step": 1412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025808064739551e-06,
+      "loss": 1.4452,
+      "step": 1413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024270788234606e-06,
+      "loss": 1.7836,
+      "step": 1414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02273243090544e-06,
+      "loss": 0.9704,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021192993165224e-06,
+      "loss": 1.2376,
+      "step": 1416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019652475427411e-06,
+      "loss": 1.086,
+      "step": 1417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01811087810575e-06,
+      "loss": 0.9673,
+      "step": 1418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016568201614275e-06,
+      "loss": 1.5457,
+      "step": 1419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015024446367315e-06,
+      "loss": 1.1934,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013479612779486e-06,
+      "loss": 1.4231,
+      "step": 1421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01193370126569e-06,
+      "loss": 1.2833,
+      "step": 1422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010386712241128e-06,
+      "loss": 1.4993,
+      "step": 1423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008838646121282e-06,
+      "loss": 0.889,
+      "step": 1424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007289503321925e-06,
+      "loss": 1.2181,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005739284259123e-06,
+      "loss": 1.5907,
+      "step": 1426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004187989349226e-06,
+      "loss": 1.2231,
+      "step": 1427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002635619008877e-06,
+      "loss": 1.0147,
+      "step": 1428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001082173655005e-06,
+      "loss": 0.9539,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999527653704829e-06,
+      "loss": 1.1554,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997972059575855e-06,
+      "loss": 1.4666,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996415391685882e-06,
+      "loss": 1.4067,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99485765045299e-06,
+      "loss": 1.1697,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993298836295556e-06,
+      "loss": 0.8879,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991738949632238e-06,
+      "loss": 1.24,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990177990881986e-06,
+      "loss": 1.4279,
+      "step": 1436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988615960464035e-06,
+      "loss": 1.2892,
+      "step": 1437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987052858797914e-06,
+      "loss": 1.3102,
+      "step": 1438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985488686303427e-06,
+      "loss": 1.2912,
+      "step": 1439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983923443400682e-06,
+      "loss": 1.1162,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982357130510064e-06,
+      "loss": 1.2143,
+      "step": 1441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980789748052245e-06,
+      "loss": 0.9437,
+      "step": 1442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979221296448189e-06,
+      "loss": 1.0904,
+      "step": 1443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977651776119145e-06,
+      "loss": 1.1795,
+      "step": 1444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97608118748665e-06,
+      "loss": 1.3661,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974509530972523e-06,
+      "loss": 1.189,
+      "step": 1446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97293680699888e-06,
+      "loss": 1.1711,
+      "step": 1447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971363015988115e-06,
+      "loss": 1.257,
+      "step": 1448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969788158362908e-06,
+      "loss": 1.1199,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968212234546235e-06,
+      "loss": 1.2658,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966635244961344e-06,
+      "loss": 1.2439,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965057190031785e-06,
+      "loss": 1.3874,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96347807018138e-06,
+      "loss": 1.1847,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961897885834247e-06,
+      "loss": 1.2611,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960316637414788e-06,
+      "loss": 1.3159,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958734325347684e-06,
+      "loss": 1.256,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95715095005791e-06,
+      "loss": 1.2112,
+      "step": 1457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955566511970721e-06,
+      "loss": 1.3089,
+      "step": 1458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953981011511661e-06,
+      "loss": 0.7546,
+      "step": 1459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95239444910656e-06,
+      "loss": 1.303,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950806825181529e-06,
+      "loss": 1.3427,
+      "step": 1461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949218140162965e-06,
+      "loss": 1.1206,
+      "step": 1462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947628394477555e-06,
+      "loss": 1.5321,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946037588552266e-06,
+      "loss": 1.1002,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944445722814348e-06,
+      "loss": 1.2454,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94285279769134e-06,
+      "loss": 1.3828,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941258813611067e-06,
+      "loss": 1.1284,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939663771001632e-06,
+      "loss": 1.3526,
+      "step": 1468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93806767029143e-06,
+      "loss": 1.0956,
+      "step": 1469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93647051190913e-06,
+      "loss": 1.2525,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934872296283697e-06,
+      "loss": 1.0334,
+      "step": 1471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93327302384437e-06,
+      "loss": 1.3629,
+      "step": 1472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931672695020677e-06,
+      "loss": 1.1118,
+      "step": 1473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930071310242429e-06,
+      "loss": 1.3959,
+      "step": 1474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92846886993972e-06,
+      "loss": 1.1314,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926865374542928e-06,
+      "loss": 1.2146,
+      "step": 1476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925260824482714e-06,
+      "loss": 1.2199,
+      "step": 1477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92365522019002e-06,
+      "loss": 1.2248,
+      "step": 1478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922048562096076e-06,
+      "loss": 1.2156,
+      "step": 1479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920440850632395e-06,
+      "loss": 1.107,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918832086230764e-06,
+      "loss": 1.0731,
+      "step": 1481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917222269323263e-06,
+      "loss": 1.2764,
+      "step": 1482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915611400342252e-06,
+      "loss": 1.3113,
+      "step": 1483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91399947972037e-06,
+      "loss": 0.8724,
+      "step": 1484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91238650789054e-06,
+      "loss": 1.3607,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91077248528597e-06,
+      "loss": 1.5892,
+      "step": 1486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90915741234015e-06,
+      "loss": 1.29,
+      "step": 1487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907541289486847e-06,
+      "loss": 1.1848,
+      "step": 1488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905924117160116e-06,
+      "loss": 1.0481,
+      "step": 1489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904305895794292e-06,
+      "loss": 1.4812,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902686625823986e-06,
+      "loss": 1.0671,
+      "step": 1491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901066307684102e-06,
+      "loss": 1.2435,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899444941809815e-06,
+      "loss": 1.4838,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89782252863659e-06,
+      "loss": 1.0384,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896199068600164e-06,
+      "loss": 1.3,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894574562136561e-06,
+      "loss": 1.2857,
+      "step": 1496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892949009682088e-06,
+      "loss": 1.214,
+      "step": 1497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89132241167333e-06,
+      "loss": 1.3121,
+      "step": 1498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889694768547148e-06,
+      "loss": 1.3038,
+      "step": 1499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888066080740692e-06,
+      "loss": 0.9818,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886436348691391e-06,
+      "loss": 1.2041,
+      "step": 1501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88480557283695e-06,
+      "loss": 1.0871,
+      "step": 1502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883173753615357e-06,
+      "loss": 1.24,
+      "step": 1503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88154089146488e-06,
+      "loss": 1.2884,
+      "step": 1504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879906986824067e-06,
+      "loss": 0.9981,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878272040131748e-06,
+      "loss": 1.3454,
+      "step": 1506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876636051827031e-06,
+      "loss": 1.1051,
+      "step": 1507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874999022349303e-06,
+      "loss": 1.0503,
+      "step": 1508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873360952138231e-06,
+      "loss": 1.3314,
+      "step": 1509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871721841633762e-06,
+      "loss": 1.1636,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870081691276124e-06,
+      "loss": 1.0907,
+      "step": 1511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868440501505822e-06,
+      "loss": 1.4804,
+      "step": 1512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866798272763641e-06,
+      "loss": 1.5832,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865155005490643e-06,
+      "loss": 1.2548,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863510700128172e-06,
+      "loss": 1.0849,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861865357117852e-06,
+      "loss": 1.3135,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860218976901582e-06,
+      "loss": 1.2664,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858571559921539e-06,
+      "loss": 1.279,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856923106620178e-06,
+      "loss": 1.3625,
+      "step": 1519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855273617440243e-06,
+      "loss": 1.0345,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85362309282474e-06,
+      "loss": 0.8889,
+      "step": 1521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851971533216968e-06,
+      "loss": 1.1995,
+      "step": 1522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85031893906049e-06,
+      "loss": 1.0464,
+      "step": 1523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848665310799156e-06,
+      "loss": 1.3505,
+      "step": 1524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847010648877095e-06,
+      "loss": 0.9002,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845354953738706e-06,
+      "loss": 1.082,
+      "step": 1526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843698225828671e-06,
+      "loss": 1.0766,
+      "step": 1527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84204046559195e-06,
+      "loss": 1.1259,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840381673473774e-06,
+      "loss": 0.952,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.83872184991966e-06,
+      "loss": 1.0735,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837060995375395e-06,
+      "loss": 1.1368,
+      "step": 1531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835399110287046e-06,
+      "loss": 1.1554,
+      "step": 1532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833736195100958e-06,
+      "loss": 1.3902,
+      "step": 1533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832072250263746e-06,
+      "loss": 0.9975,
+      "step": 1534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83040727622231e-06,
+      "loss": 1.2393,
+      "step": 1535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82874127342382e-06,
+      "loss": 1.1685,
+      "step": 1536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827074242315729e-06,
+      "loss": 1.4707,
+      "step": 1537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82540618334576e-06,
+      "loss": 1.0854,
+      "step": 1538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823737096961916e-06,
+      "loss": 1.2957,
+      "step": 1539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82206698361247e-06,
+      "loss": 1.3041,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820395843745977e-06,
+      "loss": 1.4271,
+      "step": 1541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818723677811269e-06,
+      "loss": 0.9311,
+      "step": 1542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817050486257448e-06,
+      "loss": 1.0771,
+      "step": 1543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815376269533893e-06,
+      "loss": 1.1771,
+      "step": 1544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81370102809026e-06,
+      "loss": 1.2776,
+      "step": 1545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812024762376477e-06,
+      "loss": 1.2266,
+      "step": 1546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810347472842754e-06,
+      "loss": 0.9788,
+      "step": 1547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808669159939568e-06,
+      "loss": 1.3493,
+      "step": 1548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806989824117676e-06,
+      "loss": 1.1598,
+      "step": 1549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805309465828105e-06,
+      "loss": 1.0525,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803628085522165e-06,
+      "loss": 1.6216,
+      "step": 1551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80194568365143e-06,
+      "loss": 1.1174,
+      "step": 1552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800262260667756e-06,
+      "loss": 1.2826,
+      "step": 1553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798577817023269e-06,
+      "loss": 1.4197,
+      "step": 1554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796892353170372e-06,
+      "loss": 1.2478,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795205869561742e-06,
+      "loss": 0.9227,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793518366650326e-06,
+      "loss": 1.2565,
+      "step": 1557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79182984488935e-06,
+      "loss": 1.0305,
+      "step": 1558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79014030473231e-06,
+      "loss": 1.3155,
+      "step": 1559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788449746632976e-06,
+      "loss": 1.654,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786758171045394e-06,
+      "loss": 1.2454,
+      "step": 1561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78506557842388e-06,
+      "loss": 1.317,
+      "step": 1562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783371969223022e-06,
+      "loss": 1.3954,
+      "step": 1563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781677343897687e-06,
+      "loss": 1.0937,
+      "step": 1564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779981702903012e-06,
+      "loss": 1.0799,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778285046694403e-06,
+      "loss": 1.4116,
+      "step": 1566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776587375727545e-06,
+      "loss": 1.3457,
+      "step": 1567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77488869045839e-06,
+      "loss": 1.2382,
+      "step": 1568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773188991343165e-06,
+      "loss": 1.122,
+      "step": 1569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771488278838368e-06,
+      "loss": 1.4026,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769786553400777e-06,
+      "loss": 1.2375,
+      "step": 1571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768083815487428e-06,
+      "loss": 1.5259,
+      "step": 1572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766380065555637e-06,
+      "loss": 1.2845,
+      "step": 1573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764675304062992e-06,
+      "loss": 1.4207,
+      "step": 1574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762969531467354e-06,
+      "loss": 0.9524,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76126274822685e-06,
+      "loss": 1.3831,
+      "step": 1576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759554954799886e-06,
+      "loss": 0.9257,
+      "step": 1577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75784615164513e-06,
+      "loss": 1.2425,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756136339221531e-06,
+      "loss": 1.3617,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754425517988298e-06,
+      "loss": 1.2055,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752713688404924e-06,
+      "loss": 1.0557,
+      "step": 1581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751000850931162e-06,
+      "loss": 1.1679,
+      "step": 1582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749287006027041e-06,
+      "loss": 1.0578,
+      "step": 1583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74757215415286e-06,
+      "loss": 1.3257,
+      "step": 1584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745856295769185e-06,
+      "loss": 1.2846,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74413943133686e-06,
+      "loss": 1.1522,
+      "step": 1586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74242156131699e-06,
+      "loss": 1.3173,
+      "step": 1587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740702686170955e-06,
+      "loss": 1.0076,
+      "step": 1588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738982806360405e-06,
+      "loss": 1.0425,
+      "step": 1589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73726192234726e-06,
+      "loss": 1.3819,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735540034593707e-06,
+      "loss": 1.3681,
+      "step": 1591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733817143562207e-06,
+      "loss": 0.8877,
+      "step": 1592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732093249715486e-06,
+      "loss": 1.2522,
+      "step": 1593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.73036835351654e-06,
+      "loss": 0.9746,
+      "step": 1594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72864245542864e-06,
+      "loss": 1.0649,
+      "step": 1595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726915555915317e-06,
+      "loss": 1.2024,
+      "step": 1596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725187655440376e-06,
+      "loss": 1.1883,
+      "step": 1597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723458754467893e-06,
+      "loss": 1.0137,
+      "step": 1598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721728853462207e-06,
+      "loss": 1.3832,
+      "step": 1599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719997952887932e-06,
+      "loss": 1.4824,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718266053209943e-06,
+      "loss": 1.229,
+      "step": 1601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71653315489339e-06,
+      "loss": 1.116,
+      "step": 1602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714799258403688e-06,
+      "loss": 1.1758,
+      "step": 1603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71306436420652e-06,
+      "loss": 1.2814,
+      "step": 1604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711328472767839e-06,
+      "loss": 1.3355,
+      "step": 1605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709591584553865e-06,
+      "loss": 1.531,
+      "step": 1606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707853700031084e-06,
+      "loss": 1.2097,
+      "step": 1607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706114819666249e-06,
+      "loss": 1.2805,
+      "step": 1608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704374943926386e-06,
+      "loss": 1.101,
+      "step": 1609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702634073278784e-06,
+      "loss": 0.9497,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700892208190996e-06,
+      "loss": 1.1488,
+      "step": 1611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699149349130848e-06,
+      "loss": 1.3889,
+      "step": 1612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697405496566434e-06,
+      "loss": 1.1762,
+      "step": 1613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695660650966109e-06,
+      "loss": 1.3213,
+      "step": 1614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693914812798496e-06,
+      "loss": 0.9003,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692167982532487e-06,
+      "loss": 1.0837,
+      "step": 1616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69042016063724e-06,
+      "loss": 1.0929,
+      "step": 1617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688671347582178e-06,
+      "loss": 1.4651,
+      "step": 1618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686921543836993e-06,
+      "loss": 1.3934,
+      "step": 1619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685170749871638e-06,
+      "loss": 1.1627,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683418966156337e-06,
+      "loss": 1.0866,
+      "step": 1621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681666193161578e-06,
+      "loss": 1.1169,
+      "step": 1622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67991243135811e-06,
+      "loss": 1.274,
+      "step": 1623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67815768121696e-06,
+      "loss": 1.2685,
+      "step": 1624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676401943209406e-06,
+      "loss": 1.3449,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.674645217807e-06,
+      "loss": 1.3138,
+      "step": 1626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672887505481557e-06,
+      "loss": 0.8932,
+      "step": 1627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671128806705159e-06,
+      "loss": 1.412,
+      "step": 1628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669369121950148e-06,
+      "loss": 1.3826,
+      "step": 1629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667608451689135e-06,
+      "loss": 1.0358,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665846796394995e-06,
+      "loss": 1.0848,
+      "step": 1631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664084156540864e-06,
+      "loss": 1.1015,
+      "step": 1632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66232053260015e-06,
+      "loss": 1.2883,
+      "step": 1633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660555925046518e-06,
+      "loss": 1.3709,
+      "step": 1634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.658790334353901e-06,
+      "loss": 1.0316,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657023760996497e-06,
+      "loss": 1.1841,
+      "step": 1636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65525620544876e-06,
+      "loss": 1.352,
+      "step": 1637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653487668185419e-06,
+      "loss": 1.2169,
+      "step": 1638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65171814968146e-06,
+      "loss": 1.0984,
+      "step": 1639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649947650412135e-06,
+      "loss": 0.8673,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648176170852955e-06,
+      "loss": 1.3739,
+      "step": 1641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646403711479702e-06,
+      "loss": 1.222,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644630272768413e-06,
+      "loss": 1.0763,
+      "step": 1643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642855855195394e-06,
+      "loss": 1.1184,
+      "step": 1644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64108045923721e-06,
+      "loss": 1.1733,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639304085370692e-06,
+      "loss": 1.0975,
+      "step": 1646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637526734072933e-06,
+      "loss": 1.2222,
+      "step": 1647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635748405821285e-06,
+      "loss": 1.3986,
+      "step": 1648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633969101093366e-06,
+      "loss": 1.2939,
+      "step": 1649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632188820367056e-06,
+      "loss": 1.3716,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630407564120496e-06,
+      "loss": 1.161,
+      "step": 1651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62862533283209e-06,
+      "loss": 1.3353,
+      "step": 1652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626842126980503e-06,
+      "loss": 1.3289,
+      "step": 1653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625057947044662e-06,
+      "loss": 1.1756,
+      "step": 1654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623272793503754e-06,
+      "loss": 1.1,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62148666683723e-06,
+      "loss": 1.1738,
+      "step": 1656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619699567524804e-06,
+      "loss": 0.9845,
+      "step": 1657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617911496046446e-06,
+      "loss": 1.2427,
+      "step": 1658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61612245288239e-06,
+      "loss": 1.3389,
+      "step": 1659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614332438513132e-06,
+      "loss": 1.3712,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612541453419427e-06,
+      "loss": 1.2925,
+      "step": 1661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610749498082291e-06,
+      "loss": 1.1868,
+      "step": 1662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608956572983003e-06,
+      "loss": 1.4446,
+      "step": 1663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607162678603097e-06,
+      "loss": 0.9769,
+      "step": 1664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605367815424373e-06,
+      "loss": 1.4166,
+      "step": 1665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603571983928888e-06,
+      "loss": 1.3952,
+      "step": 1666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.601775184598962e-06,
+      "loss": 1.3511,
+      "step": 1667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599977417917169e-06,
+      "loss": 1.1659,
+      "step": 1668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59817868436635e-06,
+      "loss": 1.0827,
+      "step": 1669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5963789844296e-06,
+      "loss": 1.2212,
+      "step": 1670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594578318590281e-06,
+      "loss": 1.282,
+      "step": 1671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592776687332003e-06,
+      "loss": 1.4087,
+      "step": 1672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590974091138645e-06,
+      "loss": 1.2047,
+      "step": 1673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58917053049434e-06,
+      "loss": 1.1262,
+      "step": 1674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587366005883485e-06,
+      "loss": 1.0491,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58556051779073e-06,
+      "loss": 1.2168,
+      "step": 1676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583754066700987e-06,
+      "loss": 1.2007,
+      "step": 1677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581946653099427e-06,
+      "loss": 1.0051,
+      "step": 1678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580138277471476e-06,
+      "loss": 1.1116,
+      "step": 1679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578328940302827e-06,
+      "loss": 1.4511,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576518642079418e-06,
+      "loss": 1.3017,
+      "step": 1681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574707383287459e-06,
+      "loss": 1.0846,
+      "step": 1682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572895164413407e-06,
+      "loss": 1.202,
+      "step": 1683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571081985943984e-06,
+      "loss": 1.3822,
+      "step": 1684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569267848366163e-06,
+      "loss": 1.1944,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567452752167183e-06,
+      "loss": 1.3403,
+      "step": 1686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565636697834531e-06,
+      "loss": 1.2227,
+      "step": 1687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563819685855963e-06,
+      "loss": 1.3406,
+      "step": 1688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562001716719481e-06,
+      "loss": 1.3641,
+      "step": 1689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560182790913349e-06,
+      "loss": 1.3278,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558362908926087e-06,
+      "loss": 1.2544,
+      "step": 1691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556542071246476e-06,
+      "loss": 1.2796,
+      "step": 1692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554720278363547e-06,
+      "loss": 1.5223,
+      "step": 1693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552897530766592e-06,
+      "loss": 1.223,
+      "step": 1694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551073828945156e-06,
+      "loss": 1.2412,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549249173389045e-06,
+      "loss": 1.0909,
+      "step": 1696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547423564588315e-06,
+      "loss": 1.3869,
+      "step": 1697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545597003033286e-06,
+      "loss": 1.2347,
+      "step": 1698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543769489214524e-06,
+      "loss": 1.593,
+      "step": 1699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54194102362286e-06,
+      "loss": 1.3777,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540111606749377e-06,
+      "loss": 1.1773,
+      "step": 1701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538281239085411e-06,
+      "loss": 1.0521,
+      "step": 1702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536449921122556e-06,
+      "loss": 1.1585,
+      "step": 1703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534617653352661e-06,
+      "loss": 1.0048,
+      "step": 1704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532784436267832e-06,
+      "loss": 1.2168,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530950270360425e-06,
+      "loss": 1.1519,
+      "step": 1706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529115156123055e-06,
+      "loss": 1.1027,
+      "step": 1707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52727909404859e-06,
+      "loss": 1.6184,
+      "step": 1708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525442084630153e-06,
+      "loss": 1.3211,
+      "step": 1709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523604128361123e-06,
+      "loss": 0.8486,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521765225735129e-06,
+      "loss": 1.1055,
+      "step": 1711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519925377246057e-06,
+      "loss": 1.3687,
+      "step": 1712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51808458338805e-06,
+      "loss": 1.2752,
+      "step": 1713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516242844655498e-06,
+      "loss": 1.2167,
+      "step": 1714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514400161543053e-06,
+      "loss": 1.366,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512556534545612e-06,
+      "loss": 1.1241,
+      "step": 1716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510711964158332e-06,
+      "loss": 1.5604,
+      "step": 1717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.50886645087662e-06,
+      "loss": 0.9319,
+      "step": 1718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.507019995196137e-06,
+      "loss": 1.5406,
+      "step": 1719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.5051725976128e-06,
+      "loss": 0.9036,
+      "step": 1720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503324258622773e-06,
+      "loss": 1.121,
+      "step": 1721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50147497872248e-06,
+      "loss": 1.1607,
+      "step": 1722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499624758408591e-06,
+      "loss": 1.22,
+      "step": 1723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497773598178033e-06,
+      "loss": 1.4798,
+      "step": 1724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495921498527982e-06,
+      "loss": 1.2152,
+      "step": 1725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494068459955871e-06,
+      "loss": 1.2422,
+      "step": 1726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49221448295938e-06,
+      "loss": 0.8487,
+      "step": 1727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490359568036446e-06,
+      "loss": 1.1773,
+      "step": 1728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488503715685254e-06,
+      "loss": 1.4826,
+      "step": 1729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486646926404243e-06,
+      "loss": 1.0892,
+      "step": 1730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4847892006921e-06,
+      "loss": 1.218,
+      "step": 1731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48293053904777e-06,
+      "loss": 1.3544,
+      "step": 1732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481070941970442e-06,
+      "loss": 1.0701,
+      "step": 1733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479210409959565e-06,
+      "loss": 1.2765,
+      "step": 1734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477348943514829e-06,
+      "loss": 1.5145,
+      "step": 1735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475486543136181e-06,
+      "loss": 1.0422,
+      "step": 1736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473623209323821e-06,
+      "loss": 1.1432,
+      "step": 1737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471758942578193e-06,
+      "loss": 1.2909,
+      "step": 1738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469893743399997e-06,
+      "loss": 1.1201,
+      "step": 1739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46802761229018e-06,
+      "loss": 1.128,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466160549749944e-06,
+      "loss": 1.3061,
+      "step": 1741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464292556280734e-06,
+      "loss": 1.143,
+      "step": 1742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46242363238425e-06,
+      "loss": 1.3287,
+      "step": 1743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46055377856244e-06,
+      "loss": 1.3558,
+      "step": 1744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458682995317508e-06,
+      "loss": 1.443,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456811283151896e-06,
+      "loss": 1.4394,
+      "step": 1746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454938642568306e-06,
+      "loss": 1.0564,
+      "step": 1747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453065074069682e-06,
+      "loss": 1.261,
+      "step": 1748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45119057815922e-06,
+      "loss": 1.0564,
+      "step": 1749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449315155340369e-06,
+      "loss": 1.1658,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447438806116821e-06,
+      "loss": 1.2753,
+      "step": 1751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44556153099252e-06,
+      "loss": 0.9946,
+      "step": 1752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443683330471656e-06,
+      "loss": 1.191,
+      "step": 1753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441804205058672e-06,
+      "loss": 1.2162,
+      "step": 1754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439924155258255e-06,
+      "loss": 1.1016,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43804318157534e-06,
+      "loss": 1.2654,
+      "step": 1756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436161284515117e-06,
+      "loss": 0.9103,
+      "step": 1757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434278464583018e-06,
+      "loss": 0.9341,
+      "step": 1758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43239472228472e-06,
+      "loss": 1.3441,
+      "step": 1759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430510058126156e-06,
+      "loss": 1.1083,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4286244726135e-06,
+      "loss": 1.241,
+      "step": 1761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426737966253176e-06,
+      "loss": 1.2819,
+      "step": 1762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424850539551857e-06,
+      "loss": 1.0295,
+      "step": 1763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422962193016459e-06,
+      "loss": 1.3524,
+      "step": 1764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.421072927154151e-06,
+      "loss": 1.0196,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41918274247234e-06,
+      "loss": 1.1986,
+      "step": 1766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41729163947869e-06,
+      "loss": 1.2615,
+      "step": 1767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415399618681101e-06,
+      "loss": 1.2558,
+      "step": 1768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41350668058773e-06,
+      "loss": 1.5233,
+      "step": 1769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411612825706976e-06,
+      "loss": 1.4585,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40971805454748e-06,
+      "loss": 1.3376,
+      "step": 1771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407822367618135e-06,
+      "loss": 1.2772,
+      "step": 1772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405925765428077e-06,
+      "loss": 1.283,
+      "step": 1773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40402824848669e-06,
+      "loss": 1.4666,
+      "step": 1774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402129817303602e-06,
+      "loss": 1.3215,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400230472388684e-06,
+      "loss": 1.0191,
+      "step": 1776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398330214252057e-06,
+      "loss": 1.1064,
+      "step": 1777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396429043404088e-06,
+      "loss": 1.165,
+      "step": 1778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394526960355385e-06,
+      "loss": 1.4908,
+      "step": 1779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.3926239656168e-06,
+      "loss": 1.4713,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390720059699438e-06,
+      "loss": 1.3948,
+      "step": 1781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388815243114637e-06,
+      "loss": 1.3377,
+      "step": 1782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386909516373989e-06,
+      "loss": 1.3668,
+      "step": 1783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385002879989328e-06,
+      "loss": 1.013,
+      "step": 1784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38309533447273e-06,
+      "loss": 1.53,
+      "step": 1785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381186880336518e-06,
+      "loss": 1.1368,
+      "step": 1786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379277518093257e-06,
+      "loss": 1.2184,
+      "step": 1787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377367248255757e-06,
+      "loss": 1.3491,
+      "step": 1788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.375456071337068e-06,
+      "loss": 1.4319,
+      "step": 1789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.373543987850494e-06,
+      "loss": 1.2831,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371630998309572e-06,
+      "loss": 1.1585,
+      "step": 1791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369717103228084e-06,
+      "loss": 1.2955,
+      "step": 1792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367802303120058e-06,
+      "loss": 1.403,
+      "step": 1793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365886598499766e-06,
+      "loss": 0.9073,
+      "step": 1794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36396998988172e-06,
+      "loss": 1.1775,
+      "step": 1795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362052477780677e-06,
+      "loss": 0.9867,
+      "step": 1796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360134062711632e-06,
+      "loss": 1.1414,
+      "step": 1797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35821474518983e-06,
+      "loss": 1.1365,
+      "step": 1798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356294525730753e-06,
+      "loss": 1.0915,
+      "step": 1799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354373404850124e-06,
+      "loss": 1.1481,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.352451383063916e-06,
+      "loss": 1.28,
+      "step": 1801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350528460888334e-06,
+      "loss": 0.8286,
+      "step": 1802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348604638839832e-06,
+      "loss": 1.3433,
+      "step": 1803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346679917435104e-06,
+      "loss": 1.2634,
+      "step": 1804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34475429719108e-06,
+      "loss": 1.3719,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342827778624943e-06,
+      "loss": 1.1138,
+      "step": 1806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340900362254106e-06,
+      "loss": 1.0858,
+      "step": 1807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33897204859623e-06,
+      "loss": 1.3527,
+      "step": 1808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337042838169211e-06,
+      "loss": 1.2388,
+      "step": 1809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335112731491192e-06,
+      "loss": 1.5342,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333181729080556e-06,
+      "loss": 1.2757,
+      "step": 1811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331249831455921e-06,
+      "loss": 1.3013,
+      "step": 1812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32931703913615e-06,
+      "loss": 1.2456,
+      "step": 1813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327383352640347e-06,
+      "loss": 1.3488,
+      "step": 1814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325448772487854e-06,
+      "loss": 1.4649,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323513299198252e-06,
+      "loss": 1.0413,
+      "step": 1816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321576933291365e-06,
+      "loss": 1.4795,
+      "step": 1817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319639675287255e-06,
+      "loss": 1.2239,
+      "step": 1818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317701525706226e-06,
+      "loss": 1.1387,
+      "step": 1819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315762485068815e-06,
+      "loss": 1.2473,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313822553895803e-06,
+      "loss": 0.8927,
+      "step": 1821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311881732708213e-06,
+      "loss": 1.3606,
+      "step": 1822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309940022027303e-06,
+      "loss": 1.0255,
+      "step": 1823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307997422374569e-06,
+      "loss": 1.2251,
+      "step": 1824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306053934271748e-06,
+      "loss": 1.0191,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304109558240817e-06,
+      "loss": 1.0287,
+      "step": 1826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302164294803986e-06,
+      "loss": 1.2023,
+      "step": 1827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300218144483709e-06,
+      "loss": 1.0806,
+      "step": 1828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298271107802677e-06,
+      "loss": 0.9177,
+      "step": 1829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296323185283816e-06,
+      "loss": 0.9617,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294374377450292e-06,
+      "loss": 1.071,
+      "step": 1831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292424684825514e-06,
+      "loss": 0.9883,
+      "step": 1832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290474107933116e-06,
+      "loss": 1.2767,
+      "step": 1833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28852264729698e-06,
+      "loss": 1.2299,
+      "step": 1834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286570303441223e-06,
+      "loss": 1.2329,
+      "step": 1835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284617076890199e-06,
+      "loss": 1.1056,
+      "step": 1836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282662968168495e-06,
+      "loss": 1.4812,
+      "step": 1837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280707977800944e-06,
+      "loss": 1.6674,
+      "step": 1838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278752106312606e-06,
+      "loss": 1.4291,
+      "step": 1839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276795354228785e-06,
+      "loss": 1.4803,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274837722075015e-06,
+      "loss": 0.9179,
+      "step": 1841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272879210377074e-06,
+      "loss": 1.2922,
+      "step": 1842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270919819660967e-06,
+      "loss": 1.1382,
+      "step": 1843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268959550452946e-06,
+      "loss": 1.0215,
+      "step": 1844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266998403279486e-06,
+      "loss": 1.3928,
+      "step": 1845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265036378667312e-06,
+      "loss": 1.2983,
+      "step": 1846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263073477143375e-06,
+      "loss": 0.9877,
+      "step": 1847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261109699234862e-06,
+      "loss": 0.9742,
+      "step": 1848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259145045469198e-06,
+      "loss": 1.5755,
+      "step": 1849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257179516374045e-06,
+      "loss": 1.0962,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255213112477296e-06,
+      "loss": 1.3488,
+      "step": 1851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253245834307079e-06,
+      "loss": 1.1433,
+      "step": 1852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251277682391762e-06,
+      "loss": 1.2243,
+      "step": 1853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249308657259943e-06,
+      "loss": 1.1109,
+      "step": 1854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247338759440456e-06,
+      "loss": 1.1052,
+      "step": 1855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245367989462368e-06,
+      "loss": 1.0862,
+      "step": 1856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24339634785498e-06,
+      "loss": 1.1922,
+      "step": 1857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241423835147833e-06,
+      "loss": 1.409,
+      "step": 1858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239450451870693e-06,
+      "loss": 1.285,
+      "step": 1859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237476198553567e-06,
+      "loss": 0.9902,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23550107572669e-06,
+      "loss": 1.0957,
+      "step": 1861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233525083920536e-06,
+      "loss": 1.0792,
+      "step": 1862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23154822366581e-06,
+      "loss": 1.4405,
+      "step": 1863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229570495493447e-06,
+      "loss": 1.2072,
+      "step": 1864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227591899934621e-06,
+      "loss": 1.3656,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225612437520736e-06,
+      "loss": 1.28,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223632108783428e-06,
+      "loss": 1.2186,
+      "step": 1867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221650914254566e-06,
+      "loss": 1.4985,
+      "step": 1868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219668854466254e-06,
+      "loss": 1.6419,
+      "step": 1869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217685929950823e-06,
+      "loss": 0.9819,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215702141240844e-06,
+      "loss": 1.0872,
+      "step": 1871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213717488869113e-06,
+      "loss": 0.9481,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211731973368665e-06,
+      "loss": 1.5949,
+      "step": 1873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209745595272755e-06,
+      "loss": 1.1179,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207758355114883e-06,
+      "loss": 1.5313,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205770253428775e-06,
+      "loss": 1.1437,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203781290748385e-06,
+      "loss": 1.0542,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201791467607905e-06,
+      "loss": 1.37,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199800784541752e-06,
+      "loss": 1.0513,
+      "step": 1879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197809242084575e-06,
+      "loss": 1.2766,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.195816840771262e-06,
+      "loss": 1.2273,
+      "step": 1881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193823581136919e-06,
+      "loss": 1.1538,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191829463716889e-06,
+      "loss": 0.9727,
+      "step": 1883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189834489046746e-06,
+      "loss": 1.0768,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187838657662295e-06,
+      "loss": 1.1965,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185841970099566e-06,
+      "loss": 1.1318,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183844426894825e-06,
+      "loss": 1.2806,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181846028584563e-06,
+      "loss": 1.2986,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179846775705504e-06,
+      "loss": 1.1201,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177846668794598e-06,
+      "loss": 0.8564,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17584570838903e-06,
+      "loss": 1.2629,
+      "step": 1891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173843895026207e-06,
+      "loss": 1.217,
+      "step": 1892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17184122924377e-06,
+      "loss": 1.2401,
+      "step": 1893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169837711579591e-06,
+      "loss": 1.008,
+      "step": 1894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167833342571764e-06,
+      "loss": 0.8571,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165828122758615e-06,
+      "loss": 1.0159,
+      "step": 1896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163822052678703e-06,
+      "loss": 1.2974,
+      "step": 1897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161815132870806e-06,
+      "loss": 1.2699,
+      "step": 1898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159807363873938e-06,
+      "loss": 1.3614,
+      "step": 1899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157798746227337e-06,
+      "loss": 1.1172,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15578928047047e-06,
+      "loss": 1.0937,
+      "step": 1901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153778967143035e-06,
+      "loss": 1.0979,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151767806784954e-06,
+      "loss": 0.8977,
+      "step": 1903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149755799936377e-06,
+      "loss": 1.2645,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147742947137678e-06,
+      "loss": 1.4044,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145729248929466e-06,
+      "loss": 1.2127,
+      "step": 1906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143714705852572e-06,
+      "loss": 1.1032,
+      "step": 1907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141699318448053e-06,
+      "loss": 1.3764,
+      "step": 1908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139683087257197e-06,
+      "loss": 1.1261,
+      "step": 1909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137666012821514e-06,
+      "loss": 1.0632,
+      "step": 1910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135648095682744e-06,
+      "loss": 1.406,
+      "step": 1911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13362933638285e-06,
+      "loss": 1.3376,
+      "step": 1912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131609735464027e-06,
+      "loss": 1.0672,
+      "step": 1913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129589293468689e-06,
+      "loss": 0.9565,
+      "step": 1914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12756801093948e-06,
+      "loss": 1.3258,
+      "step": 1915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125545888419269e-06,
+      "loss": 1.0856,
+      "step": 1916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123522926451149e-06,
+      "loss": 1.025,
+      "step": 1917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12149912557844e-06,
+      "loss": 1.0979,
+      "step": 1918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119474486344692e-06,
+      "loss": 1.1706,
+      "step": 1919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117449009293668e-06,
+      "loss": 1.0496,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115422694969368e-06,
+      "loss": 1.0401,
+      "step": 1921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113395543916012e-06,
+      "loss": 1.1137,
+      "step": 1922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111367556678042e-06,
+      "loss": 1.0561,
+      "step": 1923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.109338733800132e-06,
+      "loss": 1.0336,
+      "step": 1924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107309075827173e-06,
+      "loss": 1.1795,
+      "step": 1925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10527858330428e-06,
+      "loss": 1.274,
+      "step": 1926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103247256776804e-06,
+      "loss": 1.4358,
+      "step": 1927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101215096790305e-06,
+      "loss": 1.1433,
+      "step": 1928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099182103890574e-06,
+      "loss": 1.562,
+      "step": 1929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097148278623628e-06,
+      "loss": 1.2298,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.095113621535701e-06,
+      "loss": 1.1157,
+      "step": 1931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093078133173256e-06,
+      "loss": 1.0051,
+      "step": 1932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091041814082976e-06,
+      "loss": 1.1521,
+      "step": 1933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089004664811767e-06,
+      "loss": 1.1935,
+      "step": 1934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.086966685906763e-06,
+      "loss": 1.1553,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084927877915314e-06,
+      "loss": 1.0472,
+      "step": 1936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082888241384997e-06,
+      "loss": 1.103,
+      "step": 1937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080847776863609e-06,
+      "loss": 1.1667,
+      "step": 1938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078806484899174e-06,
+      "loss": 1.2402,
+      "step": 1939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07676436603993e-06,
+      "loss": 1.1888,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074721420834345e-06,
+      "loss": 1.1579,
+      "step": 1941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072677649831107e-06,
+      "loss": 1.0757,
+      "step": 1942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.070633053579122e-06,
+      "loss": 0.8852,
+      "step": 1943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068587632627521e-06,
+      "loss": 1.0428,
+      "step": 1944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.066541387525656e-06,
+      "loss": 1.2288,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064494318823102e-06,
+      "loss": 0.918,
+      "step": 1946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06244642706965e-06,
+      "loss": 1.2302,
+      "step": 1947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060397712815318e-06,
+      "loss": 1.1076,
+      "step": 1948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05834817661034e-06,
+      "loss": 1.2724,
+      "step": 1949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056297819005177e-06,
+      "loss": 1.2267,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054246640550503e-06,
+      "loss": 0.9392,
+      "step": 1951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052194641797217e-06,
+      "loss": 1.227,
+      "step": 1952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05014182329644e-06,
+      "loss": 1.0808,
+      "step": 1953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048088185599507e-06,
+      "loss": 1.484,
+      "step": 1954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046033729257979e-06,
+      "loss": 1.282,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043978454823632e-06,
+      "loss": 1.5381,
+      "step": 1956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041922362848467e-06,
+      "loss": 1.3869,
+      "step": 1957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.0398654538847e-06,
+      "loss": 1.6376,
+      "step": 1958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03780772848477e-06,
+      "loss": 1.3241,
+      "step": 1959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035749187201333e-06,
+      "loss": 1.2288,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033689830587263e-06,
+      "loss": 1.2811,
+      "step": 1961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031629659195657e-06,
+      "loss": 1.0349,
+      "step": 1962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029568673579827e-06,
+      "loss": 1.4763,
+      "step": 1963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027506874293304e-06,
+      "loss": 1.1456,
+      "step": 1964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02544426188984e-06,
+      "loss": 1.0346,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023380836923404e-06,
+      "loss": 1.2293,
+      "step": 1966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021316599948184e-06,
+      "loss": 1.0495,
+      "step": 1967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019251551518585e-06,
+      "loss": 0.9419,
+      "step": 1968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01718569218923e-06,
+      "loss": 1.1997,
+      "step": 1969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015119022514958e-06,
+      "loss": 0.9873,
+      "step": 1970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013051543050833e-06,
+      "loss": 1.2452,
+      "step": 1971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010983254352127e-06,
+      "loss": 1.3444,
+      "step": 1972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008914156974333e-06,
+      "loss": 1.0313,
+      "step": 1973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006844251473165e-06,
+      "loss": 1.2699,
+      "step": 1974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00477353840455e-06,
+      "loss": 1.4475,
+      "step": 1975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002702018324629e-06,
+      "loss": 1.2072,
+      "step": 1976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000629691789769e-06,
+      "loss": 1.2382,
+      "step": 1977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998556559356543e-06,
+      "loss": 1.1326,
+      "step": 1978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99648262158175e-06,
+      "loss": 1.1764,
+      "step": 1979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994407879022397e-06,
+      "loss": 0.9394,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992332332235711e-06,
+      "loss": 1.3708,
+      "step": 1981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990255981779139e-06,
+      "loss": 1.3066,
+      "step": 1982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988178828210336e-06,
+      "loss": 1.3636,
+      "step": 1983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.986100872087177e-06,
+      "loss": 1.2011,
+      "step": 1984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984022113967755e-06,
+      "loss": 1.0355,
+      "step": 1985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981942554410371e-06,
+      "loss": 1.2436,
+      "step": 1986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97986219397355e-06,
+      "loss": 1.0529,
+      "step": 1987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97778103321602e-06,
+      "loss": 1.3639,
+      "step": 1988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975699072696743e-06,
+      "loss": 1.4262,
+      "step": 1989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973616312974876e-06,
+      "loss": 1.2859,
+      "step": 1990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971532754609802e-06,
+      "loss": 1.1142,
+      "step": 1991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969448398161115e-06,
+      "loss": 1.2586,
+      "step": 1992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.967363244188624e-06,
+      "loss": 1.099,
+      "step": 1993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965277293252354e-06,
+      "loss": 1.274,
+      "step": 1994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96319054591254e-06,
+      "loss": 1.2425,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961103002729634e-06,
+      "loss": 1.0504,
+      "step": 1996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.959014664264299e-06,
+      "loss": 1.206,
+      "step": 1997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956925531077417e-06,
+      "loss": 0.9757,
+      "step": 1998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.954835603730078e-06,
+      "loss": 1.2309,
+      "step": 1999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952744882783587e-06,
+      "loss": 1.2319,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950653368799461e-06,
+      "loss": 1.2157,
+      "step": 2001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948561062339435e-06,
+      "loss": 1.2976,
+      "step": 2002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946467963965452e-06,
+      "loss": 1.1876,
+      "step": 2003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944374074239665e-06,
+      "loss": 0.9,
+      "step": 2004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942279393724447e-06,
+      "loss": 1.3977,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940183922982381e-06,
+      "loss": 0.8775,
+      "step": 2006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938087662576258e-06,
+      "loss": 1.1101,
+      "step": 2007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935990613069087e-06,
+      "loss": 1.1303,
+      "step": 2008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933892775024081e-06,
+      "loss": 1.3,
+      "step": 2009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931794149004675e-06,
+      "loss": 1.1126,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929694735574509e-06,
+      "loss": 1.2654,
+      "step": 2011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927594535297433e-06,
+      "loss": 0.9186,
+      "step": 2012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925493548737518e-06,
+      "loss": 1.1121,
+      "step": 2013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923391776459031e-06,
+      "loss": 1.1506,
+      "step": 2014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921289219026465e-06,
+      "loss": 0.9677,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919185877004515e-06,
+      "loss": 0.7232,
+      "step": 2016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.917081750958088e-06,
+      "loss": 1.1162,
+      "step": 2017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914976841452304e-06,
+      "loss": 1.1516,
+      "step": 2018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912871149052492e-06,
+      "loss": 1.1818,
+      "step": 2019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91076467432419e-06,
+      "loss": 1.2832,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90865741783315e-06,
+      "loss": 1.4431,
+      "step": 2021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90654938014533e-06,
+      "loss": 1.1553,
+      "step": 2022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904440561826897e-06,
+      "loss": 1.188,
+      "step": 2023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902330963444234e-06,
+      "loss": 1.2871,
+      "step": 2024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.900220585563926e-06,
+      "loss": 1.2319,
+      "step": 2025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898109428752773e-06,
+      "loss": 1.0131,
+      "step": 2026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895997493577781e-06,
+      "loss": 1.3481,
+      "step": 2027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893884780606164e-06,
+      "loss": 0.9148,
+      "step": 2028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891771290405351e-06,
+      "loss": 1.4167,
+      "step": 2029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889657023542973e-06,
+      "loss": 1.0609,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887541980586872e-06,
+      "loss": 1.1718,
+      "step": 2031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885426162105101e-06,
+      "loss": 1.3305,
+      "step": 2032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883309568665914e-06,
+      "loss": 1.4511,
+      "step": 2033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881192200837785e-06,
+      "loss": 1.2485,
+      "step": 2034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879074059189383e-06,
+      "loss": 1.2667,
+      "step": 2035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876955144289594e-06,
+      "loss": 1.0036,
+      "step": 2036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874835456707507e-06,
+      "loss": 1.1842,
+      "step": 2037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872714997012421e-06,
+      "loss": 1.224,
+      "step": 2038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870593765773842e-06,
+      "loss": 1.1051,
+      "step": 2039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868471763561482e-06,
+      "loss": 1.2916,
+      "step": 2040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86634899094526e-06,
+      "loss": 1.1447,
+      "step": 2041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.864225448495304e-06,
+      "loss": 1.1208,
+      "step": 2042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862101136781948e-06,
+      "loss": 1.3119,
+      "step": 2043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85997605637573e-06,
+      "loss": 1.1147,
+      "step": 2044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.857850207847397e-06,
+      "loss": 0.9394,
+      "step": 2045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855723591767903e-06,
+      "loss": 1.2234,
+      "step": 2046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853596208708404e-06,
+      "loss": 1.0905,
+      "step": 2047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85146805924027e-06,
+      "loss": 1.272,
+      "step": 2048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849339143935068e-06,
+      "loss": 0.7774,
+      "step": 2049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847209463364574e-06,
+      "loss": 1.2043,
+      "step": 2050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84507901810077e-06,
+      "loss": 1.2879,
+      "step": 2051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842947808715848e-06,
+      "loss": 1.1781,
+      "step": 2052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840815835782194e-06,
+      "loss": 1.0075,
+      "step": 2053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83868309987241e-06,
+      "loss": 1.2761,
+      "step": 2054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836549601559297e-06,
+      "loss": 1.0645,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834415341415862e-06,
+      "loss": 1.1682,
+      "step": 2056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83228032001532e-06,
+      "loss": 1.1766,
+      "step": 2057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830144537931082e-06,
+      "loss": 1.2819,
+      "step": 2058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828007995736774e-06,
+      "loss": 1.3992,
+      "step": 2059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825870694006217e-06,
+      "loss": 0.833,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823732633313441e-06,
+      "loss": 1.2932,
+      "step": 2061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82159381423268e-06,
+      "loss": 1.0428,
+      "step": 2062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.819454237338369e-06,
+      "loss": 1.4055,
+      "step": 2063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817313903205148e-06,
+      "loss": 1.1532,
+      "step": 2064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815172812407861e-06,
+      "loss": 1.0564,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813030965521554e-06,
+      "loss": 1.1619,
+      "step": 2066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810888363121477e-06,
+      "loss": 1.2849,
+      "step": 2067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80874500578308e-06,
+      "loss": 1.11,
+      "step": 2068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806600894082022e-06,
+      "loss": 1.1056,
+      "step": 2069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804456028594158e-06,
+      "loss": 1.1726,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80231040989555e-06,
+      "loss": 1.0745,
+      "step": 2071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80016403856246e-06,
+      "loss": 1.0943,
+      "step": 2072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798016915171353e-06,
+      "loss": 1.0216,
+      "step": 2073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795869040298895e-06,
+      "loss": 0.9484,
+      "step": 2074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793720414521955e-06,
+      "loss": 1.1281,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791571038417602e-06,
+      "loss": 1.0318,
+      "step": 2076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789420912563111e-06,
+      "loss": 1.3451,
+      "step": 2077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78727003753595e-06,
+      "loss": 1.3631,
+      "step": 2078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785118413913799e-06,
+      "loss": 1.0171,
+      "step": 2079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.782966042274529e-06,
+      "loss": 1.1125,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78081292319622e-06,
+      "loss": 1.3542,
+      "step": 2081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778659057257144e-06,
+      "loss": 1.2511,
+      "step": 2082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776504445035781e-06,
+      "loss": 1.293,
+      "step": 2083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774349087110813e-06,
+      "loss": 0.9166,
+      "step": 2084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772192984061112e-06,
+      "loss": 0.9834,
+      "step": 2085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77003613646576e-06,
+      "loss": 1.0139,
+      "step": 2086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.767878544904036e-06,
+      "loss": 1.0096,
+      "step": 2087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765720209955414e-06,
+      "loss": 1.3568,
+      "step": 2088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763561132199576e-06,
+      "loss": 0.9558,
+      "step": 2089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761401312216398e-06,
+      "loss": 0.8533,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759240750585957e-06,
+      "loss": 1.0864,
+      "step": 2091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757079447888529e-06,
+      "loss": 1.143,
+      "step": 2092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754917404704588e-06,
+      "loss": 1.4297,
+      "step": 2093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752754621614807e-06,
+      "loss": 0.9873,
+      "step": 2094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75059109920006e-06,
+      "loss": 1.3792,
+      "step": 2095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748426838041421e-06,
+      "loss": 1.1497,
+      "step": 2096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746261838720153e-06,
+      "loss": 1.107,
+      "step": 2097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744096101817731e-06,
+      "loss": 1.0386,
+      "step": 2098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741929627915814e-06,
+      "loss": 1.1931,
+      "step": 2099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73976241759627e-06,
+      "loss": 0.9805,
+      "step": 2100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73759447144116e-06,
+      "loss": 1.0927,
+      "step": 2101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73542579003274e-06,
+      "loss": 1.1926,
+      "step": 2102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733256373953471e-06,
+      "loss": 1.1355,
+      "step": 2103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731086223786006e-06,
+      "loss": 1.0814,
+      "step": 2104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.728915340113194e-06,
+      "loss": 1.1503,
+      "step": 2105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726743723518087e-06,
+      "loss": 1.1329,
+      "step": 2106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724571374583925e-06,
+      "loss": 1.3128,
+      "step": 2107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.722398293894153e-06,
+      "loss": 1.2663,
+      "step": 2108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72022448203241e-06,
+      "loss": 1.0992,
+      "step": 2109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718049939582529e-06,
+      "loss": 1.1523,
+      "step": 2110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71587466712854e-06,
+      "loss": 1.1145,
+      "step": 2111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713698665254669e-06,
+      "loss": 1.4882,
+      "step": 2112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711521934545343e-06,
+      "loss": 1.1803,
+      "step": 2113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70934447558518e-06,
+      "loss": 1.533,
+      "step": 2114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707166288958986e-06,
+      "loss": 1.1063,
+      "step": 2115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704987375251782e-06,
+      "loss": 1.3866,
+      "step": 2116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702807735048767e-06,
+      "loss": 1.0282,
+      "step": 2117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70062736893534e-06,
+      "loss": 0.9521,
+      "step": 2118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698446277497097e-06,
+      "loss": 1.0773,
+      "step": 2119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.696264461319831e-06,
+      "loss": 1.3614,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694081920989523e-06,
+      "loss": 1.1788,
+      "step": 2121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69189865709235e-06,
+      "loss": 1.0553,
+      "step": 2122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68971467021469e-06,
+      "loss": 1.1697,
+      "step": 2123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687529960943107e-06,
+      "loss": 0.8097,
+      "step": 2124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685344529864365e-06,
+      "loss": 1.161,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683158377565415e-06,
+      "loss": 1.3094,
+      "step": 2126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68097150463341e-06,
+      "loss": 1.5048,
+      "step": 2127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678783911655691e-06,
+      "loss": 1.4324,
+      "step": 2128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.676595599219794e-06,
+      "loss": 1.3312,
+      "step": 2129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674406567913447e-06,
+      "loss": 1.0739,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.672216818324574e-06,
+      "loss": 1.2828,
+      "step": 2131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67002635104129e-06,
+      "loss": 1.4752,
+      "step": 2132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.6678351666519e-06,
+      "loss": 1.1378,
+      "step": 2133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66564326574491e-06,
+      "loss": 1.0178,
+      "step": 2134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663450648909005e-06,
+      "loss": 0.8929,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661257316733078e-06,
+      "loss": 1.1675,
+      "step": 2136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659063269806202e-06,
+      "loss": 1.0494,
+      "step": 2137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656868508717648e-06,
+      "loss": 1.2702,
+      "step": 2138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654673034056875e-06,
+      "loss": 1.0256,
+      "step": 2139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652476846413537e-06,
+      "loss": 1.2183,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650279946377479e-06,
+      "loss": 0.9372,
+      "step": 2141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648082334538735e-06,
+      "loss": 1.1857,
+      "step": 2142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.645884011487534e-06,
+      "loss": 1.1845,
+      "step": 2143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64368497781429e-06,
+      "loss": 1.1847,
+      "step": 2144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641485234109615e-06,
+      "loss": 1.2877,
+      "step": 2145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639284780964307e-06,
+      "loss": 1.264,
+      "step": 2146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637083618969357e-06,
+      "loss": 1.4167,
+      "step": 2147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634881748715941e-06,
+      "loss": 1.1707,
+      "step": 2148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632679170795434e-06,
+      "loss": 1.4024,
+      "step": 2149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630475885799395e-06,
+      "loss": 1.1947,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628271894319575e-06,
+      "loss": 1.0126,
+      "step": 2151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626067196947913e-06,
+      "loss": 1.1364,
+      "step": 2152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623861794276539e-06,
+      "loss": 1.0265,
+      "step": 2153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621655686897771e-06,
+      "loss": 1.0953,
+      "step": 2154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619448875404121e-06,
+      "loss": 1.0266,
+      "step": 2155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.617241360388282e-06,
+      "loss": 1.2697,
+      "step": 2156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.615033142443143e-06,
+      "loss": 1.0968,
+      "step": 2157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612824222161781e-06,
+      "loss": 1.7315,
+      "step": 2158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610614600137456e-06,
+      "loss": 1.3581,
+      "step": 2159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608404276963623e-06,
+      "loss": 1.185,
+      "step": 2160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606193253233921e-06,
+      "loss": 1.2023,
+      "step": 2161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60398152954218e-06,
+      "loss": 0.957,
+      "step": 2162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601769106482418e-06,
+      "loss": 1.206,
+      "step": 2163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599555984648836e-06,
+      "loss": 1.3328,
+      "step": 2164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597342164635828e-06,
+      "loss": 1.3606,
+      "step": 2165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595127647037976e-06,
+      "loss": 1.0942,
+      "step": 2166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5929124324500455e-06,
+      "loss": 0.9441,
+      "step": 2167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590696521466992e-06,
+      "loss": 1.3045,
+      "step": 2168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588479914683954e-06,
+      "loss": 1.2177,
+      "step": 2169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586262612696263e-06,
+      "loss": 1.0841,
+      "step": 2170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5840446160994304e-06,
+      "loss": 1.1894,
+      "step": 2171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5818259254891614e-06,
+      "loss": 1.1739,
+      "step": 2172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579606541461344e-06,
+      "loss": 1.0031,
+      "step": 2173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577386464612049e-06,
+      "loss": 0.8777,
+      "step": 2174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5751656955375395e-06,
+      "loss": 1.3847,
+      "step": 2175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572944234834261e-06,
+      "loss": 1.0308,
+      "step": 2176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5707220830988445e-06,
+      "loss": 1.1061,
+      "step": 2177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568499240928109e-06,
+      "loss": 1.0808,
+      "step": 2178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566275708919057e-06,
+      "loss": 1.0068,
+      "step": 2179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5640514876688765e-06,
+      "loss": 1.2095,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561826577774942e-06,
+      "loss": 1.1951,
+      "step": 2181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559600979834809e-06,
+      "loss": 1.0647,
+      "step": 2182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.557374694446223e-06,
+      "loss": 1.3693,
+      "step": 2183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555147722207111e-06,
+      "loss": 1.1151,
+      "step": 2184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.552920063715585e-06,
+      "loss": 1.0637,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.550691719569944e-06,
+      "loss": 1.1645,
+      "step": 2186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548462690368665e-06,
+      "loss": 1.2795,
+      "step": 2187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546232976710413e-06,
+      "loss": 1.1517,
+      "step": 2188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544002579194041e-06,
+      "loss": 1.1376,
+      "step": 2189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541771498418575e-06,
+      "loss": 1.0132,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539539734983236e-06,
+      "loss": 1.2238,
+      "step": 2191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537307289487419e-06,
+      "loss": 1.1545,
+      "step": 2192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535074162530707e-06,
+      "loss": 1.1984,
+      "step": 2193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532840354712868e-06,
+      "loss": 1.0868,
+      "step": 2194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530605866633847e-06,
+      "loss": 1.1244,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5283706988937765e-06,
+      "loss": 1.3135,
+      "step": 2196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52613485209297e-06,
+      "loss": 1.2042,
+      "step": 2197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523898326831921e-06,
+      "loss": 1.0939,
+      "step": 2198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52166112371131e-06,
+      "loss": 1.0415,
+      "step": 2199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5194232433319955e-06,
+      "loss": 1.0923,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51718468629502e-06,
+      "loss": 1.3002,
+      "step": 2201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514945453201608e-06,
+      "loss": 1.3629,
+      "step": 2202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.512705544653161e-06,
+      "loss": 1.0718,
+      "step": 2203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510464961251271e-06,
+      "loss": 1.3819,
+      "step": 2204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5082237035977025e-06,
+      "loss": 1.1029,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505981772294404e-06,
+      "loss": 0.8424,
+      "step": 2206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.503739167943507e-06,
+      "loss": 1.4119,
+      "step": 2207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501495891147322e-06,
+      "loss": 1.1449,
+      "step": 2208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.499251942508339e-06,
+      "loss": 1.4632,
+      "step": 2209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497007322629231e-06,
+      "loss": 1.3658,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494762032112848e-06,
+      "loss": 1.3503,
+      "step": 2211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492516071562226e-06,
+      "loss": 1.3841,
+      "step": 2212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490269441580573e-06,
+      "loss": 1.1761,
+      "step": 2213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488022142771282e-06,
+      "loss": 1.0855,
+      "step": 2214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485774175737925e-06,
+      "loss": 1.0136,
+      "step": 2215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483525541084253e-06,
+      "loss": 1.2624,
+      "step": 2216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481276239414197e-06,
+      "loss": 1.3222,
+      "step": 2217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.479026271331864e-06,
+      "loss": 1.1834,
+      "step": 2218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4767756374415445e-06,
+      "loss": 0.856,
+      "step": 2219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4745243383477055e-06,
+      "loss": 1.1125,
+      "step": 2220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472272374654991e-06,
+      "loss": 1.0532,
+      "step": 2221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470019746968226e-06,
+      "loss": 1.1771,
+      "step": 2222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.467766455892416e-06,
+      "loss": 1.2954,
+      "step": 2223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4655125020327376e-06,
+      "loss": 1.249,
+      "step": 2224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.463257885994553e-06,
+      "loss": 1.2391,
+      "step": 2225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461002608383396e-06,
+      "loss": 1.1693,
+      "step": 2226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.458746669804984e-06,
+      "loss": 0.98,
+      "step": 2227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456490070865206e-06,
+      "loss": 1.4,
+      "step": 2228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454232812170135e-06,
+      "loss": 1.3389,
+      "step": 2229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4519748943260126e-06,
+      "loss": 1.0345,
+      "step": 2230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.449716317939266e-06,
+      "loss": 1.2256,
+      "step": 2231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447457083616494e-06,
+      "loss": 1.1825,
+      "step": 2232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445197191964474e-06,
+      "loss": 1.2191,
+      "step": 2233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44293664359016e-06,
+      "loss": 0.9501,
+      "step": 2234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440675439100682e-06,
+      "loss": 1.1634,
+      "step": 2235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438413579103344e-06,
+      "loss": 0.8378,
+      "step": 2236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43615106420563e-06,
+      "loss": 1.1283,
+      "step": 2237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433887895015199e-06,
+      "loss": 1.3547,
+      "step": 2238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431624072139884e-06,
+      "loss": 0.9441,
+      "step": 2239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.429359596187694e-06,
+      "loss": 0.9472,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4270944677668155e-06,
+      "loss": 1.1518,
+      "step": 2241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424828687485606e-06,
+      "loss": 1.2154,
+      "step": 2242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422562255952604e-06,
+      "loss": 1.1391,
+      "step": 2243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420295173776515e-06,
+      "loss": 1.2227,
+      "step": 2244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418027441566227e-06,
+      "loss": 1.1814,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415759059930799e-06,
+      "loss": 1.1511,
+      "step": 2246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413490029479464e-06,
+      "loss": 0.9636,
+      "step": 2247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411220350821631e-06,
+      "loss": 1.1352,
+      "step": 2248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.408950024566881e-06,
+      "loss": 1.3687,
+      "step": 2249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.406679051324972e-06,
+      "loss": 1.1154,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.404407431705833e-06,
+      "loss": 1.1844,
+      "step": 2251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402135166319567e-06,
+      "loss": 0.9242,
+      "step": 2252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399862255776449e-06,
+      "loss": 1.2831,
+      "step": 2253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397588700686933e-06,
+      "loss": 1.2665,
+      "step": 2254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395314501661639e-06,
+      "loss": 1.4586,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393039659311366e-06,
+      "loss": 1.0681,
+      "step": 2256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3907641742470815e-06,
+      "loss": 1.3012,
+      "step": 2257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388488047079927e-06,
+      "loss": 1.2348,
+      "step": 2258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386211278421217e-06,
+      "loss": 1.015,
+      "step": 2259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383933868882438e-06,
+      "loss": 1.2557,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.381655819075249e-06,
+      "loss": 1.0186,
+      "step": 2261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379377129611478e-06,
+      "loss": 1.1969,
+      "step": 2262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377097801103129e-06,
+      "loss": 1.135,
+      "step": 2263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374817834162378e-06,
+      "loss": 0.9644,
+      "step": 2264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372537229401565e-06,
+      "loss": 1.4496,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3702559874332125e-06,
+      "loss": 1.0887,
+      "step": 2266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367974108870003e-06,
+      "loss": 1.3321,
+      "step": 2267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3656915943247984e-06,
+      "loss": 1.2338,
+      "step": 2268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363408444410629e-06,
+      "loss": 0.9731,
+      "step": 2269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3611246597406925e-06,
+      "loss": 1.0954,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3588402409283625e-06,
+      "loss": 1.1108,
+      "step": 2271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356555188587178e-06,
+      "loss": 1.1147,
+      "step": 2272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354269503330849e-06,
+      "loss": 1.2188,
+      "step": 2273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351983185773259e-06,
+      "loss": 1.0709,
+      "step": 2274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349696236528458e-06,
+      "loss": 1.4559,
+      "step": 2275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347408656210666e-06,
+      "loss": 1.0845,
+      "step": 2276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.345120445434275e-06,
+      "loss": 1.3354,
+      "step": 2277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342831604813844e-06,
+      "loss": 1.485,
+      "step": 2278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3405421349641e-06,
+      "loss": 1.0551,
+      "step": 2279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338252036499941e-06,
+      "loss": 1.0369,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3359613100364345e-06,
+      "loss": 1.2566,
+      "step": 2281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333669956188815e-06,
+      "loss": 1.2311,
+      "step": 2282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331377975572486e-06,
+      "loss": 1.0912,
+      "step": 2283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3290853688030196e-06,
+      "loss": 1.1679,
+      "step": 2284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326792136496157e-06,
+      "loss": 1.3485,
+      "step": 2285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324498279267803e-06,
+      "loss": 0.8295,
+      "step": 2286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322203797734038e-06,
+      "loss": 1.4665,
+      "step": 2287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319908692511103e-06,
+      "loss": 1.2184,
+      "step": 2288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317612964215408e-06,
+      "loss": 1.1884,
+      "step": 2289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315316613463535e-06,
+      "loss": 1.5718,
+      "step": 2290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3130196408722265e-06,
+      "loss": 1.2075,
+      "step": 2291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310722047058396e-06,
+      "loss": 1.3012,
+      "step": 2292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308423832639125e-06,
+      "loss": 1.2377,
+      "step": 2293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306124998231655e-06,
+      "loss": 1.275,
+      "step": 2294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303825544453405e-06,
+      "loss": 1.1195,
+      "step": 2295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301525471921949e-06,
+      "loss": 1.0071,
+      "step": 2296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2992247812550335e-06,
+      "loss": 0.9832,
+      "step": 2297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296923473070571e-06,
+      "loss": 1.2111,
+      "step": 2298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2946215479866375e-06,
+      "loss": 1.1063,
+      "step": 2299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292319006621477e-06,
+      "loss": 1.2826,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2900158495934955e-06,
+      "loss": 1.2149,
+      "step": 2301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2877120775212685e-06,
+      "loss": 1.2204,
+      "step": 2302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285407691023536e-06,
+      "loss": 1.096,
+      "step": 2303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283102690719198e-06,
+      "loss": 0.9988,
+      "step": 2304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.280797077227328e-06,
+      "loss": 1.0751,
+      "step": 2305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278490851167155e-06,
+      "loss": 1.2458,
+      "step": 2306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276184013158079e-06,
+      "loss": 1.109,
+      "step": 2307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2738765638196625e-06,
+      "loss": 1.2614,
+      "step": 2308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271568503771632e-06,
+      "loss": 1.3032,
+      "step": 2309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269259833633877e-06,
+      "loss": 0.9006,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266950554026453e-06,
+      "loss": 1.1741,
+      "step": 2311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264640665569577e-06,
+      "loss": 1.3564,
+      "step": 2312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262330168883631e-06,
+      "loss": 1.1983,
+      "step": 2313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26001906458916e-06,
+      "loss": 1.4665,
+      "step": 2314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257707353306873e-06,
+      "loss": 0.9153,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255395035657639e-06,
+      "loss": 1.0543,
+      "step": 2316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253082112262494e-06,
+      "loss": 1.3319,
+      "step": 2317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.250768583742634e-06,
+      "loss": 1.0657,
+      "step": 2318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248454450719416e-06,
+      "loss": 0.977,
+      "step": 2319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246139713814365e-06,
+      "loss": 1.4327,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2438243736491646e-06,
+      "loss": 1.2737,
+      "step": 2321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241508430845656e-06,
+      "loss": 1.2005,
+      "step": 2322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239191886025854e-06,
+      "loss": 1.1725,
+      "step": 2323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236874739811921e-06,
+      "loss": 1.2275,
+      "step": 2324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234556992826192e-06,
+      "loss": 1.0092,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232238645691157e-06,
+      "loss": 1.3079,
+      "step": 2326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229919699029472e-06,
+      "loss": 1.0473,
+      "step": 2327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227600153463947e-06,
+      "loss": 1.2638,
+      "step": 2328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.225280009617563e-06,
+      "loss": 1.0689,
+      "step": 2329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.222959268113452e-06,
+      "loss": 1.1313,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2206379295749115e-06,
+      "loss": 1.0656,
+      "step": 2331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218315994625397e-06,
+      "loss": 0.875,
+      "step": 2332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215993463888526e-06,
+      "loss": 1.4941,
+      "step": 2333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213670337988079e-06,
+      "loss": 1.2223,
+      "step": 2334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211346617547987e-06,
+      "loss": 1.2894,
+      "step": 2335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209022303192351e-06,
+      "loss": 1.0399,
+      "step": 2336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206697395545425e-06,
+      "loss": 1.0961,
+      "step": 2337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.204371895231623e-06,
+      "loss": 1.264,
+      "step": 2338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2020458028755235e-06,
+      "loss": 0.9762,
+      "step": 2339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199719119101858e-06,
+      "loss": 1.0987,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1973918445355175e-06,
+      "loss": 0.8767,
+      "step": 2341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195063979801554e-06,
+      "loss": 1.2217,
+      "step": 2342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192735525525179e-06,
+      "loss": 1.2255,
+      "step": 2343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.190406482331757e-06,
+      "loss": 1.4677,
+      "step": 2344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1880768508468165e-06,
+      "loss": 1.0671,
+      "step": 2345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18574663169604e-06,
+      "loss": 1.3154,
+      "step": 2346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183415825505272e-06,
+      "loss": 1.2165,
+      "step": 2347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1810844329005095e-06,
+      "loss": 1.5042,
+      "step": 2348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.178752454507911e-06,
+      "loss": 0.8667,
+      "step": 2349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176419890953788e-06,
+      "loss": 1.1014,
+      "step": 2350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1740867428646165e-06,
+      "loss": 1.1527,
+      "step": 2351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171753010867023e-06,
+      "loss": 1.2294,
+      "step": 2352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169418695587791e-06,
+      "loss": 0.9737,
+      "step": 2353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167083797653866e-06,
+      "loss": 1.1157,
+      "step": 2354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164748317692344e-06,
+      "loss": 1.1024,
+      "step": 2355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162412256330481e-06,
+      "loss": 1.1676,
+      "step": 2356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160075614195688e-06,
+      "loss": 1.3796,
+      "step": 2357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157738391915531e-06,
+      "loss": 1.1848,
+      "step": 2358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1554005901177324e-06,
+      "loss": 1.024,
+      "step": 2359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.153062209430174e-06,
+      "loss": 1.304,
+      "step": 2360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1507232504808854e-06,
+      "loss": 1.0988,
+      "step": 2361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.148383713898058e-06,
+      "loss": 1.1509,
+      "step": 2362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1460436003100365e-06,
+      "loss": 1.3859,
+      "step": 2363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143702910345318e-06,
+      "loss": 1.0316,
+      "step": 2364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.141361644632561e-06,
+      "loss": 1.3581,
+      "step": 2365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139019803800569e-06,
+      "loss": 1.1722,
+      "step": 2366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.136677388478309e-06,
+      "loss": 1.0703,
+      "step": 2367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.134334399294897e-06,
+      "loss": 1.1066,
+      "step": 2368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1319908368796055e-06,
+      "loss": 1.3224,
+      "step": 2369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129646701861858e-06,
+      "loss": 1.2441,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127301994871237e-06,
+      "loss": 1.174,
+      "step": 2371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124956716537471e-06,
+      "loss": 1.2845,
+      "step": 2372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1226108674904515e-06,
+      "loss": 1.3163,
+      "step": 2373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120264448360214e-06,
+      "loss": 1.119,
+      "step": 2374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117917459776955e-06,
+      "loss": 0.915,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115569902371018e-06,
+      "loss": 1.1496,
+      "step": 2376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.113221776772902e-06,
+      "loss": 1.0112,
+      "step": 2377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110873083613259e-06,
+      "loss": 1.353,
+      "step": 2378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.108523823522891e-06,
+      "loss": 1.3838,
+      "step": 2379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106173997132755e-06,
+      "loss": 1.0035,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.103823605073958e-06,
+      "loss": 1.2118,
+      "step": 2381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101472647977761e-06,
+      "loss": 1.3795,
+      "step": 2382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099121126475577e-06,
+      "loss": 0.8687,
+      "step": 2383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096769041198964e-06,
+      "loss": 1.3729,
+      "step": 2384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.094416392779642e-06,
+      "loss": 0.9658,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0920631818494745e-06,
+      "loss": 1.3014,
+      "step": 2386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089709409040479e-06,
+      "loss": 1.0117,
+      "step": 2387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087355074984823e-06,
+      "loss": 1.2103,
+      "step": 2388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085000180314825e-06,
+      "loss": 1.2967,
+      "step": 2389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.082644725662954e-06,
+      "loss": 1.3539,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080288711661831e-06,
+      "loss": 1.1581,
+      "step": 2391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077932138944225e-06,
+      "loss": 1.3043,
+      "step": 2392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.075575008143055e-06,
+      "loss": 0.8765,
+      "step": 2393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073217319891391e-06,
+      "loss": 1.3286,
+      "step": 2394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070859074822454e-06,
+      "loss": 1.1912,
+      "step": 2395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068500273569612e-06,
+      "loss": 1.109,
+      "step": 2396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066140916766382e-06,
+      "loss": 1.2519,
+      "step": 2397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063781005046433e-06,
+      "loss": 0.821,
+      "step": 2398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061420539043582e-06,
+      "loss": 1.0499,
+      "step": 2399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059059519391794e-06,
+      "loss": 1.5203,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0566979467251835e-06,
+      "loss": 1.4596,
+      "step": 2401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.054335821678012e-06,
+      "loss": 1.3515,
+      "step": 2402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.051973144884692e-06,
+      "loss": 1.171,
+      "step": 2403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.049609916979782e-06,
+      "loss": 1.1736,
+      "step": 2404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.047246138597989e-06,
+      "loss": 1.0802,
+      "step": 2405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044881810374169e-06,
+      "loss": 1.116,
+      "step": 2406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.042516932943324e-06,
+      "loss": 1.147,
+      "step": 2407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040151506940605e-06,
+      "loss": 1.1664,
+      "step": 2408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03778553300131e-06,
+      "loss": 1.1771,
+      "step": 2409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035419011760882e-06,
+      "loss": 0.7954,
+      "step": 2410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033051943854915e-06,
+      "loss": 1.1214,
+      "step": 2411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0306843299191465e-06,
+      "loss": 1.1605,
+      "step": 2412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02831617058946e-06,
+      "loss": 1.2669,
+      "step": 2413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0259474665018915e-06,
+      "loss": 0.9503,
+      "step": 2414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023578218292615e-06,
+      "loss": 1.1169,
+      "step": 2415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0212084265979575e-06,
+      "loss": 1.2737,
+      "step": 2416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0188380920543864e-06,
+      "loss": 1.0036,
+      "step": 2417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016467215298519e-06,
+      "loss": 1.6136,
+      "step": 2418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01409579696712e-06,
+      "loss": 1.193,
+      "step": 2419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011723837697091e-06,
+      "loss": 1.2726,
+      "step": 2420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009351338125487e-06,
+      "loss": 1.1831,
+      "step": 2421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0069782988895056e-06,
+      "loss": 1.1566,
+      "step": 2422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.004604720626489e-06,
+      "loss": 1.3971,
+      "step": 2423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002230603973924e-06,
+      "loss": 0.7277,
+      "step": 2424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9998559495694415e-06,
+      "loss": 1.2749,
+      "step": 2425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9974807580508205e-06,
+      "loss": 1.045,
+      "step": 2426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.995105030055978e-06,
+      "loss": 0.911,
+      "step": 2427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992728766222982e-06,
+      "loss": 0.9996,
+      "step": 2428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990351967190037e-06,
+      "loss": 0.749,
+      "step": 2429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987974633595498e-06,
+      "loss": 1.1398,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98559676607786e-06,
+      "loss": 1.1913,
+      "step": 2431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9832183652757625e-06,
+      "loss": 1.1481,
+      "step": 2432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980839431827986e-06,
+      "loss": 1.2305,
+      "step": 2433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978459966373458e-06,
+      "loss": 1.0897,
+      "step": 2434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976079969551246e-06,
+      "loss": 1.2582,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973699442000561e-06,
+      "loss": 1.4118,
+      "step": 2436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9713183843607575e-06,
+      "loss": 1.0791,
+      "step": 2437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96893679727133e-06,
+      "loss": 1.0377,
+      "step": 2438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966554681371917e-06,
+      "loss": 1.0665,
+      "step": 2439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9641720373022996e-06,
+      "loss": 1.0327,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9617888657023976e-06,
+      "loss": 1.2373,
+      "step": 2441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959405167212278e-06,
+      "loss": 1.3552,
+      "step": 2442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9570209424721425e-06,
+      "loss": 1.4616,
+      "step": 2443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954636192122339e-06,
+      "loss": 1.1943,
+      "step": 2444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.952250916803358e-06,
+      "loss": 1.1906,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949865117155823e-06,
+      "loss": 1.1066,
+      "step": 2446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947478793820508e-06,
+      "loss": 1.2142,
+      "step": 2447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94509194743832e-06,
+      "loss": 1.2306,
+      "step": 2448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.942704578650312e-06,
+      "loss": 1.0503,
+      "step": 2449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940316688097675e-06,
+      "loss": 0.9296,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.937928276421736e-06,
+      "loss": 1.1381,
+      "step": 2451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935539344263971e-06,
+      "loss": 1.1462,
+      "step": 2452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933149892265989e-06,
+      "loss": 1.3275,
+      "step": 2453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93075992106954e-06,
+      "loss": 1.1633,
+      "step": 2454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9283694313165135e-06,
+      "loss": 1.1657,
+      "step": 2455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925978423648941e-06,
+      "loss": 1.0371,
+      "step": 2456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9235868987089875e-06,
+      "loss": 1.1797,
+      "step": 2457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921194857138963e-06,
+      "loss": 1.1524,
+      "step": 2458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.918802299581312e-06,
+      "loss": 1.1839,
+      "step": 2459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91640922667862e-06,
+      "loss": 1.1972,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914015639073608e-06,
+      "loss": 0.9741,
+      "step": 2461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911621537409139e-06,
+      "loss": 1.3813,
+      "step": 2462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.909226922328212e-06,
+      "loss": 1.4291,
+      "step": 2463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906831794473963e-06,
+      "loss": 1.2965,
+      "step": 2464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904436154489668e-06,
+      "loss": 1.4911,
+      "step": 2465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9020400030187394e-06,
+      "loss": 1.2243,
+      "step": 2466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899643340704727e-06,
+      "loss": 1.2912,
+      "step": 2467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897246168191317e-06,
+      "loss": 0.9077,
+      "step": 2468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894848486122333e-06,
+      "loss": 0.953,
+      "step": 2469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892450295141737e-06,
+      "loss": 1.1874,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8900515958936276e-06,
+      "loss": 0.8451,
+      "step": 2471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887652389022236e-06,
+      "loss": 1.4001,
+      "step": 2472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885252675171935e-06,
+      "loss": 1.2285,
+      "step": 2473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.88285245498723e-06,
+      "loss": 1.1768,
+      "step": 2474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880451729112762e-06,
+      "loss": 1.0537,
+      "step": 2475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878050498193314e-06,
+      "loss": 1.3438,
+      "step": 2476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8756487628737945e-06,
+      "loss": 1.1138,
+      "step": 2477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873246523799256e-06,
+      "loss": 1.0983,
+      "step": 2478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870843781614884e-06,
+      "loss": 1.2078,
+      "step": 2479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868440536965997e-06,
+      "loss": 0.9049,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866036790498049e-06,
+      "loss": 1.0915,
+      "step": 2481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863632542856632e-06,
+      "loss": 0.9461,
+      "step": 2482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8612277946874675e-06,
+      "loss": 1.1273,
+      "step": 2483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858822546636417e-06,
+      "loss": 1.1824,
+      "step": 2484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85641679934947e-06,
+      "loss": 1.0943,
+      "step": 2485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.854010553472757e-06,
+      "loss": 1.0107,
+      "step": 2486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851603809652535e-06,
+      "loss": 1.464,
+      "step": 2487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849196568535201e-06,
+      "loss": 1.3,
+      "step": 2488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8467888307672845e-06,
+      "loss": 1.3774,
+      "step": 2489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8443805969954445e-06,
+      "loss": 1.2512,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.841971867866478e-06,
+      "loss": 1.1162,
+      "step": 2491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839562644027311e-06,
+      "loss": 0.9748,
+      "step": 2492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837152926125004e-06,
+      "loss": 1.4741,
+      "step": 2493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834742714806754e-06,
+      "loss": 1.1971,
+      "step": 2494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832332010719882e-06,
+      "loss": 1.3961,
+      "step": 2495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8299208145118475e-06,
+      "loss": 1.0982,
+      "step": 2496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827509126830245e-06,
+      "loss": 1.1559,
+      "step": 2497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825096948322791e-06,
+      "loss": 1.2634,
+      "step": 2498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822684279637345e-06,
+      "loss": 0.9554,
+      "step": 2499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820271121421889e-06,
+      "loss": 1.1957,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817857474324542e-06,
+      "loss": 1.0693,
+      "step": 2501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815443338993554e-06,
+      "loss": 1.4312,
+      "step": 2502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.813028716077301e-06,
+      "loss": 0.9365,
+      "step": 2503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810613606224299e-06,
+      "loss": 1.1189,
+      "step": 2504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808198010083184e-06,
+      "loss": 1.1932,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.805781928302732e-06,
+      "loss": 1.3271,
+      "step": 2506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803365361531844e-06,
+      "loss": 1.1111,
+      "step": 2507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800948310419554e-06,
+      "loss": 1.4027,
+      "step": 2508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798530775615024e-06,
+      "loss": 0.6963,
+      "step": 2509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796112757767547e-06,
+      "loss": 1.3557,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793694257526545e-06,
+      "loss": 1.2016,
+      "step": 2511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7912752755415716e-06,
+      "loss": 1.24,
+      "step": 2512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788855812462307e-06,
+      "loss": 1.0356,
+      "step": 2513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786435868938561e-06,
+      "loss": 1.297,
+      "step": 2514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784015445620278e-06,
+      "loss": 1.1936,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78159454315752e-06,
+      "loss": 0.9793,
+      "step": 2516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.779173162200489e-06,
+      "loss": 1.22,
+      "step": 2517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776751303399509e-06,
+      "loss": 1.2374,
+      "step": 2518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774328967405035e-06,
+      "loss": 1.0736,
+      "step": 2519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.771906154867649e-06,
+      "loss": 1.0719,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7694828664380605e-06,
+      "loss": 1.002,
+      "step": 2521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767059102767109e-06,
+      "loss": 1.2167,
+      "step": 2522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764634864505759e-06,
+      "loss": 1.8472,
+      "step": 2523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7622101523051045e-06,
+      "loss": 1.0165,
+      "step": 2524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759784966816366e-06,
+      "loss": 1.1918,
+      "step": 2525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757359308690889e-06,
+      "loss": 1.3162,
+      "step": 2526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754933178580152e-06,
+      "loss": 0.9484,
+      "step": 2527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7525065771357546e-06,
+      "loss": 1.1928,
+      "step": 2528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750079505009424e-06,
+      "loss": 0.9565,
+      "step": 2529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7476519628530145e-06,
+      "loss": 1.118,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.745223951318507e-06,
+      "loss": 1.0342,
+      "step": 2531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742795471058009e-06,
+      "loss": 1.2093,
+      "step": 2532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740366522723753e-06,
+      "loss": 1.2465,
+      "step": 2533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737937106968094e-06,
+      "loss": 1.0797,
+      "step": 2534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.73550722444352e-06,
+      "loss": 1.3077,
+      "step": 2535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7330768758026374e-06,
+      "loss": 1.4044,
+      "step": 2536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730646061698183e-06,
+      "loss": 1.1518,
+      "step": 2537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728214782783013e-06,
+      "loss": 1.0573,
+      "step": 2538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725783039710115e-06,
+      "loss": 1.1181,
+      "step": 2539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723350833132596e-06,
+      "loss": 1.1557,
+      "step": 2540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720918163703689e-06,
+      "loss": 1.0922,
+      "step": 2541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7184850320767505e-06,
+      "loss": 1.245,
+      "step": 2542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.716051438905266e-06,
+      "loss": 1.1326,
+      "step": 2543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7136173848428375e-06,
+      "loss": 1.0508,
+      "step": 2544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.711182870543196e-06,
+      "loss": 1.0633,
+      "step": 2545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708747896660196e-06,
+      "loss": 1.235,
+      "step": 2546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70631246384781e-06,
+      "loss": 1.2071,
+      "step": 2547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703876572760144e-06,
+      "loss": 0.9675,
+      "step": 2548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.701440224051415e-06,
+      "loss": 1.0037,
+      "step": 2549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6990034183759726e-06,
+      "loss": 1.0475,
+      "step": 2550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696566156388283e-06,
+      "loss": 1.3159,
+      "step": 2551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694128438742939e-06,
+      "loss": 1.4496,
+      "step": 2552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691690266094652e-06,
+      "loss": 1.3318,
+      "step": 2553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689251639098261e-06,
+      "loss": 1.0636,
+      "step": 2554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686812558408721e-06,
+      "loss": 1.2303,
+      "step": 2555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684373024681112e-06,
+      "loss": 1.0244,
+      "step": 2556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.681933038570635e-06,
+      "loss": 0.9443,
+      "step": 2557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679492600732614e-06,
+      "loss": 1.0463,
+      "step": 2558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677051711822491e-06,
+      "loss": 1.6848,
+      "step": 2559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674610372495832e-06,
+      "loss": 1.0602,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.672168583408326e-06,
+      "loss": 1.1215,
+      "step": 2561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669726345215776e-06,
+      "loss": 1.0696,
+      "step": 2562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667283658574111e-06,
+      "loss": 1.1115,
+      "step": 2563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66484052413938e-06,
+      "loss": 1.2751,
+      "step": 2564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662396942567749e-06,
+      "loss": 1.2111,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.659952914515508e-06,
+      "loss": 0.8877,
+      "step": 2566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.657508440639066e-06,
+      "loss": 1.2603,
+      "step": 2567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65506352159495e-06,
+      "loss": 1.1308,
+      "step": 2568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652618158039808e-06,
+      "loss": 1.1378,
+      "step": 2569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650172350630406e-06,
+      "loss": 1.1295,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647726100023631e-06,
+      "loss": 1.0191,
+      "step": 2571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645279406876488e-06,
+      "loss": 0.9749,
+      "step": 2572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6428322718461e-06,
+      "loss": 1.1638,
+      "step": 2573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640384695589714e-06,
+      "loss": 1.3979,
+      "step": 2574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637936678764685e-06,
+      "loss": 1.0107,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635488222028497e-06,
+      "loss": 1.1013,
+      "step": 2576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633039326038746e-06,
+      "loss": 1.0004,
+      "step": 2577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630589991453148e-06,
+      "loss": 1.1581,
+      "step": 2578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6281402189295375e-06,
+      "loss": 1.2739,
+      "step": 2579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6256900091258644e-06,
+      "loss": 0.9289,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623239362700197e-06,
+      "loss": 1.012,
+      "step": 2581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.620788280310722e-06,
+      "loss": 1.4703,
+      "step": 2582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618336762615742e-06,
+      "loss": 1.1492,
+      "step": 2583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615884810273678e-06,
+      "loss": 1.0489,
+      "step": 2584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.613432423943066e-06,
+      "loss": 1.1367,
+      "step": 2585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610979604282557e-06,
+      "loss": 0.9442,
+      "step": 2586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608526351950923e-06,
+      "loss": 1.0145,
+      "step": 2587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.606072667607048e-06,
+      "loss": 1.1778,
+      "step": 2588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.603618551909935e-06,
+      "loss": 1.149,
+      "step": 2589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601164005518702e-06,
+      "loss": 1.104,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59870902909258e-06,
+      "loss": 1.1177,
+      "step": 2591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59625362329092e-06,
+      "loss": 1.2589,
+      "step": 2592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593797788773186e-06,
+      "loss": 0.958,
+      "step": 2593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.591341526198955e-06,
+      "loss": 1.0527,
+      "step": 2594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588884836227925e-06,
+      "loss": 0.7294,
+      "step": 2595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.586427719519901e-06,
+      "loss": 0.9698,
+      "step": 2596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583970176734809e-06,
+      "loss": 1.1056,
+      "step": 2597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581512208532685e-06,
+      "loss": 1.1159,
+      "step": 2598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.579053815573684e-06,
+      "loss": 1.4222,
+      "step": 2599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.576594998518071e-06,
+      "loss": 0.8984,
+      "step": 2600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.574135758026226e-06,
+      "loss": 1.2699,
+      "step": 2601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5716760947586425e-06,
+      "loss": 1.3234,
+      "step": 2602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56921600937593e-06,
+      "loss": 0.9414,
+      "step": 2603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566755502538806e-06,
+      "loss": 0.7012,
+      "step": 2604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56429457490811e-06,
+      "loss": 1.2965,
+      "step": 2605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.561833227144784e-06,
+      "loss": 1.2086,
+      "step": 2606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55937145990989e-06,
+      "loss": 1.192,
+      "step": 2607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.556909273864601e-06,
+      "loss": 1.1715,
+      "step": 2608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554446669670201e-06,
+      "loss": 0.9643,
+      "step": 2609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551983647988089e-06,
+      "loss": 1.5582,
+      "step": 2610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.549520209479773e-06,
+      "loss": 0.9649,
+      "step": 2611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.547056354806874e-06,
+      "loss": 1.4916,
+      "step": 2612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.544592084631127e-06,
+      "loss": 0.9034,
+      "step": 2613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.542127399614376e-06,
+      "loss": 1.1493,
+      "step": 2614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.539662300418578e-06,
+      "loss": 1.2504,
+      "step": 2615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5371967877058e-06,
+      "loss": 1.214,
+      "step": 2616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.53473086213822e-06,
+      "loss": 1.1208,
+      "step": 2617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.532264524378128e-06,
+      "loss": 1.0828,
+      "step": 2618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5297977750879235e-06,
+      "loss": 1.3292,
+      "step": 2619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52733061493012e-06,
+      "loss": 1.0971,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5248630445673335e-06,
+      "loss": 1.3247,
+      "step": 2621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522395064662299e-06,
+      "loss": 1.1521,
+      "step": 2622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519926675877857e-06,
+      "loss": 1.1487,
+      "step": 2623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.517457878876958e-06,
+      "loss": 1.4269,
+      "step": 2624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514988674322662e-06,
+      "loss": 1.0947,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512519062878142e-06,
+      "loss": 1.1534,
+      "step": 2626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.510049045206672e-06,
+      "loss": 1.1432,
+      "step": 2627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507578621971646e-06,
+      "loss": 1.1865,
+      "step": 2628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505107793836561e-06,
+      "loss": 0.7714,
+      "step": 2629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502636561465018e-06,
+      "loss": 1.1799,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500164925520737e-06,
+      "loss": 0.8865,
+      "step": 2631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497692886667537e-06,
+      "loss": 1.2264,
+      "step": 2632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495220445569353e-06,
+      "loss": 1.1817,
+      "step": 2633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.492747602890223e-06,
+      "loss": 0.9267,
+      "step": 2634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490274359294295e-06,
+      "loss": 1.1592,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487800715445822e-06,
+      "loss": 1.1531,
+      "step": 2636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.485326672009168e-06,
+      "loss": 1.3936,
+      "step": 2637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4828522296488014e-06,
+      "loss": 0.9079,
+      "step": 2638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.480377389029301e-06,
+      "loss": 1.2084,
+      "step": 2639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477902150815347e-06,
+      "loss": 1.166,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475426515671736e-06,
+      "loss": 0.9637,
+      "step": 2641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472950484263359e-06,
+      "loss": 1.209,
+      "step": 2642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.470474057255222e-06,
+      "loss": 0.8194,
+      "step": 2643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467997235312437e-06,
+      "loss": 1.1342,
+      "step": 2644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465520019100218e-06,
+      "loss": 1.2446,
+      "step": 2645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463042409283885e-06,
+      "loss": 1.0653,
+      "step": 2646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.460564406528869e-06,
+      "loss": 1.0097,
+      "step": 2647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458086011500703e-06,
+      "loss": 1.0457,
+      "step": 2648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.455607224865022e-06,
+      "loss": 0.9267,
+      "step": 2649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453128047287573e-06,
+      "loss": 0.8994,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450648479434201e-06,
+      "loss": 1.1499,
+      "step": 2651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448168521970865e-06,
+      "loss": 1.1198,
+      "step": 2652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4456881755636165e-06,
+      "loss": 1.4086,
+      "step": 2653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.443207440878624e-06,
+      "loss": 1.2137,
+      "step": 2654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44072631858215e-06,
+      "loss": 1.0211,
+      "step": 2655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438244809340568e-06,
+      "loss": 1.1283,
+      "step": 2656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.435762913820352e-06,
+      "loss": 1.374,
+      "step": 2657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43328063268808e-06,
+      "loss": 1.0765,
+      "step": 2658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.430797966610436e-06,
+      "loss": 1.0198,
+      "step": 2659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.428314916254203e-06,
+      "loss": 1.1017,
+      "step": 2660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.425831482286273e-06,
+      "loss": 1.1984,
+      "step": 2661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.423347665373633e-06,
+      "loss": 1.0706,
+      "step": 2662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.420863466183383e-06,
+      "loss": 1.2971,
+      "step": 2663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.418378885382716e-06,
+      "loss": 1.0196,
+      "step": 2664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.415893923638936e-06,
+      "loss": 1.2107,
+      "step": 2665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41340858161944e-06,
+      "loss": 1.4184,
+      "step": 2666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4109228599917374e-06,
+      "loss": 1.0145,
+      "step": 2667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408436759423431e-06,
+      "loss": 1.229,
+      "step": 2668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4059502805822295e-06,
+      "loss": 1.4235,
+      "step": 2669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.403463424135943e-06,
+      "loss": 1.2013,
+      "step": 2670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.400976190752482e-06,
+      "loss": 1.0585,
+      "step": 2671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.398488581099859e-06,
+      "loss": 1.1443,
+      "step": 2672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.396000595846188e-06,
+      "loss": 1.4333,
+      "step": 2673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.393512235659681e-06,
+      "loss": 1.0433,
+      "step": 2674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3910235012086555e-06,
+      "loss": 1.1739,
+      "step": 2675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388534393161525e-06,
+      "loss": 1.1363,
+      "step": 2676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.386044912186806e-06,
+      "loss": 1.241,
+      "step": 2677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383555058953115e-06,
+      "loss": 0.9487,
+      "step": 2678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381064834129167e-06,
+      "loss": 1.3098,
+      "step": 2679
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378574238383776e-06,
+      "loss": 1.1222,
+      "step": 2680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37608327238586e-06,
+      "loss": 1.2366,
+      "step": 2681
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373591936804433e-06,
+      "loss": 1.236,
+      "step": 2682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.371100232308609e-06,
+      "loss": 1.1111,
+      "step": 2683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3686081595676e-06,
+      "loss": 1.0585,
+      "step": 2684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366115719250719e-06,
+      "loss": 1.1636,
+      "step": 2685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3636229120273766e-06,
+      "loss": 1.04,
+      "step": 2686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.361129738567081e-06,
+      "loss": 1.3193,
+      "step": 2687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3586361995394415e-06,
+      "loss": 1.3104,
+      "step": 2688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.356142295614163e-06,
+      "loss": 0.9609,
+      "step": 2689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353648027461048e-06,
+      "loss": 1.3286,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351153395750002e-06,
+      "loss": 1.2418,
+      "step": 2691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348658401151018e-06,
+      "loss": 0.8418,
+      "step": 2692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346163044334199e-06,
+      "loss": 1.3778,
+      "step": 2693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343667325969736e-06,
+      "loss": 1.1822,
+      "step": 2694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.341171246727919e-06,
+      "loss": 1.2911,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3386748072791395e-06,
+      "loss": 0.9015,
+      "step": 2696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.336178008293878e-06,
+      "loss": 1.1631,
+      "step": 2697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33368085044272e-06,
+      "loss": 0.9928,
+      "step": 2698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331183334396341e-06,
+      "loss": 1.1521,
+      "step": 2699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.328685460825512e-06,
+      "loss": 1.3827,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.326187230401109e-06,
+      "loss": 1.1708,
+      "step": 2701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323688643794094e-06,
+      "loss": 1.2465,
+      "step": 2702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3211897016755295e-06,
+      "loss": 1.1711,
+      "step": 2703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.318690404716572e-06,
+      "loss": 1.0793,
+      "step": 2704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.316190753588474e-06,
+      "loss": 1.0836,
+      "step": 2705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.313690748962582e-06,
+      "loss": 0.8649,
+      "step": 2706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31119039151034e-06,
+      "loss": 1.313,
+      "step": 2707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3086896819032814e-06,
+      "loss": 0.9761,
+      "step": 2708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306188620813044e-06,
+      "loss": 1.0621,
+      "step": 2709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.303687208911348e-06,
+      "loss": 1.2529,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.301185446870017e-06,
+      "loss": 1.2113,
+      "step": 2711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298683335360962e-06,
+      "loss": 1.6126,
+      "step": 2712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296180875056195e-06,
+      "loss": 0.8841,
+      "step": 2713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293678066627816e-06,
+      "loss": 1.1716,
+      "step": 2714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2911749107480206e-06,
+      "loss": 1.477,
+      "step": 2715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288671408089098e-06,
+      "loss": 1.0344,
+      "step": 2716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2861675593234275e-06,
+      "loss": 1.1181,
+      "step": 2717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.283663365123486e-06,
+      "loss": 1.1031,
+      "step": 2718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2811588261618405e-06,
+      "loss": 1.1086,
+      "step": 2719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.278653943111152e-06,
+      "loss": 1.0066,
+      "step": 2720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.27614871664417e-06,
+      "loss": 1.154,
+      "step": 2721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.273643147433743e-06,
+      "loss": 1.3212,
+      "step": 2722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.271137236152806e-06,
+      "loss": 0.9781,
+      "step": 2723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.268630983474388e-06,
+      "loss": 1.1313,
+      "step": 2724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26612439007161e-06,
+      "loss": 1.3778,
+      "step": 2725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263617456617681e-06,
+      "loss": 1.1765,
+      "step": 2726
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.261110183785909e-06,
+      "loss": 1.2386,
+      "step": 2727
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.258602572249683e-06,
+      "loss": 1.4837,
+      "step": 2728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256094622682493e-06,
+      "loss": 1.0228,
+      "step": 2729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2535863357579105e-06,
+      "loss": 0.9574,
+      "step": 2730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.251077712149608e-06,
+      "loss": 1.2453,
+      "step": 2731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248568752531337e-06,
+      "loss": 1.1986,
+      "step": 2732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.246059457576948e-06,
+      "loss": 1.0091,
+      "step": 2733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243549827960378e-06,
+      "loss": 1.0974,
+      "step": 2734
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241039864355654e-06,
+      "loss": 1.3838,
+      "step": 2735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238529567436892e-06,
+      "loss": 1.1205,
+      "step": 2736
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.236018937878301e-06,
+      "loss": 1.1349,
+      "step": 2737
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233507976354174e-06,
+      "loss": 1.356,
+      "step": 2738
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230996683538898e-06,
+      "loss": 1.195,
+      "step": 2739
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.228485060106948e-06,
+      "loss": 1.2069,
+      "step": 2740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.225973106732883e-06,
+      "loss": 1.2825,
+      "step": 2741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.223460824091358e-06,
+      "loss": 0.9558,
+      "step": 2742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220948212857112e-06,
+      "loss": 1.0563,
+      "step": 2743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218435273704973e-06,
+      "loss": 1.1776,
+      "step": 2744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.215922007309856e-06,
+      "loss": 1.2167,
+      "step": 2745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213408414346765e-06,
+      "loss": 1.1573,
+      "step": 2746
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.210894495490795e-06,
+      "loss": 1.171,
+      "step": 2747
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208380251417122e-06,
+      "loss": 1.0759,
+      "step": 2748
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.205865682801014e-06,
+      "loss": 1.3649,
+      "step": 2749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203350790317825e-06,
+      "loss": 1.0374,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.200835574642995e-06,
+      "loss": 0.9424,
+      "step": 2751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198320036452051e-06,
+      "loss": 0.8477,
+      "step": 2752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.19580417642061e-06,
+      "loss": 1.1315,
+      "step": 2753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.193287995224371e-06,
+      "loss": 1.1466,
+      "step": 2754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.19077149353912e-06,
+      "loss": 1.217,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18825467204073e-06,
+      "loss": 1.3183,
+      "step": 2756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.185737531405162e-06,
+      "loss": 0.9355,
+      "step": 2757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.183220072308459e-06,
+      "loss": 1.2816,
+      "step": 2758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180702295426752e-06,
+      "loss": 1.1704,
+      "step": 2759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178184201436256e-06,
+      "loss": 1.1971,
+      "step": 2760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1756657910132714e-06,
+      "loss": 1.1807,
+      "step": 2761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173147064834183e-06,
+      "loss": 1.1733,
+      "step": 2762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1706280235754636e-06,
+      "loss": 1.5338,
+      "step": 2763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168108667913666e-06,
+      "loss": 1.3372,
+      "step": 2764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1655889985254315e-06,
+      "loss": 1.2423,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163069016087483e-06,
+      "loss": 0.9558,
+      "step": 2766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.160548721276627e-06,
+      "loss": 1.1828,
+      "step": 2767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158028114769758e-06,
+      "loss": 1.0127,
+      "step": 2768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155507197243849e-06,
+      "loss": 1.3,
+      "step": 2769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152985969375962e-06,
+      "loss": 1.1065,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150464431843236e-06,
+      "loss": 1.1365,
+      "step": 2771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147942585322898e-06,
+      "loss": 1.1999,
+      "step": 2772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.145420430492257e-06,
+      "loss": 1.1602,
+      "step": 2773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.142897968028704e-06,
+      "loss": 1.1291,
+      "step": 2774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.140375198609712e-06,
+      "loss": 0.9586,
+      "step": 2775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.137852122912839e-06,
+      "loss": 1.2556,
+      "step": 2776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1353287416157204e-06,
+      "loss": 1.114,
+      "step": 2777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1328050553960804e-06,
+      "loss": 1.3262,
+      "step": 2778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.130281064931721e-06,
+      "loss": 0.9981,
+      "step": 2779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1277567709005245e-06,
+      "loss": 1.2009,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.12523217398046e-06,
+      "loss": 1.4091,
+      "step": 2781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122707274849572e-06,
+      "loss": 1.4017,
+      "step": 2782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.120182074185988e-06,
+      "loss": 1.1258,
+      "step": 2783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117656572667921e-06,
+      "loss": 1.1138,
+      "step": 2784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115130770973658e-06,
+      "loss": 1.1248,
+      "step": 2785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.112604669781572e-06,
+      "loss": 1.3584,
+      "step": 2786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.110078269770113e-06,
+      "loss": 0.8821,
+      "step": 2787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107551571617813e-06,
+      "loss": 1.3212,
+      "step": 2788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.105024576003281e-06,
+      "loss": 0.7359,
+      "step": 2789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1024972836052135e-06,
+      "loss": 1.206,
+      "step": 2790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0999696951023765e-06,
+      "loss": 1.1672,
+      "step": 2791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0974418111736235e-06,
+      "loss": 1.1479,
+      "step": 2792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094913632497882e-06,
+      "loss": 1.1807,
+      "step": 2793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.092385159754165e-06,
+      "loss": 1.2989,
+      "step": 2794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.089856393621558e-06,
+      "loss": 1.1855,
+      "step": 2795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0873273347792275e-06,
+      "loss": 0.9834,
+      "step": 2796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0847979839064216e-06,
+      "loss": 1.1508,
+      "step": 2797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0822683416824625e-06,
+      "loss": 1.2125,
+      "step": 2798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.079738408786753e-06,
+      "loss": 1.1003,
+      "step": 2799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077208185898772e-06,
+      "loss": 1.1544,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.074677673698079e-06,
+      "loss": 0.8693,
+      "step": 2801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07214687286431e-06,
+      "loss": 1.1865,
+      "step": 2802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.069615784077178e-06,
+      "loss": 1.0269,
+      "step": 2803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.067084408016475e-06,
+      "loss": 1.4922,
+      "step": 2804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.064552745362067e-06,
+      "loss": 1.1805,
+      "step": 2805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0620207967939e-06,
+      "loss": 1.0885,
+      "step": 2806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.059488562991994e-06,
+      "loss": 0.8339,
+      "step": 2807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0569560446364495e-06,
+      "loss": 1.2256,
+      "step": 2808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.05442324240744e-06,
+      "loss": 0.8611,
+      "step": 2809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051890156985217e-06,
+      "loss": 0.9923,
+      "step": 2810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049356789050107e-06,
+      "loss": 0.9592,
+      "step": 2811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046823139282515e-06,
+      "loss": 1.2752,
+      "step": 2812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044289208362914e-06,
+      "loss": 1.1951,
+      "step": 2813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.041754996971866e-06,
+      "loss": 1.0043,
+      "step": 2814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.039220505789994e-06,
+      "loss": 1.1297,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.036685735498004e-06,
+      "loss": 1.18,
+      "step": 2816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.034150686776677e-06,
+      "loss": 1.2191,
+      "step": 2817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031615360306867e-06,
+      "loss": 1.1919,
+      "step": 2818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029079756769503e-06,
+      "loss": 1.0355,
+      "step": 2819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026543876845586e-06,
+      "loss": 1.1276,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0240077212161965e-06,
+      "loss": 1.0177,
+      "step": 2821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.021471290562484e-06,
+      "loss": 1.1378,
+      "step": 2822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.018934585565675e-06,
+      "loss": 1.0444,
+      "step": 2823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.016397606907069e-06,
+      "loss": 0.8274,
+      "step": 2824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013860355268039e-06,
+      "loss": 1.1387,
+      "step": 2825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.011322831330028e-06,
+      "loss": 0.8574,
+      "step": 2826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0087850357745566e-06,
+      "loss": 1.1486,
+      "step": 2827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0062469692832205e-06,
+      "loss": 1.0735,
+      "step": 2828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0037086325376785e-06,
+      "loss": 1.0362,
+      "step": 2829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001170026219673e-06,
+      "loss": 1.0148,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.998631151011009e-06,
+      "loss": 1.167,
+      "step": 2831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.996092007593572e-06,
+      "loss": 1.0852,
+      "step": 2832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993552596649317e-06,
+      "loss": 1.2319,
+      "step": 2833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9910129188602665e-06,
+      "loss": 1.1112,
+      "step": 2834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98847297490852e-06,
+      "loss": 1.1611,
+      "step": 2835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985932765476246e-06,
+      "loss": 1.1321,
+      "step": 2836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.983392291245685e-06,
+      "loss": 1.177,
+      "step": 2837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9808515528991486e-06,
+      "loss": 0.9476,
+      "step": 2838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.978310551119017e-06,
+      "loss": 1.2181,
+      "step": 2839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.975769286587747e-06,
+      "loss": 1.253,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.973227759987862e-06,
+      "loss": 1.6661,
+      "step": 2841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970685972001953e-06,
+      "loss": 1.3359,
+      "step": 2842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.968143923312687e-06,
+      "loss": 1.1386,
+      "step": 2843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.965601614602798e-06,
+      "loss": 0.9751,
+      "step": 2844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.963059046555088e-06,
+      "loss": 1.0818,
+      "step": 2845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.960516219852433e-06,
+      "loss": 1.0342,
+      "step": 2846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.957973135177776e-06,
+      "loss": 1.2795,
+      "step": 2847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.955429793214129e-06,
+      "loss": 0.9018,
+      "step": 2848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.952886194644573e-06,
+      "loss": 1.2385,
+      "step": 2849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950342340152261e-06,
+      "loss": 1.1693,
+      "step": 2850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947798230420409e-06,
+      "loss": 1.3747,
+      "step": 2851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945253866132308e-06,
+      "loss": 0.9318,
+      "step": 2852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942709247971312e-06,
+      "loss": 1.084,
+      "step": 2853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.940164376620847e-06,
+      "loss": 0.9273,
+      "step": 2854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937619252764405e-06,
+      "loss": 1.3084,
+      "step": 2855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935073877085546e-06,
+      "loss": 1.3007,
+      "step": 2856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.932528250267897e-06,
+      "loss": 1.1121,
+      "step": 2857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9299823729951544e-06,
+      "loss": 0.9726,
+      "step": 2858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.927436245951081e-06,
+      "loss": 0.8868,
+      "step": 2859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9248898698195054e-06,
+      "loss": 1.2645,
+      "step": 2860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.922343245284325e-06,
+      "loss": 1.08,
+      "step": 2861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.919796373029504e-06,
+      "loss": 1.1377,
+      "step": 2862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.917249253739069e-06,
+      "loss": 1.076,
+      "step": 2863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.914701888097121e-06,
+      "loss": 1.139,
+      "step": 2864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.912154276787817e-06,
+      "loss": 1.1904,
+      "step": 2865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90960642049539e-06,
+      "loss": 0.8818,
+      "step": 2866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90705831990413e-06,
+      "loss": 1.064,
+      "step": 2867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904509975698399e-06,
+      "loss": 1.0627,
+      "step": 2868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9019613885626235e-06,
+      "loss": 1.0457,
+      "step": 2869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8994125591812914e-06,
+      "loss": 1.0497,
+      "step": 2870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.89686348823896e-06,
+      "loss": 1.1227,
+      "step": 2871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894314176420247e-06,
+      "loss": 1.1657,
+      "step": 2872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.89176462440984e-06,
+      "loss": 1.07,
+      "step": 2873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.889214832892489e-06,
+      "loss": 1.0578,
+      "step": 2874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.886664802553004e-06,
+      "loss": 1.0784,
+      "step": 2875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8841145340762665e-06,
+      "loss": 0.9453,
+      "step": 2876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881564028147218e-06,
+      "loss": 0.9973,
+      "step": 2877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879013285450863e-06,
+      "loss": 1.5174,
+      "step": 2878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876462306672273e-06,
+      "loss": 1.0409,
+      "step": 2879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.873911092496577e-06,
+      "loss": 1.0857,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.871359643608974e-06,
+      "loss": 1.177,
+      "step": 2881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8688079606947226e-06,
+      "loss": 0.7796,
+      "step": 2882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.866256044439144e-06,
+      "loss": 1.2589,
+      "step": 2883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8637038955276225e-06,
+      "loss": 1.3127,
+      "step": 2884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.861151514645607e-06,
+      "loss": 1.107,
+      "step": 2885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.858598902478604e-06,
+      "loss": 1.1583,
+      "step": 2886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8560460597121885e-06,
+      "loss": 1.2135,
+      "step": 2887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.853492987031989e-06,
+      "loss": 1.3394,
+      "step": 2888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850939685123704e-06,
+      "loss": 1.0921,
+      "step": 2889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8483861546730915e-06,
+      "loss": 1.0822,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.845832396365965e-06,
+      "loss": 1.0325,
+      "step": 2891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.843278410888208e-06,
+      "loss": 1.1696,
+      "step": 2892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.840724198925757e-06,
+      "loss": 1.4267,
+      "step": 2893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.838169761164616e-06,
+      "loss": 1.1493,
+      "step": 2894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8356150982908466e-06,
+      "loss": 1.1926,
+      "step": 2895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.83306021099057e-06,
+      "loss": 1.0708,
+      "step": 2896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.830505099949969e-06,
+      "loss": 1.1812,
+      "step": 2897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.827949765855285e-06,
+      "loss": 1.2712,
+      "step": 2898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.825394209392823e-06,
+      "loss": 0.9998,
+      "step": 2899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822838431248943e-06,
+      "loss": 1.3093,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820282432110068e-06,
+      "loss": 1.266,
+      "step": 2901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.817726212662678e-06,
+      "loss": 0.9607,
+      "step": 2902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8151697735933146e-06,
+      "loss": 1.1351,
+      "step": 2903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812613115588575e-06,
+      "loss": 1.1109,
+      "step": 2904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81005623933512e-06,
+      "loss": 1.0977,
+      "step": 2905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807499145519663e-06,
+      "loss": 0.978,
+      "step": 2906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804941834828981e-06,
+      "loss": 0.9403,
+      "step": 2907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802384307949909e-06,
+      "loss": 0.9316,
+      "step": 2908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7998265655693345e-06,
+      "loss": 0.984,
+      "step": 2909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79726860837421e-06,
+      "loss": 1.1675,
+      "step": 2910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7947104370515405e-06,
+      "loss": 0.99,
+      "step": 2911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.792152052288391e-06,
+      "loss": 0.9169,
+      "step": 2912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.789593454771886e-06,
+      "loss": 1.1199,
+      "step": 2913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787034645189199e-06,
+      "loss": 1.1686,
+      "step": 2914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78447562422757e-06,
+      "loss": 1.2833,
+      "step": 2915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7819163925742915e-06,
+      "loss": 1.1309,
+      "step": 2916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7793569509167116e-06,
+      "loss": 0.9774,
+      "step": 2917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.776797299942236e-06,
+      "loss": 1.5849,
+      "step": 2918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774237440338327e-06,
+      "loss": 0.9836,
+      "step": 2919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.771677372792502e-06,
+      "loss": 0.9559,
+      "step": 2920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.769117097992336e-06,
+      "loss": 1.0702,
+      "step": 2921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766556616625456e-06,
+      "loss": 1.2473,
+      "step": 2922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.76399592937955e-06,
+      "loss": 0.8066,
+      "step": 2923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7614350369423555e-06,
+      "loss": 1.028,
+      "step": 2924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.758873940001668e-06,
+      "loss": 1.0741,
+      "step": 2925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7563126392453415e-06,
+      "loss": 1.2575,
+      "step": 2926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7537511353612775e-06,
+      "loss": 0.9128,
+      "step": 2927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.751189429037435e-06,
+      "loss": 1.0898,
+      "step": 2928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748627520961829e-06,
+      "loss": 1.127,
+      "step": 2929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.746065411822528e-06,
+      "loss": 1.1789,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.743503102307655e-06,
+      "loss": 1.1589,
+      "step": 2931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.740940593105383e-06,
+      "loss": 1.1059,
+      "step": 2932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738377884903943e-06,
+      "loss": 1.2368,
+      "step": 2933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73581497839162e-06,
+      "loss": 1.083,
+      "step": 2934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.733251874256747e-06,
+      "loss": 0.9607,
+      "step": 2935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.730688573187715e-06,
+      "loss": 1.0335,
+      "step": 2936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7281250758729656e-06,
+      "loss": 1.3263,
+      "step": 2937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725561383000994e-06,
+      "loss": 1.1758,
+      "step": 2938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.722997495260348e-06,
+      "loss": 1.496,
+      "step": 2939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.720433413339627e-06,
+      "loss": 1.3721,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.717869137927483e-06,
+      "loss": 1.3884,
+      "step": 2941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.71530466971262e-06,
+      "loss": 0.8853,
+      "step": 2942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.712740009383793e-06,
+      "loss": 1.2363,
+      "step": 2943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.710175157629812e-06,
+      "loss": 0.9393,
+      "step": 2944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7076101151395315e-06,
+      "loss": 1.1573,
+      "step": 2945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705044882601862e-06,
+      "loss": 0.9477,
+      "step": 2946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7024794607057695e-06,
+      "loss": 1.1838,
+      "step": 2947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.69991385014026e-06,
+      "loss": 1.0842,
+      "step": 2948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697348051594399e-06,
+      "loss": 0.9878,
+      "step": 2949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.694782065757298e-06,
+      "loss": 1.2003,
+      "step": 2950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.692215893318121e-06,
+      "loss": 1.0107,
+      "step": 2951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.689649534966083e-06,
+      "loss": 0.9393,
+      "step": 2952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.687082991390443e-06,
+      "loss": 1.18,
+      "step": 2953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684516263280519e-06,
+      "loss": 1.3162,
+      "step": 2954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.68194935132567e-06,
+      "loss": 1.2683,
+      "step": 2955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.679382256215311e-06,
+      "loss": 0.9013,
+      "step": 2956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6768149786389e-06,
+      "loss": 1.3536,
+      "step": 2957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.674247519285951e-06,
+      "loss": 1.1732,
+      "step": 2958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.671679878846017e-06,
+      "loss": 1.1994,
+      "step": 2959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6691120580087126e-06,
+      "loss": 1.2559,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6665440574636895e-06,
+      "loss": 1.2849,
+      "step": 2961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6639758779006535e-06,
+      "loss": 1.1466,
+      "step": 2962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.661407520009358e-06,
+      "loss": 1.4137,
+      "step": 2963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6588389844796e-06,
+      "loss": 1.2053,
+      "step": 2964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656270272001232e-06,
+      "loss": 1.1982,
+      "step": 2965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.653701383264147e-06,
+      "loss": 1.1421,
+      "step": 2966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651132318958288e-06,
+      "loss": 1.1756,
+      "step": 2967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648563079773646e-06,
+      "loss": 1.0172,
+      "step": 2968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645993666400256e-06,
+      "loss": 1.3621,
+      "step": 2969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6434240795282045e-06,
+      "loss": 0.9656,
+      "step": 2970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6408543198476204e-06,
+      "loss": 1.2173,
+      "step": 2971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63828438804868e-06,
+      "loss": 1.0267,
+      "step": 2972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63571428482161e-06,
+      "loss": 1.1145,
+      "step": 2973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6331440108566735e-06,
+      "loss": 1.0423,
+      "step": 2974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.630573566844191e-06,
+      "loss": 1.2485,
+      "step": 2975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.628002953474521e-06,
+      "loss": 1.2606,
+      "step": 2976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625432171438068e-06,
+      "loss": 1.2041,
+      "step": 2977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.622861221425286e-06,
+      "loss": 1.287,
+      "step": 2978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.620290104126669e-06,
+      "loss": 1.0499,
+      "step": 2979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.617718820232762e-06,
+      "loss": 0.8083,
+      "step": 2980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.615147370434147e-06,
+      "loss": 1.279,
+      "step": 2981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.612575755421459e-06,
+      "loss": 1.1598,
+      "step": 2982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61000397588537e-06,
+      "loss": 1.0594,
+      "step": 2983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.607432032516601e-06,
+      "loss": 1.0281,
+      "step": 2984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.604859926005913e-06,
+      "loss": 0.9886,
+      "step": 2985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.602287657044116e-06,
+      "loss": 1.0072,
+      "step": 2986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.59971522632206e-06,
+      "loss": 0.8619,
+      "step": 2987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597142634530639e-06,
+      "loss": 1.5137,
+      "step": 2988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.594569882360788e-06,
+      "loss": 1.3302,
+      "step": 2989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5919969705034914e-06,
+      "loss": 1.1364,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5894238996497715e-06,
+      "loss": 1.1525,
+      "step": 2991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.586850670490694e-06,
+      "loss": 1.2559,
+      "step": 2992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.584277283717367e-06,
+      "loss": 1.0004,
+      "step": 2993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581703740020943e-06,
+      "loss": 1.2068,
+      "step": 2994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.579130040092613e-06,
+      "loss": 1.2143,
+      "step": 2995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.576556184623615e-06,
+      "loss": 1.4352,
+      "step": 2996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.573982174305223e-06,
+      "loss": 1.2887,
+      "step": 2997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.571408009828757e-06,
+      "loss": 1.2839,
+      "step": 2998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.568833691885576e-06,
+      "loss": 1.0589,
+      "step": 2999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.56625922116708e-06,
+      "loss": 0.9333,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.563684598364717e-06,
+      "loss": 1.0621,
+      "step": 3001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.561109824169962e-06,
+      "loss": 0.9508,
+      "step": 3002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.558534899274344e-06,
+      "loss": 0.9824,
+      "step": 3003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.555959824369426e-06,
+      "loss": 1.2521,
+      "step": 3004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.55338460014681e-06,
+      "loss": 1.546,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550809227298144e-06,
+      "loss": 1.2777,
+      "step": 3006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548233706515109e-06,
+      "loss": 1.0248,
+      "step": 3007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.545658038489433e-06,
+      "loss": 0.9262,
+      "step": 3008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.543082223912875e-06,
+      "loss": 1.227,
+      "step": 3009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.540506263477243e-06,
+      "loss": 1.3781,
+      "step": 3010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.537930157874375e-06,
+      "loss": 1.0925,
+      "step": 3011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535353907796155e-06,
+      "loss": 1.1422,
+      "step": 3012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532777513934502e-06,
+      "loss": 1.0412,
+      "step": 3013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.530200976981375e-06,
+      "loss": 0.7671,
+      "step": 3014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.527624297628768e-06,
+      "loss": 1.0635,
+      "step": 3015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525047476568722e-06,
+      "loss": 1.2962,
+      "step": 3016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522470514493306e-06,
+      "loss": 1.2822,
+      "step": 3017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.519893412094631e-06,
+      "loss": 1.1042,
+      "step": 3018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.51731617006485e-06,
+      "loss": 1.2849,
+      "step": 3019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.514738789096146e-06,
+      "loss": 1.691,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5121612698807435e-06,
+      "loss": 1.1596,
+      "step": 3021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509583613110904e-06,
+      "loss": 0.973,
+      "step": 3022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.507005819478924e-06,
+      "loss": 0.839,
+      "step": 3023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.504427889677141e-06,
+      "loss": 1.0551,
+      "step": 3024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5018498243979216e-06,
+      "loss": 1.1772,
+      "step": 3025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.499271624333676e-06,
+      "loss": 1.3605,
+      "step": 3026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.496693290176849e-06,
+      "loss": 1.2769,
+      "step": 3027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494114822619918e-06,
+      "loss": 0.9487,
+      "step": 3028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4915362223554e-06,
+      "loss": 1.0584,
+      "step": 3029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488957490075846e-06,
+      "loss": 0.87,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.486378626473842e-06,
+      "loss": 0.8297,
+      "step": 3031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483799632242012e-06,
+      "loss": 1.1847,
+      "step": 3032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.481220508073009e-06,
+      "loss": 1.079,
+      "step": 3033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478641254659528e-06,
+      "loss": 1.1535,
+      "step": 3034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.476061872694296e-06,
+      "loss": 1.3601,
+      "step": 3035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473482362870073e-06,
+      "loss": 0.9582,
+      "step": 3036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.470902725879656e-06,
+      "loss": 1.1008,
+      "step": 3037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.468322962415871e-06,
+      "loss": 1.0386,
+      "step": 3038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.465743073171584e-06,
+      "loss": 0.9377,
+      "step": 3039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.463163058839694e-06,
+      "loss": 0.9073,
+      "step": 3040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.46058292011313e-06,
+      "loss": 1.1804,
+      "step": 3041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4580026576848565e-06,
+      "loss": 1.0375,
+      "step": 3042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.455422272247871e-06,
+      "loss": 1.4804,
+      "step": 3043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.452841764495203e-06,
+      "loss": 1.2924,
+      "step": 3044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450261135119921e-06,
+      "loss": 1.035,
+      "step": 3045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4476803848151146e-06,
+      "loss": 1.4844,
+      "step": 3046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.445099514273917e-06,
+      "loss": 1.1499,
+      "step": 3047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442518524189489e-06,
+      "loss": 1.3416,
+      "step": 3048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.439937415255021e-06,
+      "loss": 1.479,
+      "step": 3049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4373561881637405e-06,
+      "loss": 1.2302,
+      "step": 3050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.434774843608901e-06,
+      "loss": 1.041,
+      "step": 3051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.432193382283794e-06,
+      "loss": 0.904,
+      "step": 3052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.429611804881739e-06,
+      "loss": 1.0129,
+      "step": 3053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4270301120960856e-06,
+      "loss": 0.9948,
+      "step": 3054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424448304620215e-06,
+      "loss": 1.2524,
+      "step": 3055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.421866383147541e-06,
+      "loss": 1.2489,
+      "step": 3056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4192843483715064e-06,
+      "loss": 0.9594,
+      "step": 3057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416702200985585e-06,
+      "loss": 1.1853,
+      "step": 3058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41411994168328e-06,
+      "loss": 1.1974,
+      "step": 3059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411537571158127e-06,
+      "loss": 1.0449,
+      "step": 3060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.408955090103688e-06,
+      "loss": 1.2154,
+      "step": 3061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.406372499213557e-06,
+      "loss": 0.9364,
+      "step": 3062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.403789799181358e-06,
+      "loss": 1.0508,
+      "step": 3063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.401206990700741e-06,
+      "loss": 1.1967,
+      "step": 3064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.398624074465391e-06,
+      "loss": 1.1963,
+      "step": 3065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396041051169016e-06,
+      "loss": 1.2442,
+      "step": 3066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.393457921505355e-06,
+      "loss": 1.4519,
+      "step": 3067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390874686168176e-06,
+      "loss": 1.0362,
+      "step": 3068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3882913458512766e-06,
+      "loss": 1.0674,
+      "step": 3069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385707901248478e-06,
+      "loss": 1.0844,
+      "step": 3070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383124353053636e-06,
+      "loss": 0.8278,
+      "step": 3071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380540701960627e-06,
+      "loss": 1.4636,
+      "step": 3072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3779569486633624e-06,
+      "loss": 1.1985,
+      "step": 3073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.375373093855774e-06,
+      "loss": 1.049,
+      "step": 3074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372789138231827e-06,
+      "loss": 1.0741,
+      "step": 3075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.37020508248551e-06,
+      "loss": 1.2127,
+      "step": 3076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.367620927310839e-06,
+      "loss": 1.028,
+      "step": 3077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365036673401857e-06,
+      "loss": 0.9801,
+      "step": 3078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.362452321452636e-06,
+      "loss": 0.9936,
+      "step": 3079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359867872157267e-06,
+      "loss": 1.1208,
+      "step": 3080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357283326209879e-06,
+      "loss": 1.0279,
+      "step": 3081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.354698684304613e-06,
+      "loss": 1.0857,
+      "step": 3082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.352113947135648e-06,
+      "loss": 1.1836,
+      "step": 3083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3495291153971806e-06,
+      "loss": 1.0619,
+      "step": 3084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.346944189783437e-06,
+      "loss": 1.1804,
+      "step": 3085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344359170988668e-06,
+      "loss": 1.2892,
+      "step": 3086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341774059707147e-06,
+      "loss": 1.3354,
+      "step": 3087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339188856633173e-06,
+      "loss": 1.0558,
+      "step": 3088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336603562461074e-06,
+      "loss": 1.1833,
+      "step": 3089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3340181778851954e-06,
+      "loss": 1.0532,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.331432703599914e-06,
+      "loss": 1.0001,
+      "step": 3091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.328847140299624e-06,
+      "loss": 1.2745,
+      "step": 3092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326261488678749e-06,
+      "loss": 1.1197,
+      "step": 3093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.323675749431732e-06,
+      "loss": 1.1317,
+      "step": 3094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.321089923253046e-06,
+      "loss": 1.2694,
+      "step": 3095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.318504010837175e-06,
+      "loss": 1.1589,
+      "step": 3096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3159180128786424e-06,
+      "loss": 1.4024,
+      "step": 3097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.313331930071981e-06,
+      "loss": 1.0911,
+      "step": 3098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.310745763111754e-06,
+      "loss": 1.0157,
+      "step": 3099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308159512692544e-06,
+      "loss": 1.0622,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.305573179508956e-06,
+      "loss": 1.0594,
+      "step": 3101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.302986764255621e-06,
+      "loss": 1.2197,
+      "step": 3102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.300400267627186e-06,
+      "loss": 1.2751,
+      "step": 3103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.297813690318325e-06,
+      "loss": 1.4772,
+      "step": 3104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2952270330237295e-06,
+      "loss": 0.9799,
+      "step": 3105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.292640296438116e-06,
+      "loss": 1.0888,
+      "step": 3106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.290053481256223e-06,
+      "loss": 0.99,
+      "step": 3107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287466588172804e-06,
+      "loss": 1.1691,
+      "step": 3108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.28487961788264e-06,
+      "loss": 0.9573,
+      "step": 3109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2822925710805305e-06,
+      "loss": 1.0802,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279705448461293e-06,
+      "loss": 0.9707,
+      "step": 3111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27711825071977e-06,
+      "loss": 1.2094,
+      "step": 3112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274530978550822e-06,
+      "loss": 1.1489,
+      "step": 3113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2719436326493255e-06,
+      "loss": 1.0844,
+      "step": 3114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.269356213710185e-06,
+      "loss": 1.0107,
+      "step": 3115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266768722428318e-06,
+      "loss": 0.9319,
+      "step": 3116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.264181159498665e-06,
+      "loss": 1.0785,
+      "step": 3117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.261593525616181e-06,
+      "loss": 0.9261,
+      "step": 3118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259005821475847e-06,
+      "loss": 1.0549,
+      "step": 3119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256418047772659e-06,
+      "loss": 1.2602,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25383020520163e-06,
+      "loss": 1.1854,
+      "step": 3121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.251242294457796e-06,
+      "loss": 1.1426,
+      "step": 3122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.248654316236207e-06,
+      "loss": 1.2859,
+      "step": 3123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2460662712319335e-06,
+      "loss": 1.2157,
+      "step": 3124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.243478160140063e-06,
+      "loss": 1.111,
+      "step": 3125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.240889983655701e-06,
+      "loss": 1.0224,
+      "step": 3126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.238301742473969e-06,
+      "loss": 1.2264,
+      "step": 3127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.235713437290012e-06,
+      "loss": 1.0542,
+      "step": 3128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.233125068798983e-06,
+      "loss": 1.1913,
+      "step": 3129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.230536637696062e-06,
+      "loss": 1.1495,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227948144676434e-06,
+      "loss": 1.3771,
+      "step": 3131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.225359590435312e-06,
+      "loss": 1.0435,
+      "step": 3132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2227709756679215e-06,
+      "loss": 1.1052,
+      "step": 3133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220182301069499e-06,
+      "loss": 1.1197,
+      "step": 3134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.217593567335304e-06,
+      "loss": 1.192,
+      "step": 3135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.215004775160608e-06,
+      "loss": 1.0788,
+      "step": 3136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2124159252407e-06,
+      "loss": 1.2536,
+      "step": 3137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209827018270886e-06,
+      "loss": 1.0251,
+      "step": 3138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.207238054946483e-06,
+      "loss": 1.0758,
+      "step": 3139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.204649035962825e-06,
+      "loss": 0.9948,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.202059962015266e-06,
+      "loss": 0.9496,
+      "step": 3141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.199470833799164e-06,
+      "loss": 1.3037,
+      "step": 3142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196881652009903e-06,
+      "loss": 1.3232,
+      "step": 3143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1942924173428725e-06,
+      "loss": 1.3087,
+      "step": 3144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1917031304934805e-06,
+      "loss": 1.2722,
+      "step": 3145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.18911379215715e-06,
+      "loss": 1.0509,
+      "step": 3146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.186524403029315e-06,
+      "loss": 1.3807,
+      "step": 3147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1839349638054245e-06,
+      "loss": 1.2853,
+      "step": 3148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181345475180941e-06,
+      "loss": 1.1689,
+      "step": 3149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.178755937851341e-06,
+      "loss": 0.9564,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.176166352512111e-06,
+      "loss": 1.0837,
+      "step": 3151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.173576719858755e-06,
+      "loss": 1.177,
+      "step": 3152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170987040586784e-06,
+      "loss": 1.1626,
+      "step": 3153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.168397315391729e-06,
+      "loss": 1.0035,
+      "step": 3154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.165807544969125e-06,
+      "loss": 1.0515,
+      "step": 3155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1632177300145255e-06,
+      "loss": 1.3149,
+      "step": 3156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.160627871223493e-06,
+      "loss": 1.1955,
+      "step": 3157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1580379692916025e-06,
+      "loss": 1.0611,
+      "step": 3158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.155448024914442e-06,
+      "loss": 1.1719,
+      "step": 3159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152858038787608e-06,
+      "loss": 0.8195,
+      "step": 3160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150268011606711e-06,
+      "loss": 0.9905,
+      "step": 3161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.147677944067368e-06,
+      "loss": 1.0742,
+      "step": 3162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.145087836865213e-06,
+      "loss": 1.0381,
+      "step": 3163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142497690695888e-06,
+      "loss": 1.4661,
+      "step": 3164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.139907506255043e-06,
+      "loss": 1.3047,
+      "step": 3165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137317284238344e-06,
+      "loss": 0.9332,
+      "step": 3166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.134727025341458e-06,
+      "loss": 1.1271,
+      "step": 3167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1321367302600726e-06,
+      "loss": 1.0521,
+      "step": 3168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.129546399689878e-06,
+      "loss": 0.9529,
+      "step": 3169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.126956034326573e-06,
+      "loss": 0.997,
+      "step": 3170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1243656348658744e-06,
+      "loss": 1.0733,
+      "step": 3171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121775202003499e-06,
+      "loss": 0.8724,
+      "step": 3172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1191847364351765e-06,
+      "loss": 1.0593,
+      "step": 3173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116594238856645e-06,
+      "loss": 1.2661,
+      "step": 3174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114003709963649e-06,
+      "loss": 1.2313,
+      "step": 3175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111413150451948e-06,
+      "loss": 1.1048,
+      "step": 3176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108822561017302e-06,
+      "loss": 1.0362,
+      "step": 3177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1062319423554815e-06,
+      "loss": 0.9587,
+      "step": 3178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.103641295162267e-06,
+      "loss": 1.1104,
+      "step": 3179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.101050620133447e-06,
+      "loss": 0.9725,
+      "step": 3180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.098459917964812e-06,
+      "loss": 1.1143,
+      "step": 3181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.095869189352166e-06,
+      "loss": 1.1081,
+      "step": 3182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.093278434991315e-06,
+      "loss": 1.0899,
+      "step": 3183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090687655578078e-06,
+      "loss": 0.9636,
+      "step": 3184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.088096851808276e-06,
+      "loss": 1.2841,
+      "step": 3185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0855060243777366e-06,
+      "loss": 1.0703,
+      "step": 3186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0829151739822964e-06,
+      "loss": 1.4639,
+      "step": 3187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080324301317795e-06,
+      "loss": 1.0132,
+      "step": 3188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.077733407080081e-06,
+      "loss": 1.2351,
+      "step": 3189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0751424919650085e-06,
+      "loss": 1.2272,
+      "step": 3190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.072551556668434e-06,
+      "loss": 1.124,
+      "step": 3191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069960601886224e-06,
+      "loss": 1.1536,
+      "step": 3192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.067369628314244e-06,
+      "loss": 1.135,
+      "step": 3193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.064778636648371e-06,
+      "loss": 1.2125,
+      "step": 3194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.062187627584486e-06,
+      "loss": 0.9428,
+      "step": 3195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.05959660181847e-06,
+      "loss": 1.2556,
+      "step": 3196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.057005560046211e-06,
+      "loss": 1.0727,
+      "step": 3197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054414502963605e-06,
+      "loss": 1.2098,
+      "step": 3198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.051823431266546e-06,
+      "loss": 1.1625,
+      "step": 3199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.049232345650936e-06,
+      "loss": 1.2143,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.046641246812678e-06,
+      "loss": 0.9835,
+      "step": 3201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.044050135447682e-06,
+      "loss": 1.2682,
+      "step": 3202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.041459012251859e-06,
+      "loss": 0.9995,
+      "step": 3203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038867877921124e-06,
+      "loss": 1.0748,
+      "step": 3204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.036276733151392e-06,
+      "loss": 1.1606,
+      "step": 3205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.033685578638586e-06,
+      "loss": 0.8947,
+      "step": 3206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.031094415078628e-06,
+      "loss": 0.9662,
+      "step": 3207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.028503243167443e-06,
+      "loss": 0.7643,
+      "step": 3208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0259120636009596e-06,
+      "loss": 1.1659,
+      "step": 3209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023320877075107e-06,
+      "loss": 1.3092,
+      "step": 3210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.020729684285817e-06,
+      "loss": 1.3883,
+      "step": 3211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0181384859290215e-06,
+      "loss": 1.0639,
+      "step": 3212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.015547282700658e-06,
+      "loss": 0.9419,
+      "step": 3213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.01295607529666e-06,
+      "loss": 1.0204,
+      "step": 3214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.010364864412966e-06,
+      "loss": 1.131,
+      "step": 3215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.007773650745514e-06,
+      "loss": 1.2141,
+      "step": 3216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.005182434990242e-06,
+      "loss": 1.2615,
+      "step": 3217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0025912178430925e-06,
+      "loss": 1.1517,
+      "step": 3218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.8857,
+      "step": 3219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.997408782156909e-06,
+      "loss": 1.3265,
+      "step": 3220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.994817565009758e-06,
+      "loss": 1.0494,
+      "step": 3221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9922263492544885e-06,
+      "loss": 1.1799,
+      "step": 3222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.989635135587036e-06,
+      "loss": 1.2585,
+      "step": 3223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.987043924703342e-06,
+      "loss": 1.2112,
+      "step": 3224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.984452717299344e-06,
+      "loss": 1.3751,
+      "step": 3225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.981861514070979e-06,
+      "loss": 1.09,
+      "step": 3226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.979270315714187e-06,
+      "loss": 1.378,
+      "step": 3227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976679122924896e-06,
+      "loss": 1.2178,
+      "step": 3228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974087936399042e-06,
+      "loss": 1.0864,
+      "step": 3229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.971496756832557e-06,
+      "loss": 0.8009,
+      "step": 3230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.968905584921374e-06,
+      "loss": 0.9135,
+      "step": 3231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.966314421361416e-06,
+      "loss": 1.2485,
+      "step": 3232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.963723266848609e-06,
+      "loss": 0.9962,
+      "step": 3233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9611321220788775e-06,
+      "loss": 1.1287,
+      "step": 3234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9585409877481424e-06,
+      "loss": 1.0091,
+      "step": 3235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.955949864552318e-06,
+      "loss": 1.1507,
+      "step": 3236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.953358753187322e-06,
+      "loss": 0.9746,
+      "step": 3237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.950767654349067e-06,
+      "loss": 1.054,
+      "step": 3238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.948176568733456e-06,
+      "loss": 0.8394,
+      "step": 3239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945585497036396e-06,
+      "loss": 1.0526,
+      "step": 3240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.942994439953789e-06,
+      "loss": 1.0657,
+      "step": 3241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.940403398181531e-06,
+      "loss": 0.7147,
+      "step": 3242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.937812372415517e-06,
+      "loss": 1.0339,
+      "step": 3243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.935221363351631e-06,
+      "loss": 1.2555,
+      "step": 3244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.932630371685758e-06,
+      "loss": 1.2104,
+      "step": 3245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930039398113779e-06,
+      "loss": 1.2027,
+      "step": 3246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927448443331567e-06,
+      "loss": 1.0211,
+      "step": 3247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.924857508034994e-06,
+      "loss": 1.0812,
+      "step": 3248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9222665929199195e-06,
+      "loss": 0.8809,
+      "step": 3249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919675698682206e-06,
+      "loss": 0.8311,
+      "step": 3250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917084826017704e-06,
+      "loss": 1.0487,
+      "step": 3251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914493975622263e-06,
+      "loss": 1.0147,
+      "step": 3252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.911903148191726e-06,
+      "loss": 0.9312,
+      "step": 3253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909312344421923e-06,
+      "loss": 1.3829,
+      "step": 3254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.906721565008686e-06,
+      "loss": 1.0181,
+      "step": 3255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.904130810647836e-06,
+      "loss": 0.8743,
+      "step": 3256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.90154008203519e-06,
+      "loss": 1.0781,
+      "step": 3257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.898949379866556e-06,
+      "loss": 1.0752,
+      "step": 3258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.896358704837734e-06,
+      "loss": 0.9388,
+      "step": 3259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.893768057644519e-06,
+      "loss": 0.9703,
+      "step": 3260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8911774389827e-06,
+      "loss": 1.0734,
+      "step": 3261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.888586849548053e-06,
+      "loss": 1.08,
+      "step": 3262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8859962900363515e-06,
+      "loss": 1.2405,
+      "step": 3263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883405761143357e-06,
+      "loss": 1.0719,
+      "step": 3264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.880815263564825e-06,
+      "loss": 1.0954,
+      "step": 3265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.878224797996502e-06,
+      "loss": 0.9571,
+      "step": 3266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8756343651341255e-06,
+      "loss": 0.8799,
+      "step": 3267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.873043965673427e-06,
+      "loss": 1.1062,
+      "step": 3268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8704536003101255e-06,
+      "loss": 1.1032,
+      "step": 3269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.86786326973993e-06,
+      "loss": 1.2753,
+      "step": 3270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.865272974658543e-06,
+      "loss": 1.2712,
+      "step": 3271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.862682715761658e-06,
+      "loss": 0.7928,
+      "step": 3272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.860092493744957e-06,
+      "loss": 1.5432,
+      "step": 3273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.857502309304114e-06,
+      "loss": 1.2657,
+      "step": 3274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.854912163134788e-06,
+      "loss": 0.8474,
+      "step": 3275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.852322055932633e-06,
+      "loss": 1.2802,
+      "step": 3276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849731988393291e-06,
+      "loss": 1.1298,
+      "step": 3277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8471419612123925e-06,
+      "loss": 0.8967,
+      "step": 3278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84455197508556e-06,
+      "loss": 1.2221,
+      "step": 3279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841962030708398e-06,
+      "loss": 1.2962,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8393721287765085e-06,
+      "loss": 1.1143,
+      "step": 3281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836782269985475e-06,
+      "loss": 1.1937,
+      "step": 3282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.834192455030876e-06,
+      "loss": 1.1835,
+      "step": 3283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.831602684608274e-06,
+      "loss": 1.1891,
+      "step": 3284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829012959413217e-06,
+      "loss": 1.0687,
+      "step": 3285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.826423280141247e-06,
+      "loss": 1.0946,
+      "step": 3286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82383364748789e-06,
+      "loss": 1.0018,
+      "step": 3287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82124406214866e-06,
+      "loss": 1.3732,
+      "step": 3288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8186545248190604e-06,
+      "loss": 1.1505,
+      "step": 3289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816065036194576e-06,
+      "loss": 1.2252,
+      "step": 3290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813475596970687e-06,
+      "loss": 1.3086,
+      "step": 3291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.810886207842852e-06,
+      "loss": 0.9928,
+      "step": 3292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.80829686950652e-06,
+      "loss": 1.0968,
+      "step": 3293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.80570758265713e-06,
+      "loss": 1.2403,
+      "step": 3294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8031183479901e-06,
+      "loss": 0.8596,
+      "step": 3295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.800529166200837e-06,
+      "loss": 1.0261,
+      "step": 3296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797940037984736e-06,
+      "loss": 1.2552,
+      "step": 3297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795350964037174e-06,
+      "loss": 0.9902,
+      "step": 3298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.792761945053518e-06,
+      "loss": 0.9665,
+      "step": 3299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.790172981729116e-06,
+      "loss": 1.1836,
+      "step": 3300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.787584074759301e-06,
+      "loss": 1.2009,
+      "step": 3301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.784995224839394e-06,
+      "loss": 1.2382,
+      "step": 3302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.782406432664698e-06,
+      "loss": 0.9908,
+      "step": 3303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779817698930502e-06,
+      "loss": 1.2543,
+      "step": 3304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777229024332082e-06,
+      "loss": 1.2751,
+      "step": 3305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774640409564688e-06,
+      "loss": 1.1302,
+      "step": 3306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7720518553235666e-06,
+      "loss": 1.1781,
+      "step": 3307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76946336230394e-06,
+      "loss": 1.0743,
+      "step": 3308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766874931201017e-06,
+      "loss": 0.9859,
+      "step": 3309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76428656270999e-06,
+      "loss": 0.9316,
+      "step": 3310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7616982575260325e-06,
+      "loss": 1.0894,
+      "step": 3311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.759110016344302e-06,
+      "loss": 1.1622,
+      "step": 3312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.756521839859939e-06,
+      "loss": 0.9999,
+      "step": 3313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753933728768069e-06,
+      "loss": 1.1636,
+      "step": 3314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751345683763795e-06,
+      "loss": 1.1265,
+      "step": 3315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.748757705542205e-06,
+      "loss": 1.183,
+      "step": 3316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746169794798371e-06,
+      "loss": 1.0122,
+      "step": 3317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743581952227342e-06,
+      "loss": 0.9094,
+      "step": 3318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740994178524153e-06,
+      "loss": 1.1176,
+      "step": 3319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73840647438382e-06,
+      "loss": 1.2029,
+      "step": 3320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735818840501339e-06,
+      "loss": 1.0087,
+      "step": 3321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733231277571683e-06,
+      "loss": 1.0286,
+      "step": 3322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.730643786289816e-06,
+      "loss": 1.2596,
+      "step": 3323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7280563673506745e-06,
+      "loss": 1.1351,
+      "step": 3324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.725469021449179e-06,
+      "loss": 1.4661,
+      "step": 3325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722881749280232e-06,
+      "loss": 1.21,
+      "step": 3326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7202945515387085e-06,
+      "loss": 1.058,
+      "step": 3327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717707428919471e-06,
+      "loss": 1.1046,
+      "step": 3328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715120382117361e-06,
+      "loss": 1.1166,
+      "step": 3329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.712533411827197e-06,
+      "loss": 1.1501,
+      "step": 3330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70994651874378e-06,
+      "loss": 1.2691,
+      "step": 3331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707359703561885e-06,
+      "loss": 1.1866,
+      "step": 3332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.704772966976272e-06,
+      "loss": 0.9646,
+      "step": 3333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.702186309681677e-06,
+      "loss": 1.1465,
+      "step": 3334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.699599732372815e-06,
+      "loss": 1.3858,
+      "step": 3335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.697013235744382e-06,
+      "loss": 1.0767,
+      "step": 3336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.694426820491045e-06,
+      "loss": 1.2208,
+      "step": 3337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6918404873074574e-06,
+      "loss": 0.996,
+      "step": 3338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.689254236888247e-06,
+      "loss": 1.2997,
+      "step": 3339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68666806992802e-06,
+      "loss": 0.8596,
+      "step": 3340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68408198712136e-06,
+      "loss": 0.9703,
+      "step": 3341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.681495989162826e-06,
+      "loss": 1.1707,
+      "step": 3342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.678910076746957e-06,
+      "loss": 0.7708,
+      "step": 3343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.676324250568269e-06,
+      "loss": 0.9041,
+      "step": 3344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.673738511321252e-06,
+      "loss": 1.4249,
+      "step": 3345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.671152859700377e-06,
+      "loss": 1.2114,
+      "step": 3346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.668567296400088e-06,
+      "loss": 1.1601,
+      "step": 3347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665981822114805e-06,
+      "loss": 1.2015,
+      "step": 3348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663396437538928e-06,
+      "loss": 1.5078,
+      "step": 3349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660811143366828e-06,
+      "loss": 1.2451,
+      "step": 3350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.658225940292857e-06,
+      "loss": 1.1335,
+      "step": 3351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.655640829011335e-06,
+      "loss": 1.1058,
+      "step": 3352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.653055810216564e-06,
+      "loss": 1.078,
+      "step": 3353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65047088460282e-06,
+      "loss": 1.2433,
+      "step": 3354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.647886052864353e-06,
+      "loss": 1.4397,
+      "step": 3355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.645301315695387e-06,
+      "loss": 1.1341,
+      "step": 3356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6427166737901245e-06,
+      "loss": 0.9828,
+      "step": 3357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6401321278427334e-06,
+      "loss": 0.813,
+      "step": 3358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.637547678547366e-06,
+      "loss": 1.1279,
+      "step": 3359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634963326598143e-06,
+      "loss": 1.1773,
+      "step": 3360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6323790726891615e-06,
+      "loss": 0.9566,
+      "step": 3361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.629794917514492e-06,
+      "loss": 1.0631,
+      "step": 3362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627210861768175e-06,
+      "loss": 1.0816,
+      "step": 3363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.624626906144227e-06,
+      "loss": 0.9813,
+      "step": 3364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.622043051336639e-06,
+      "loss": 1.2012,
+      "step": 3365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619459298039373e-06,
+      "loss": 0.9988,
+      "step": 3366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616875646946367e-06,
+      "loss": 1.368,
+      "step": 3367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614292098751524e-06,
+      "loss": 1.3726,
+      "step": 3368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611708654148725e-06,
+      "loss": 1.1682,
+      "step": 3369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609125313831826e-06,
+      "loss": 1.0886,
+      "step": 3370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606542078494646e-06,
+      "loss": 1.0114,
+      "step": 3371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603958948830985e-06,
+      "loss": 1.2046,
+      "step": 3372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601375925534609e-06,
+      "loss": 0.9526,
+      "step": 3373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5987930092992596e-06,
+      "loss": 1.6075,
+      "step": 3374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.596210200818643e-06,
+      "loss": 1.195,
+      "step": 3375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.593627500786444e-06,
+      "loss": 1.0319,
+      "step": 3376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.591044909896313e-06,
+      "loss": 1.0879,
+      "step": 3377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588462428841875e-06,
+      "loss": 0.9528,
+      "step": 3378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585880058316722e-06,
+      "loss": 1.1315,
+      "step": 3379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5832977990144165e-06,
+      "loss": 1.0836,
+      "step": 3380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.580715651628494e-06,
+      "loss": 1.1476,
+      "step": 3381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.578133616852462e-06,
+      "loss": 1.3174,
+      "step": 3382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.575551695379787e-06,
+      "loss": 1.0298,
+      "step": 3383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.572969887903916e-06,
+      "loss": 0.9037,
+      "step": 3384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.570388195118262e-06,
+      "loss": 1.1765,
+      "step": 3385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5678066177162065e-06,
+      "loss": 1.1845,
+      "step": 3386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565225156391099e-06,
+      "loss": 1.2091,
+      "step": 3387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562643811836263e-06,
+      "loss": 1.2729,
+      "step": 3388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.560062584744981e-06,
+      "loss": 1.3217,
+      "step": 3389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.557481475810512e-06,
+      "loss": 0.9114,
+      "step": 3390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.554900485726083e-06,
+      "loss": 1.0323,
+      "step": 3391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5523196151848846e-06,
+      "loss": 1.0772,
+      "step": 3392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549738864880082e-06,
+      "loss": 1.1604,
+      "step": 3393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547158235504797e-06,
+      "loss": 1.1239,
+      "step": 3394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.54457772775213e-06,
+      "loss": 1.0928,
+      "step": 3395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.541997342315145e-06,
+      "loss": 1.4959,
+      "step": 3396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.539417079886872e-06,
+      "loss": 1.1472,
+      "step": 3397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536836941160308e-06,
+      "loss": 1.1507,
+      "step": 3398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.534256926828417e-06,
+      "loss": 1.0992,
+      "step": 3399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5316770375841315e-06,
+      "loss": 1.0654,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.529097274120347e-06,
+      "loss": 1.1549,
+      "step": 3401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.526517637129927e-06,
+      "loss": 1.2342,
+      "step": 3402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523938127305705e-06,
+      "loss": 1.2437,
+      "step": 3403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5213587453404736e-06,
+      "loss": 1.0582,
+      "step": 3404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518779491926992e-06,
+      "loss": 0.9466,
+      "step": 3405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5162003677579905e-06,
+      "loss": 1.1334,
+      "step": 3406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513621373526159e-06,
+      "loss": 1.2139,
+      "step": 3407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511042509924157e-06,
+      "loss": 0.9551,
+      "step": 3408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5084637776446026e-06,
+      "loss": 1.3642,
+      "step": 3409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505885177380083e-06,
+      "loss": 1.2119,
+      "step": 3410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.503306709823153e-06,
+      "loss": 0.8467,
+      "step": 3411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5007283756663245e-06,
+      "loss": 1.1585,
+      "step": 3412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.498150175602081e-06,
+      "loss": 0.9179,
+      "step": 3413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495572110322862e-06,
+      "loss": 1.092,
+      "step": 3414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4929941805210765e-06,
+      "loss": 1.26,
+      "step": 3415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490416386889097e-06,
+      "loss": 1.1168,
+      "step": 3416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.487838730119257e-06,
+      "loss": 1.1143,
+      "step": 3417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485261210903854e-06,
+      "loss": 1.13,
+      "step": 3418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.482683829935152e-06,
+      "loss": 1.3796,
+      "step": 3419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.48010658790537e-06,
+      "loss": 1.0084,
+      "step": 3420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477529485506696e-06,
+      "loss": 1.1196,
+      "step": 3421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47495252343128e-06,
+      "loss": 0.9829,
+      "step": 3422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.472375702371232e-06,
+      "loss": 1.312,
+      "step": 3423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.469799023018628e-06,
+      "loss": 1.2767,
+      "step": 3424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4672224860655e-06,
+      "loss": 1.084,
+      "step": 3425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464646092203846e-06,
+      "loss": 1.0278,
+      "step": 3426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4620698421256265e-06,
+      "loss": 0.9285,
+      "step": 3427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459493736522759e-06,
+      "loss": 1.1708,
+      "step": 3428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4569177760871255e-06,
+      "loss": 1.138,
+      "step": 3429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4543419615105685e-06,
+      "loss": 1.1945,
+      "step": 3430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451766293484892e-06,
+      "loss": 1.0762,
+      "step": 3431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.449190772701857e-06,
+      "loss": 0.9171,
+      "step": 3432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4466153998531904e-06,
+      "loss": 1.0694,
+      "step": 3433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.444040175630577e-06,
+      "loss": 1.0376,
+      "step": 3434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.441465100725658e-06,
+      "loss": 0.945,
+      "step": 3435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.438890175830039e-06,
+      "loss": 0.9259,
+      "step": 3436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436315401635285e-06,
+      "loss": 1.5451,
+      "step": 3437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.433740778832919e-06,
+      "loss": 1.18,
+      "step": 3438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.431166308114427e-06,
+      "loss": 1.2157,
+      "step": 3439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.428591990171246e-06,
+      "loss": 1.0495,
+      "step": 3440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.42601782569478e-06,
+      "loss": 0.9386,
+      "step": 3441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.423443815376387e-06,
+      "loss": 1.0726,
+      "step": 3442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.420869959907388e-06,
+      "loss": 1.2402,
+      "step": 3443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.41829625997906e-06,
+      "loss": 1.148,
+      "step": 3444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.415722716282635e-06,
+      "loss": 0.9065,
+      "step": 3445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.413149329509307e-06,
+      "loss": 1.3061,
+      "step": 3446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.410576100350229e-06,
+      "loss": 1.3403,
+      "step": 3447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4080030294965085e-06,
+      "loss": 1.1839,
+      "step": 3448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.405430117639212e-06,
+      "loss": 1.2231,
+      "step": 3449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.402857365469364e-06,
+      "loss": 0.9221,
+      "step": 3450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.400284773677942e-06,
+      "loss": 0.7947,
+      "step": 3451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397712342955885e-06,
+      "loss": 1.1069,
+      "step": 3452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3951400739940885e-06,
+      "loss": 1.0492,
+      "step": 3453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392567967483401e-06,
+      "loss": 1.1839,
+      "step": 3454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.389996024114632e-06,
+      "loss": 0.6984,
+      "step": 3455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.387424244578543e-06,
+      "loss": 1.1461,
+      "step": 3456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.384852629565854e-06,
+      "loss": 1.1714,
+      "step": 3457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.38228117976724e-06,
+      "loss": 1.0813,
+      "step": 3458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.379709895873331e-06,
+      "loss": 0.9584,
+      "step": 3459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377138778574716e-06,
+      "loss": 1.4428,
+      "step": 3460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3745678285619346e-06,
+      "loss": 1.0833,
+      "step": 3461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371997046525481e-06,
+      "loss": 1.3132,
+      "step": 3462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36942643315581e-06,
+      "loss": 1.0063,
+      "step": 3463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366855989143326e-06,
+      "loss": 1.2985,
+      "step": 3464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3642857151783934e-06,
+      "loss": 0.8136,
+      "step": 3465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3617156119513206e-06,
+      "loss": 1.0003,
+      "step": 3466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.359145680152381e-06,
+      "loss": 0.9939,
+      "step": 3467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356575920471796e-06,
+      "loss": 1.222,
+      "step": 3468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.354006333599744e-06,
+      "loss": 1.1516,
+      "step": 3469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351436920226357e-06,
+      "loss": 0.9077,
+      "step": 3470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.348867681041714e-06,
+      "loss": 1.075,
+      "step": 3471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.346298616735855e-06,
+      "loss": 1.146,
+      "step": 3472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.343729727998769e-06,
+      "loss": 1.2394,
+      "step": 3473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3411610155204e-06,
+      "loss": 1.1322,
+      "step": 3474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.338592479990643e-06,
+      "loss": 1.2711,
+      "step": 3475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.336024122099348e-06,
+      "loss": 1.3663,
+      "step": 3476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.333455942536312e-06,
+      "loss": 1.2259,
+      "step": 3477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.330887941991288e-06,
+      "loss": 1.1834,
+      "step": 3478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.328320121153982e-06,
+      "loss": 0.9325,
+      "step": 3479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.325752480714052e-06,
+      "loss": 1.2157,
+      "step": 3480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.323185021361101e-06,
+      "loss": 1.2219,
+      "step": 3481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.320617743784691e-06,
+      "loss": 1.0446,
+      "step": 3482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.318050648674331e-06,
+      "loss": 1.0433,
+      "step": 3483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315483736719482e-06,
+      "loss": 0.8768,
+      "step": 3484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.312917008609557e-06,
+      "loss": 1.2072,
+      "step": 3485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.310350465033919e-06,
+      "loss": 1.153,
+      "step": 3486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30778410668188e-06,
+      "loss": 0.9737,
+      "step": 3487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.305217934242703e-06,
+      "loss": 1.0602,
+      "step": 3488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.302651948405603e-06,
+      "loss": 1.0672,
+      "step": 3489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30008614985974e-06,
+      "loss": 1.3978,
+      "step": 3490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297520539294233e-06,
+      "loss": 1.2493,
+      "step": 3491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.294955117398139e-06,
+      "loss": 1.1284,
+      "step": 3492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.29238988486047e-06,
+      "loss": 0.9943,
+      "step": 3493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.28982484237019e-06,
+      "loss": 1.1033,
+      "step": 3494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2872599906162075e-06,
+      "loss": 0.9746,
+      "step": 3495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.284695330287383e-06,
+      "loss": 0.9831,
+      "step": 3496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.282130862072519e-06,
+      "loss": 0.8497,
+      "step": 3497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.279566586660375e-06,
+      "loss": 1.0416,
+      "step": 3498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.277002504739653e-06,
+      "loss": 0.949,
+      "step": 3499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.274438616999007e-06,
+      "loss": 1.3626,
+      "step": 3500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.271874924127038e-06,
+      "loss": 1.3085,
+      "step": 3501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269311426812287e-06,
+      "loss": 0.9438,
+      "step": 3502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2667481257432555e-06,
+      "loss": 1.0658,
+      "step": 3503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264185021608382e-06,
+      "loss": 0.9665,
+      "step": 3504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.261622115096058e-06,
+      "loss": 1.0285,
+      "step": 3505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259059406894619e-06,
+      "loss": 1.2433,
+      "step": 3506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2564968976923485e-06,
+      "loss": 0.8073,
+      "step": 3507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253934588177473e-06,
+      "loss": 1.1116,
+      "step": 3508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.251372479038172e-06,
+      "loss": 1.2294,
+      "step": 3509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248810570962567e-06,
+      "loss": 1.0029,
+      "step": 3510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.246248864638724e-06,
+      "loss": 1.1139,
+      "step": 3511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24368736075466e-06,
+      "loss": 1.0694,
+      "step": 3512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.241126059998333e-06,
+      "loss": 1.0083,
+      "step": 3513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238564963057646e-06,
+      "loss": 0.925,
+      "step": 3514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.236004070620452e-06,
+      "loss": 1.0676,
+      "step": 3515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233443383374545e-06,
+      "loss": 1.1974,
+      "step": 3516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.230882902007667e-06,
+      "loss": 0.9102,
+      "step": 3517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.228322627207499e-06,
+      "loss": 1.3405,
+      "step": 3518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225762559661675e-06,
+      "loss": 1.4332,
+      "step": 3519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223202700057765e-06,
+      "loss": 1.2224,
+      "step": 3520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.220643049083289e-06,
+      "loss": 0.9539,
+      "step": 3521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21808360742571e-06,
+      "loss": 0.8282,
+      "step": 3522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.215524375772431e-06,
+      "loss": 1.1544,
+      "step": 3523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.212965354810802e-06,
+      "loss": 1.0753,
+      "step": 3524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.210406545228116e-06,
+      "loss": 1.0515,
+      "step": 3525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.207847947711609e-06,
+      "loss": 1.2346,
+      "step": 3526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205289562948462e-06,
+      "loss": 1.1248,
+      "step": 3527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202731391625793e-06,
+      "loss": 1.0294,
+      "step": 3528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.200173434430667e-06,
+      "loss": 1.1169,
+      "step": 3529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1976156920500935e-06,
+      "loss": 0.8974,
+      "step": 3530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195058165171019e-06,
+      "loss": 0.8308,
+      "step": 3531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.19250085448034e-06,
+      "loss": 1.0429,
+      "step": 3532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1899437606648826e-06,
+      "loss": 1.0082,
+      "step": 3533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.187386884411426e-06,
+      "loss": 1.1558,
+      "step": 3534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184830226406687e-06,
+      "loss": 1.0387,
+      "step": 3535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.182273787337323e-06,
+      "loss": 1.1302,
+      "step": 3536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179717567889933e-06,
+      "loss": 0.832,
+      "step": 3537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.177161568751058e-06,
+      "loss": 1.2,
+      "step": 3538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.174605790607178e-06,
+      "loss": 1.1536,
+      "step": 3539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.172050234144716e-06,
+      "loss": 1.0071,
+      "step": 3540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.169494900050033e-06,
+      "loss": 1.2821,
+      "step": 3541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16693978900943e-06,
+      "loss": 1.2434,
+      "step": 3542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.164384901709154e-06,
+      "loss": 0.9652,
+      "step": 3543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.161830238835386e-06,
+      "loss": 1.1786,
+      "step": 3544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.159275801074244e-06,
+      "loss": 1.0318,
+      "step": 3545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156721589111794e-06,
+      "loss": 0.9536,
+      "step": 3546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.154167603634036e-06,
+      "loss": 0.7792,
+      "step": 3547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151613845326912e-06,
+      "loss": 1.1224,
+      "step": 3548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1490603148762976e-06,
+      "loss": 1.1987,
+      "step": 3549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.146507012968013e-06,
+      "loss": 1.0888,
+      "step": 3550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.143953940287814e-06,
+      "loss": 1.0807,
+      "step": 3551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.141401097521396e-06,
+      "loss": 1.166,
+      "step": 3552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.138848485354396e-06,
+      "loss": 1.2148,
+      "step": 3553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.136296104472378e-06,
+      "loss": 1.0244,
+      "step": 3554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.133743955560857e-06,
+      "loss": 1.0613,
+      "step": 3555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.131192039305278e-06,
+      "loss": 0.9783,
+      "step": 3556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128640356391027e-06,
+      "loss": 1.0558,
+      "step": 3557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1260889075034254e-06,
+      "loss": 1.1805,
+      "step": 3558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.12353769332773e-06,
+      "loss": 1.1775,
+      "step": 3559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120986714549139e-06,
+      "loss": 0.943,
+      "step": 3560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.118435971852783e-06,
+      "loss": 0.857,
+      "step": 3561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.115885465923734e-06,
+      "loss": 0.8834,
+      "step": 3562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1133351974469985e-06,
+      "loss": 1.2297,
+      "step": 3563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.110785167107514e-06,
+      "loss": 1.508,
+      "step": 3564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.108235375590161e-06,
+      "loss": 1.1704,
+      "step": 3565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1056858235797545e-06,
+      "loss": 1.1348,
+      "step": 3566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.103136511761042e-06,
+      "loss": 1.0438,
+      "step": 3567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.100587440818709e-06,
+      "loss": 0.93,
+      "step": 3568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098038611437377e-06,
+      "loss": 1.1288,
+      "step": 3569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0954900243016016e-06,
+      "loss": 0.9519,
+      "step": 3570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.092941680095871e-06,
+      "loss": 1.4101,
+      "step": 3571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090393579504612e-06,
+      "loss": 1.1709,
+      "step": 3572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0878457232121835e-06,
+      "loss": 1.1238,
+      "step": 3573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.085298111902882e-06,
+      "loss": 0.8512,
+      "step": 3574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.082750746260932e-06,
+      "loss": 0.8522,
+      "step": 3575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.080203626970498e-06,
+      "loss": 1.3723,
+      "step": 3576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0776567547156755e-06,
+      "loss": 0.941,
+      "step": 3577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0751101301804945e-06,
+      "loss": 1.3791,
+      "step": 3578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0725637540489214e-06,
+      "loss": 0.9876,
+      "step": 3579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070017627004847e-06,
+      "loss": 1.0643,
+      "step": 3580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0674717497321045e-06,
+      "loss": 1.1835,
+      "step": 3581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0649261229144554e-06,
+      "loss": 1.1976,
+      "step": 3582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0623807472355954e-06,
+      "loss": 1.1308,
+      "step": 3583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059835623379155e-06,
+      "loss": 1.0116,
+      "step": 3584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0572907520286894e-06,
+      "loss": 0.9539,
+      "step": 3585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054746133867693e-06,
+      "loss": 1.166,
+      "step": 3586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0522017695795915e-06,
+      "loss": 1.126,
+      "step": 3587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0496576598477396e-06,
+      "loss": 0.9609,
+      "step": 3588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047113805355428e-06,
+      "loss": 1.1544,
+      "step": 3589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044570206785874e-06,
+      "loss": 1.1433,
+      "step": 3590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0420268648222255e-06,
+      "loss": 1.0502,
+      "step": 3591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.039483780147568e-06,
+      "loss": 0.9945,
+      "step": 3592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0369409534449134e-06,
+      "loss": 1.0192,
+      "step": 3593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0343983853972045e-06,
+      "loss": 1.2885,
+      "step": 3594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0318560766873145e-06,
+      "loss": 1.0732,
+      "step": 3595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029314027998049e-06,
+      "loss": 1.0868,
+      "step": 3596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02677224001214e-06,
+      "loss": 0.9785,
+      "step": 3597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.024230713412253e-06,
+      "loss": 1.3793,
+      "step": 3598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.021689448880983e-06,
+      "loss": 1.1572,
+      "step": 3599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019148447100855e-06,
+      "loss": 1.2076,
+      "step": 3600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.016607708754318e-06,
+      "loss": 1.217,
+      "step": 3601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.014067234523756e-06,
+      "loss": 1.0618,
+      "step": 3602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.011527025091481e-06,
+      "loss": 0.8656,
+      "step": 3603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008987081139734e-06,
+      "loss": 1.0321,
+      "step": 3604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.006447403350686e-06,
+      "loss": 1.1114,
+      "step": 3605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0039079924064285e-06,
+      "loss": 1.0236,
+      "step": 3606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0013688489889914e-06,
+      "loss": 1.1597,
+      "step": 3607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998829973780329e-06,
+      "loss": 0.9936,
+      "step": 3608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9962913674623215e-06,
+      "loss": 0.8168,
+      "step": 3609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.993753030716783e-06,
+      "loss": 1.1718,
+      "step": 3610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991214964225444e-06,
+      "loss": 0.8558,
+      "step": 3611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.988677168669974e-06,
+      "loss": 1.3544,
+      "step": 3612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.986139644731964e-06,
+      "loss": 0.9634,
+      "step": 3613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983602393092931e-06,
+      "loss": 1.4247,
+      "step": 3614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.981065414434327e-06,
+      "loss": 1.102,
+      "step": 3615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.978528709437518e-06,
+      "loss": 1.0511,
+      "step": 3616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.975992278783805e-06,
+      "loss": 0.8942,
+      "step": 3617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973456123154415e-06,
+      "loss": 1.3633,
+      "step": 3618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.970920243230499e-06,
+      "loss": 1.1029,
+      "step": 3619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9683846396931345e-06,
+      "loss": 1.3772,
+      "step": 3620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.965849313223324e-06,
+      "loss": 0.8087,
+      "step": 3621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9633142645019965e-06,
+      "loss": 1.0717,
+      "step": 3622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960779494210008e-06,
+      "loss": 1.1277,
+      "step": 3623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.958245003028136e-06,
+      "loss": 1.1218,
+      "step": 3624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955710791637085e-06,
+      "loss": 1.153,
+      "step": 3625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953176860717488e-06,
+      "loss": 1.066,
+      "step": 3626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950643210949894e-06,
+      "loss": 1.1926,
+      "step": 3627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948109843014784e-06,
+      "loss": 0.9058,
+      "step": 3628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.945576757592561e-06,
+      "loss": 1.3323,
+      "step": 3629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9430439553635504e-06,
+      "loss": 0.9325,
+      "step": 3630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9405114370080075e-06,
+      "loss": 0.9553,
+      "step": 3631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.937979203206103e-06,
+      "loss": 0.9215,
+      "step": 3632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9354472546379345e-06,
+      "loss": 0.9547,
+      "step": 3633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.932915591983526e-06,
+      "loss": 0.9221,
+      "step": 3634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.930384215922822e-06,
+      "loss": 0.9865,
+      "step": 3635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.927853127135692e-06,
+      "loss": 0.8816,
+      "step": 3636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.925322326301923e-06,
+      "loss": 1.1486,
+      "step": 3637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.92279181410123e-06,
+      "loss": 0.9503,
+      "step": 3638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.920261591213249e-06,
+      "loss": 1.0177,
+      "step": 3639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.917731658317538e-06,
+      "loss": 1.4402,
+      "step": 3640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91520201609358e-06,
+      "loss": 1.2434,
+      "step": 3641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.912672665220773e-06,
+      "loss": 1.0527,
+      "step": 3642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.910143606378443e-06,
+      "loss": 0.8135,
+      "step": 3643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.907614840245836e-06,
+      "loss": 1.0291,
+      "step": 3644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905086367502118e-06,
+      "loss": 1.1848,
+      "step": 3645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.90255818882638e-06,
+      "loss": 0.951,
+      "step": 3646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.900030304897626e-06,
+      "loss": 0.9205,
+      "step": 3647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897502716394789e-06,
+      "loss": 1.0397,
+      "step": 3648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89497542399672e-06,
+      "loss": 1.4627,
+      "step": 3649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892448428382189e-06,
+      "loss": 1.1199,
+      "step": 3650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889921730229889e-06,
+      "loss": 0.9411,
+      "step": 3651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887395330218429e-06,
+      "loss": 1.5667,
+      "step": 3652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884869229026343e-06,
+      "loss": 1.0166,
+      "step": 3653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8823434273320794e-06,
+      "loss": 1.3662,
+      "step": 3654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.879817925814012e-06,
+      "loss": 1.1078,
+      "step": 3655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.877292725150429e-06,
+      "loss": 1.0418,
+      "step": 3656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8747678260195424e-06,
+      "loss": 1.103,
+      "step": 3657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.872243229099476e-06,
+      "loss": 1.1933,
+      "step": 3658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86971893506828e-06,
+      "loss": 1.0312,
+      "step": 3659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86719494460392e-06,
+      "loss": 1.099,
+      "step": 3660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8646712583842795e-06,
+      "loss": 1.1832,
+      "step": 3661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8621478770871645e-06,
+      "loss": 1.0741,
+      "step": 3662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859624801390291e-06,
+      "loss": 1.0747,
+      "step": 3663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.857102031971298e-06,
+      "loss": 0.9971,
+      "step": 3664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.854579569507744e-06,
+      "loss": 1.0698,
+      "step": 3665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852057414677102e-06,
+      "loss": 1.1627,
+      "step": 3666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849535568156766e-06,
+      "loss": 1.009,
+      "step": 3667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.84701403062404e-06,
+      "loss": 0.8708,
+      "step": 3668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8444928027561515e-06,
+      "loss": 1.0453,
+      "step": 3669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841971885230243e-06,
+      "loss": 1.2248,
+      "step": 3670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.839451278723373e-06,
+      "loss": 0.9757,
+      "step": 3671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.83693098391252e-06,
+      "loss": 1.476,
+      "step": 3672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.83441100147457e-06,
+      "loss": 1.0389,
+      "step": 3673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8318913320863355e-06,
+      "loss": 1.3277,
+      "step": 3674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.829371976424537e-06,
+      "loss": 0.9179,
+      "step": 3675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826852935165818e-06,
+      "loss": 1.0787,
+      "step": 3676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.824334208986731e-06,
+      "loss": 1.2443,
+      "step": 3677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8218157985637465e-06,
+      "loss": 1.1848,
+      "step": 3678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.81929770457325e-06,
+      "loss": 0.9503,
+      "step": 3679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.816779927691542e-06,
+      "loss": 1.1618,
+      "step": 3680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.814262468594839e-06,
+      "loss": 1.1218,
+      "step": 3681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.811745327959271e-06,
+      "loss": 1.2704,
+      "step": 3682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8092285064608826e-06,
+      "loss": 0.9724,
+      "step": 3683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8067120047756313e-06,
+      "loss": 1.1755,
+      "step": 3684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8041958235793906e-06,
+      "loss": 1.1685,
+      "step": 3685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.801679963547949e-06,
+      "loss": 1.0555,
+      "step": 3686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7991644253570053e-06,
+      "loss": 1.0293,
+      "step": 3687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7966492096821773e-06,
+      "loss": 1.2505,
+      "step": 3688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7941343171989876e-06,
+      "loss": 0.9021,
+      "step": 3689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7916197485828793e-06,
+      "loss": 0.7455,
+      "step": 3690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.789105504509206e-06,
+      "loss": 0.9235,
+      "step": 3691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.786591585653235e-06,
+      "loss": 1.1398,
+      "step": 3692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7840779926901465e-06,
+      "loss": 1.2633,
+      "step": 3693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7815647262950293e-06,
+      "loss": 0.8928,
+      "step": 3694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7790517871428895e-06,
+      "loss": 1.1645,
+      "step": 3695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7765391759086424e-06,
+      "loss": 1.178,
+      "step": 3696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.774026893267117e-06,
+      "loss": 1.3383,
+      "step": 3697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.771514939893055e-06,
+      "loss": 1.2571,
+      "step": 3698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7690033164611036e-06,
+      "loss": 1.3845,
+      "step": 3699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766492023645827e-06,
+      "loss": 1.1064,
+      "step": 3700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7639810621217004e-06,
+      "loss": 1.1108,
+      "step": 3701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.761470432563109e-06,
+      "loss": 1.1777,
+      "step": 3702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7589601356443484e-06,
+      "loss": 0.9357,
+      "step": 3703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7564501720396242e-06,
+      "loss": 0.9946,
+      "step": 3704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.753940542423053e-06,
+      "loss": 1.2709,
+      "step": 3705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7514312474686643e-06,
+      "loss": 1.1689,
+      "step": 3706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.748922287850394e-06,
+      "loss": 1.1104,
+      "step": 3707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74641366424209e-06,
+      "loss": 1.0034,
+      "step": 3708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7439053773175092e-06,
+      "loss": 1.1768,
+      "step": 3709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7413974277503183e-06,
+      "loss": 0.954,
+      "step": 3710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.738889816214093e-06,
+      "loss": 0.9458,
+      "step": 3711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7363825433823187e-06,
+      "loss": 1.116,
+      "step": 3712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7338756099283925e-06,
+      "loss": 0.8238,
+      "step": 3713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7313690165256134e-06,
+      "loss": 0.9508,
+      "step": 3714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7288627638471953e-06,
+      "loss": 1.0531,
+      "step": 3715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7263568525662574e-06,
+      "loss": 1.2173,
+      "step": 3716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.72385128335583e-06,
+      "loss": 1.1929,
+      "step": 3717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7213460568888493e-06,
+      "loss": 0.9797,
+      "step": 3718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.718841173838161e-06,
+      "loss": 1.273,
+      "step": 3719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.716336634876516e-06,
+      "loss": 1.2142,
+      "step": 3720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.713832440676574e-06,
+      "loss": 1.1192,
+      "step": 3721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.711328591910904e-06,
+      "loss": 1.2625,
+      "step": 3722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7088250892519794e-06,
+      "loss": 0.9117,
+      "step": 3723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7063219333721857e-06,
+      "loss": 1.163,
+      "step": 3724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.703819124943806e-06,
+      "loss": 1.2368,
+      "step": 3725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7013166646390384e-06,
+      "loss": 1.1528,
+      "step": 3726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6988145531299846e-06,
+      "loss": 1.089,
+      "step": 3727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6963127910886526e-06,
+      "loss": 1.1084,
+      "step": 3728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.693811379186959e-06,
+      "loss": 0.8275,
+      "step": 3729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691310318096719e-06,
+      "loss": 1.4433,
+      "step": 3730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.688809608489662e-06,
+      "loss": 1.0027,
+      "step": 3731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6863092510374198e-06,
+      "loss": 1.0583,
+      "step": 3732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6838092464115276e-06,
+      "loss": 1.1341,
+      "step": 3733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68130959528343e-06,
+      "loss": 0.9081,
+      "step": 3734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6788102983244718e-06,
+      "loss": 1.3512,
+      "step": 3735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6763113562059077e-06,
+      "loss": 1.0125,
+      "step": 3736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6738127695988925e-06,
+      "loss": 1.0361,
+      "step": 3737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6713145391744877e-06,
+      "loss": 1.1401,
+      "step": 3738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6688166656036627e-06,
+      "loss": 1.3513,
+      "step": 3739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6663191495572827e-06,
+      "loss": 1.1501,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.663821991706123e-06,
+      "loss": 1.2903,
+      "step": 3741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661325192720862e-06,
+      "loss": 0.91,
+      "step": 3742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6588287532720814e-06,
+      "loss": 1.2449,
+      "step": 3743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6563326740302664e-06,
+      "loss": 1.1436,
+      "step": 3744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6538369556658026e-06,
+      "loss": 1.0816,
+      "step": 3745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6513415988489824e-06,
+      "loss": 1.1438,
+      "step": 3746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6488466042500003e-06,
+      "loss": 1.3798,
+      "step": 3747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6463519725389516e-06,
+      "loss": 1.4862,
+      "step": 3748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6438577043858373e-06,
+      "loss": 0.8742,
+      "step": 3749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.64136380046056e-06,
+      "loss": 1.0531,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6388702614329203e-06,
+      "loss": 1.095,
+      "step": 3751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6363770879726247e-06,
+      "loss": 0.9457,
+      "step": 3752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6338842807492813e-06,
+      "loss": 1.2902,
+      "step": 3753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6313918404324e-06,
+      "loss": 0.9784,
+      "step": 3754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.628899767691393e-06,
+      "loss": 0.9013,
+      "step": 3755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6264080631955683e-06,
+      "loss": 1.2915,
+      "step": 3756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6239167276141408e-06,
+      "loss": 1.1385,
+      "step": 3757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621425761616224e-06,
+      "loss": 1.0373,
+      "step": 3758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6189351658708347e-06,
+      "loss": 1.0676,
+      "step": 3759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.616444941046887e-06,
+      "loss": 0.9789,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6139550878131947e-06,
+      "loss": 1.0381,
+      "step": 3761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6114656068384767e-06,
+      "loss": 1.1466,
+      "step": 3762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6089764987913458e-06,
+      "loss": 1.306,
+      "step": 3763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6064877643403194e-06,
+      "loss": 0.932,
+      "step": 3764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.603999404153814e-06,
+      "loss": 1.2892,
+      "step": 3765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601511418900143e-06,
+      "loss": 1.0784,
+      "step": 3766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.599023809247519e-06,
+      "loss": 1.0534,
+      "step": 3767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5965365758640587e-06,
+      "loss": 0.9334,
+      "step": 3768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5940497194177713e-06,
+      "loss": 1.2891,
+      "step": 3769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.591563240576572e-06,
+      "loss": 0.8542,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5890771400082647e-06,
+      "loss": 1.0782,
+      "step": 3771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5865914183805606e-06,
+      "loss": 1.3088,
+      "step": 3772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5841060763610657e-06,
+      "loss": 1.0789,
+      "step": 3773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.581621114617284e-06,
+      "loss": 0.8727,
+      "step": 3774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5791365338166173e-06,
+      "loss": 1.0292,
+      "step": 3775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5766523346263682e-06,
+      "loss": 1.197,
+      "step": 3776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.57416851771373e-06,
+      "loss": 1.2475,
+      "step": 3777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.571685083745798e-06,
+      "loss": 1.1168,
+      "step": 3778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.569202033389565e-06,
+      "loss": 1.2665,
+      "step": 3779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.56671936731192e-06,
+      "loss": 0.9924,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5642370861796504e-06,
+      "loss": 1.126,
+      "step": 3781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.561755190659434e-06,
+      "loss": 1.0271,
+      "step": 3782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5592736814178508e-06,
+      "loss": 1.1344,
+      "step": 3783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.556792559121377e-06,
+      "loss": 0.8872,
+      "step": 3784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5543118244363843e-06,
+      "loss": 1.2383,
+      "step": 3785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5518314780291384e-06,
+      "loss": 1.1382,
+      "step": 3786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5493515205658004e-06,
+      "loss": 1.3549,
+      "step": 3787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5468719527124294e-06,
+      "loss": 1.0729,
+      "step": 3788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.54439277513498e-06,
+      "loss": 0.9682,
+      "step": 3789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.541913988499299e-06,
+      "loss": 1.1914,
+      "step": 3790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.539435593471132e-06,
+      "loss": 1.3182,
+      "step": 3791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5369575907161167e-06,
+      "loss": 1.0863,
+      "step": 3792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5344799808997845e-06,
+      "loss": 0.881,
+      "step": 3793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5320027646875643e-06,
+      "loss": 0.9651,
+      "step": 3794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.529525942744778e-06,
+      "loss": 1.1921,
+      "step": 3795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5270495157366434e-06,
+      "loss": 1.0246,
+      "step": 3796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.524573484328267e-06,
+      "loss": 1.3119,
+      "step": 3797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5220978491846534e-06,
+      "loss": 1.0982,
+      "step": 3798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.519622610970701e-06,
+      "loss": 0.9217,
+      "step": 3799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517147770351199e-06,
+      "loss": 0.9926,
+      "step": 3800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5146733279908345e-06,
+      "loss": 1.0631,
+      "step": 3801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5121992845541797e-06,
+      "loss": 1.1706,
+      "step": 3802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5097256407057066e-06,
+      "loss": 1.1034,
+      "step": 3803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.507252397109777e-06,
+      "loss": 0.9654,
+      "step": 3804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5047795544306473e-06,
+      "loss": 1.1643,
+      "step": 3805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5023071133324627e-06,
+      "loss": 1.4018,
+      "step": 3806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4998350744792664e-06,
+      "loss": 1.0576,
+      "step": 3807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497363438534984e-06,
+      "loss": 1.2316,
+      "step": 3808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4948922061634415e-06,
+      "loss": 0.9829,
+      "step": 3809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4924213780283545e-06,
+      "loss": 1.1431,
+      "step": 3810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4899509547933275e-06,
+      "loss": 1.0803,
+      "step": 3811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4874809371218608e-06,
+      "loss": 0.9799,
+      "step": 3812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4850113256773388e-06,
+      "loss": 0.9871,
+      "step": 3813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4825421211230437e-06,
+      "loss": 1.1964,
+      "step": 3814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4800733241221444e-06,
+      "loss": 1.0029,
+      "step": 3815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4776049353377016e-06,
+      "loss": 1.2056,
+      "step": 3816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.475136955432668e-06,
+      "loss": 1.0686,
+      "step": 3817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4726693850698824e-06,
+      "loss": 1.068,
+      "step": 3818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4702022249120777e-06,
+      "loss": 1.0704,
+      "step": 3819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.467735475621873e-06,
+      "loss": 1.5442,
+      "step": 3820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.465269137861781e-06,
+      "loss": 1.2997,
+      "step": 3821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4628032122942024e-06,
+      "loss": 1.2096,
+      "step": 3822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4603376995814235e-06,
+      "loss": 1.0632,
+      "step": 3823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4578726003856245e-06,
+      "loss": 0.8235,
+      "step": 3824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4554079153688735e-06,
+      "loss": 1.004,
+      "step": 3825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4529436451931263e-06,
+      "loss": 1.175,
+      "step": 3826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4504797905202303e-06,
+      "loss": 1.0871,
+      "step": 3827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.448016352011914e-06,
+      "loss": 1.1858,
+      "step": 3828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4455533303298007e-06,
+      "loss": 1.0425,
+      "step": 3829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4430907261354e-06,
+      "loss": 1.2494,
+      "step": 3830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4406285400901107e-06,
+      "loss": 1.0295,
+      "step": 3831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.438166772855218e-06,
+      "loss": 1.2794,
+      "step": 3832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4357054250918926e-06,
+      "loss": 0.7011,
+      "step": 3833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4332444974611946e-06,
+      "loss": 1.1695,
+      "step": 3834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4307839906240714e-06,
+      "loss": 1.053,
+      "step": 3835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.428323905241358e-06,
+      "loss": 1.1211,
+      "step": 3836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4258642419737748e-06,
+      "loss": 1.0228,
+      "step": 3837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4234050014819308e-06,
+      "loss": 0.885,
+      "step": 3838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.420946184426317e-06,
+      "loss": 1.4031,
+      "step": 3839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4184877914673155e-06,
+      "loss": 0.9956,
+      "step": 3840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4160298232651923e-06,
+      "loss": 0.9869,
+      "step": 3841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4135722804801004e-06,
+      "loss": 1.1684,
+      "step": 3842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4111151637720776e-06,
+      "loss": 1.1475,
+      "step": 3843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4086584738010455e-06,
+      "loss": 1.0635,
+      "step": 3844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4062022112268157e-06,
+      "loss": 0.9365,
+      "step": 3845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4037463767090807e-06,
+      "loss": 1.0338,
+      "step": 3846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4012909709074205e-06,
+      "loss": 1.015,
+      "step": 3847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3988359944812997e-06,
+      "loss": 1.2185,
+      "step": 3848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3963814480900665e-06,
+      "loss": 1.0147,
+      "step": 3849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3939273323929533e-06,
+      "loss": 0.8815,
+      "step": 3850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3914736480490785e-06,
+      "loss": 1.0199,
+      "step": 3851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3890203957174437e-06,
+      "loss": 1.4076,
+      "step": 3852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.386567576056937e-06,
+      "loss": 1.1723,
+      "step": 3853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3841151897263234e-06,
+      "loss": 0.9866,
+      "step": 3854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3816632373842583e-06,
+      "loss": 0.9932,
+      "step": 3855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.379211719689278e-06,
+      "loss": 1.0076,
+      "step": 3856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3767606372998033e-06,
+      "loss": 1.2272,
+      "step": 3857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3743099908741385e-06,
+      "loss": 1.3112,
+      "step": 3858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3718597810704646e-06,
+      "loss": 1.2238,
+      "step": 3859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3694100085468535e-06,
+      "loss": 0.9162,
+      "step": 3860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3669606739612554e-06,
+      "loss": 1.0541,
+      "step": 3861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.364511777971504e-06,
+      "loss": 0.8268,
+      "step": 3862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3620633212353182e-06,
+      "loss": 0.988,
+      "step": 3863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3596153044102897e-06,
+      "loss": 1.0131,
+      "step": 3864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3571677281539013e-06,
+      "loss": 1.172,
+      "step": 3865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.354720593123514e-06,
+      "loss": 0.9498,
+      "step": 3866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3522738999763706e-06,
+      "loss": 1.0938,
+      "step": 3867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.349827649369596e-06,
+      "loss": 0.9773,
+      "step": 3868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3473818419601944e-06,
+      "loss": 1.0058,
+      "step": 3869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3449364784050515e-06,
+      "loss": 1.1356,
+      "step": 3870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.342491559360934e-06,
+      "loss": 1.0162,
+      "step": 3871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3400470854844925e-06,
+      "loss": 0.9046,
+      "step": 3872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3376030574322515e-06,
+      "loss": 1.0277,
+      "step": 3873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3351594758606222e-06,
+      "loss": 0.9072,
+      "step": 3874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3327163414258913e-06,
+      "loss": 1.1547,
+      "step": 3875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3302736547842263e-06,
+      "loss": 1.2122,
+      "step": 3876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.327831416591676e-06,
+      "loss": 0.9048,
+      "step": 3877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3253896275041677e-06,
+      "loss": 0.9835,
+      "step": 3878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3229482881775117e-06,
+      "loss": 1.3615,
+      "step": 3879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3205073992673885e-06,
+      "loss": 1.0804,
+      "step": 3880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.318066961429367e-06,
+      "loss": 1.2456,
+      "step": 3881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3156269753188895e-06,
+      "loss": 1.3762,
+      "step": 3882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.31318744159128e-06,
+      "loss": 0.917,
+      "step": 3883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.310748360901741e-06,
+      "loss": 1.1002,
+      "step": 3884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.308309733905349e-06,
+      "loss": 0.9725,
+      "step": 3885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3058715612570623e-06,
+      "loss": 0.9475,
+      "step": 3886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3034338436117175e-06,
+      "loss": 0.8693,
+      "step": 3887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.300996581624028e-06,
+      "loss": 1.0921,
+      "step": 3888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2985597759485866e-06,
+      "loss": 0.9613,
+      "step": 3889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2961234272398578e-06,
+      "loss": 1.0472,
+      "step": 3890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2936875361521902e-06,
+      "loss": 0.9804,
+      "step": 3891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.291252103339806e-06,
+      "loss": 1.1418,
+      "step": 3892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2888171294568044e-06,
+      "loss": 1.0886,
+      "step": 3893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2863826151571654e-06,
+      "loss": 0.9429,
+      "step": 3894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2839485610947363e-06,
+      "loss": 1.1843,
+      "step": 3895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2815149679232507e-06,
+      "loss": 0.941,
+      "step": 3896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2790818362963127e-06,
+      "loss": 0.9916,
+      "step": 3897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.276649166867406e-06,
+      "loss": 0.7612,
+      "step": 3898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.274216960289886e-06,
+      "loss": 1.1152,
+      "step": 3899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271785217216987e-06,
+      "loss": 1.2457,
+      "step": 3900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.269353938301818e-06,
+      "loss": 0.9869,
+      "step": 3901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.266923124197363e-06,
+      "loss": 1.2022,
+      "step": 3902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2644927755564816e-06,
+      "loss": 1.1088,
+      "step": 3903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2620628930319065e-06,
+      "loss": 0.9614,
+      "step": 3904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2596334772762504e-06,
+      "loss": 1.4511,
+      "step": 3905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.257204528941993e-06,
+      "loss": 1.4979,
+      "step": 3906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.254776048681494e-06,
+      "loss": 1.063,
+      "step": 3907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2523480371469863e-06,
+      "loss": 0.949,
+      "step": 3908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2499204949905767e-06,
+      "loss": 0.8922,
+      "step": 3909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2474934228642475e-06,
+      "loss": 1.5122,
+      "step": 3910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2450668214198487e-06,
+      "loss": 1.3233,
+      "step": 3911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.242640691309111e-06,
+      "loss": 0.8582,
+      "step": 3912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2402150331836356e-06,
+      "loss": 1.2862,
+      "step": 3913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2377898476948964e-06,
+      "loss": 1.0249,
+      "step": 3914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.235365135494244e-06,
+      "loss": 0.9198,
+      "step": 3915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2329408972328934e-06,
+      "loss": 0.8462,
+      "step": 3916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.230517133561941e-06,
+      "loss": 1.0822,
+      "step": 3917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2280938451323524e-06,
+      "loss": 1.3487,
+      "step": 3918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.225671032594966e-06,
+      "loss": 1.0439,
+      "step": 3919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.223248696600493e-06,
+      "loss": 1.3498,
+      "step": 3920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2208268377995127e-06,
+      "loss": 0.9455,
+      "step": 3921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2184054568424817e-06,
+      "loss": 1.0608,
+      "step": 3922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2159845543797243e-06,
+      "loss": 0.9915,
+      "step": 3923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2135641310614383e-06,
+      "loss": 0.8314,
+      "step": 3924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2111441875376943e-06,
+      "loss": 1.0622,
+      "step": 3925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.20872472445843e-06,
+      "loss": 1.0473,
+      "step": 3926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2063057424734554e-06,
+      "loss": 1.0416,
+      "step": 3927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.203887242232455e-06,
+      "loss": 1.2256,
+      "step": 3928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2014692243849772e-06,
+      "loss": 1.1437,
+      "step": 3929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1990516895804467e-06,
+      "loss": 1.133,
+      "step": 3930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.196634638468157e-06,
+      "loss": 0.7445,
+      "step": 3931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1942180716972698e-06,
+      "loss": 0.8686,
+      "step": 3932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1918019899168174e-06,
+      "loss": 1.1597,
+      "step": 3933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.189386393775703e-06,
+      "loss": 1.0514,
+      "step": 3934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1869712839226987e-06,
+      "loss": 0.9889,
+      "step": 3935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1845566610064487e-06,
+      "loss": 1.2385,
+      "step": 3936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.182142525675459e-06,
+      "loss": 0.8807,
+      "step": 3937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179728878578112e-06,
+      "loss": 1.3264,
+      "step": 3938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177315720362656e-06,
+      "loss": 0.9778,
+      "step": 3939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1749030516772084e-06,
+      "loss": 1.4229,
+      "step": 3940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172490873169758e-06,
+      "loss": 1.0342,
+      "step": 3941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170079185488153e-06,
+      "loss": 1.0105,
+      "step": 3942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1676679892801197e-06,
+      "loss": 1.0994,
+      "step": 3943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.165257285193248e-06,
+      "loss": 1.4044,
+      "step": 3944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.162847073874996e-06,
+      "loss": 0.882,
+      "step": 3945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1604373559726915e-06,
+      "loss": 1.2029,
+      "step": 3946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.158028132133524e-06,
+      "loss": 1.1909,
+      "step": 3947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1556194030045563e-06,
+      "loss": 1.3913,
+      "step": 3948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1532111692327167e-06,
+      "loss": 0.8474,
+      "step": 3949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1508034314647994e-06,
+      "loss": 1.1184,
+      "step": 3950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148396190347468e-06,
+      "loss": 1.1629,
+      "step": 3951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1459894465272467e-06,
+      "loss": 0.9028,
+      "step": 3952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.143583200650532e-06,
+      "loss": 1.0749,
+      "step": 3953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1411774533635854e-06,
+      "loss": 1.1454,
+      "step": 3954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1387722053125338e-06,
+      "loss": 0.953,
+      "step": 3955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136367457143369e-06,
+      "loss": 0.9607,
+      "step": 3956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.133963209501952e-06,
+      "loss": 1.1474,
+      "step": 3957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1315594630340052e-06,
+      "loss": 1.2487,
+      "step": 3958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1291562183851165e-06,
+      "loss": 1.305,
+      "step": 3959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1267534762007435e-06,
+      "loss": 1.4417,
+      "step": 3960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.124351237126205e-06,
+      "loss": 1.0251,
+      "step": 3961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1219495018066888e-06,
+      "loss": 0.8578,
+      "step": 3962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.119548270887239e-06,
+      "loss": 1.1151,
+      "step": 3963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1171475450127717e-06,
+      "loss": 1.0341,
+      "step": 3964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114747324828066e-06,
+      "loss": 1.1329,
+      "step": 3965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.112347610977764e-06,
+      "loss": 0.8516,
+      "step": 3966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.109948404106375e-06,
+      "loss": 1.6569,
+      "step": 3967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1075497048582635e-06,
+      "loss": 0.8668,
+      "step": 3968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1051515138776677e-06,
+      "loss": 1.1869,
+      "step": 3969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.102753831808685e-06,
+      "loss": 1.0843,
+      "step": 3970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.100356659295274e-06,
+      "loss": 0.9327,
+      "step": 3971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.097959996981263e-06,
+      "loss": 1.2688,
+      "step": 3972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0955638455103336e-06,
+      "loss": 0.9375,
+      "step": 3973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.093168205526038e-06,
+      "loss": 1.0696,
+      "step": 3974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0907730776717892e-06,
+      "loss": 1.0612,
+      "step": 3975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0883784625908618e-06,
+      "loss": 1.0627,
+      "step": 3976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0859843609263946e-06,
+      "loss": 0.9353,
+      "step": 3977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.083590773321383e-06,
+      "loss": 1.227,
+      "step": 3978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.081197700418689e-06,
+      "loss": 0.9116,
+      "step": 3979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0788051428610377e-06,
+      "loss": 1.1006,
+      "step": 3980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0764131012910137e-06,
+      "loss": 1.2131,
+      "step": 3981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0740215763510617e-06,
+      "loss": 1.0653,
+      "step": 3982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.071630568683488e-06,
+      "loss": 1.0559,
+      "step": 3983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.069240078930461e-06,
+      "loss": 1.1045,
+      "step": 3984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.066850107734013e-06,
+      "loss": 0.7339,
+      "step": 3985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0644606557360303e-06,
+      "loss": 1.0195,
+      "step": 3986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0620717235782638e-06,
+      "loss": 1.0764,
+      "step": 3987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0596833119023283e-06,
+      "loss": 1.0807,
+      "step": 3988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0572954213496897e-06,
+      "loss": 0.9656,
+      "step": 3989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.054908052561681e-06,
+      "loss": 0.6501,
+      "step": 3990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0525212061794927e-06,
+      "loss": 1.0137,
+      "step": 3991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0501348828441767e-06,
+      "loss": 1.0028,
+      "step": 3992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.047749083196645e-06,
+      "loss": 1.1355,
+      "step": 3993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0453638078776614e-06,
+      "loss": 1.2163,
+      "step": 3994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0429790575278587e-06,
+      "loss": 1.0711,
+      "step": 3995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0405948327877233e-06,
+      "loss": 1.0192,
+      "step": 3996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.038211134297603e-06,
+      "loss": 1.1523,
+      "step": 3997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0358279626977034e-06,
+      "loss": 1.1744,
+      "step": 3998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.033445318628085e-06,
+      "loss": 0.9579,
+      "step": 3999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0310632027286717e-06,
+      "loss": 1.0107,
+      "step": 4000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.028681615639244e-06,
+      "loss": 1.0497,
+      "step": 4001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.026300557999439e-06,
+      "loss": 1.1648,
+      "step": 4002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.023920030448756e-06,
+      "loss": 1.0788,
+      "step": 4003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.021540033626544e-06,
+      "loss": 1.0102,
+      "step": 4004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0191605681720154e-06,
+      "loss": 1.067,
+      "step": 4005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0167816347242396e-06,
+      "loss": 1.181,
+      "step": 4006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0144032339221407e-06,
+      "loss": 0.9066,
+      "step": 4007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.012025366404504e-06,
+      "loss": 1.2606,
+      "step": 4008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0096480328099644e-06,
+      "loss": 0.9465,
+      "step": 4009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.00727123377702e-06,
+      "loss": 1.1475,
+      "step": 4010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0048949699440234e-06,
+      "loss": 1.1024,
+      "step": 4011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.002519241949181e-06,
+      "loss": 0.9957,
+      "step": 4012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0001440504305598e-06,
+      "loss": 1.141,
+      "step": 4013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.997769396026078e-06,
+      "loss": 0.9652,
+      "step": 4014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9953952793735132e-06,
+      "loss": 0.9442,
+      "step": 4015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9930217011104957e-06,
+      "loss": 1.0011,
+      "step": 4016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.990648661874514e-06,
+      "loss": 0.9746,
+      "step": 4017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.98827616230291e-06,
+      "loss": 1.0474,
+      "step": 4018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9859042030328833e-06,
+      "loss": 1.3674,
+      "step": 4019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9835327847014816e-06,
+      "loss": 1.1283,
+      "step": 4020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.981161907945615e-06,
+      "loss": 1.2607,
+      "step": 4021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9787915734020446e-06,
+      "loss": 1.0432,
+      "step": 4022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9764217817073853e-06,
+      "loss": 1.0273,
+      "step": 4023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9740525334981105e-06,
+      "loss": 1.1394,
+      "step": 4024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.971683829410541e-06,
+      "loss": 0.9162,
+      "step": 4025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9693156700808556e-06,
+      "loss": 0.8667,
+      "step": 4026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.966948056145086e-06,
+      "loss": 0.7884,
+      "step": 4027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9645809882391187e-06,
+      "loss": 1.0532,
+      "step": 4028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9622144669986925e-06,
+      "loss": 0.9941,
+      "step": 4029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.959848493059396e-06,
+      "loss": 0.9378,
+      "step": 4030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.957483067056677e-06,
+      "loss": 0.9437,
+      "step": 4031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9551181896258317e-06,
+      "loss": 1.0757,
+      "step": 4032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9527538614020117e-06,
+      "loss": 1.2698,
+      "step": 4033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9503900830202202e-06,
+      "loss": 0.9847,
+      "step": 4034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.94802685511531e-06,
+      "loss": 1.1535,
+      "step": 4035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9456641783219897e-06,
+      "loss": 1.1342,
+      "step": 4036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.943302053274818e-06,
+      "loss": 1.0992,
+      "step": 4037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9409404806082077e-06,
+      "loss": 0.9526,
+      "step": 4038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.93857946095642e-06,
+      "loss": 0.9534,
+      "step": 4039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.936218994953568e-06,
+      "loss": 0.972,
+      "step": 4040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.93385908323362e-06,
+      "loss": 1.2975,
+      "step": 4041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.93149972643039e-06,
+      "loss": 0.8343,
+      "step": 4042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.929140925177547e-06,
+      "loss": 0.9271,
+      "step": 4043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9267826801086103e-06,
+      "loss": 0.914,
+      "step": 4044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.924424991856947e-06,
+      "loss": 1.113,
+      "step": 4045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9220678610557773e-06,
+      "loss": 1.2238,
+      "step": 4046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.91971128833817e-06,
+      "loss": 1.1019,
+      "step": 4047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9173552743370454e-06,
+      "loss": 1.1667,
+      "step": 4048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.914999819685176e-06,
+      "loss": 0.8872,
+      "step": 4049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912644925015179e-06,
+      "loss": 1.1429,
+      "step": 4050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.910290590959523e-06,
+      "loss": 1.3381,
+      "step": 4051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9079368181505263e-06,
+      "loss": 1.1734,
+      "step": 4052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.905583607220359e-06,
+      "loss": 0.9263,
+      "step": 4053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9032309588010372e-06,
+      "loss": 1.0515,
+      "step": 4054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9008788735244264e-06,
+      "loss": 1.0493,
+      "step": 4055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8985273520222414e-06,
+      "loss": 0.9325,
+      "step": 4056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.896176394926043e-06,
+      "loss": 1.0472,
+      "step": 4057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.893826002867247e-06,
+      "loss": 0.9136,
+      "step": 4058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8914761764771093e-06,
+      "loss": 1.1085,
+      "step": 4059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.889126916386744e-06,
+      "loss": 0.9564,
+      "step": 4060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.886778223227099e-06,
+      "loss": 1.2366,
+      "step": 4061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.884430097628984e-06,
+      "loss": 1.4054,
+      "step": 4062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8820825402230455e-06,
+      "loss": 1.0769,
+      "step": 4063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.879735551639787e-06,
+      "loss": 0.8965,
+      "step": 4064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.877389132509551e-06,
+      "loss": 1.1885,
+      "step": 4065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8750432834625312e-06,
+      "loss": 1.2225,
+      "step": 4066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8726980051287654e-06,
+      "loss": 1.0917,
+      "step": 4067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8703532981381437e-06,
+      "loss": 0.9406,
+      "step": 4068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8680091631203954e-06,
+      "loss": 0.9852,
+      "step": 4069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8656656007051055e-06,
+      "loss": 1.0063,
+      "step": 4070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8633226115216916e-06,
+      "loss": 1.1868,
+      "step": 4071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.860980196199432e-06,
+      "loss": 1.0326,
+      "step": 4072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8586383553674394e-06,
+      "loss": 1.1687,
+      "step": 4073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8562970896546815e-06,
+      "loss": 0.9485,
+      "step": 4074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8539563996899634e-06,
+      "loss": 1.1274,
+      "step": 4075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8516162861019437e-06,
+      "loss": 1.142,
+      "step": 4076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8492767495191154e-06,
+      "loss": 1.1187,
+      "step": 4077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.846937790569828e-06,
+      "loss": 1.1378,
+      "step": 4078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.844599409882269e-06,
+      "loss": 0.9482,
+      "step": 4079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84226160808447e-06,
+      "loss": 1.1316,
+      "step": 4080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8399243858043136e-06,
+      "loss": 1.0479,
+      "step": 4081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.837587743669521e-06,
+      "loss": 1.1451,
+      "step": 4082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.835251682307658e-06,
+      "loss": 1.2627,
+      "step": 4083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8329162023461355e-06,
+      "loss": 1.0306,
+      "step": 4084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.83058130441221e-06,
+      "loss": 1.3517,
+      "step": 4085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.82824698913298e-06,
+      "loss": 0.6795,
+      "step": 4086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8259132571353864e-06,
+      "loss": 0.9609,
+      "step": 4087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.823580109046212e-06,
+      "loss": 1.2432,
+      "step": 4088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8212475454920916e-06,
+      "loss": 0.8994,
+      "step": 4089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8189155670994913e-06,
+      "loss": 1.0342,
+      "step": 4090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.816584174494731e-06,
+      "loss": 1.0476,
+      "step": 4091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.814253368303961e-06,
+      "loss": 1.1012,
+      "step": 4092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8119231491531856e-06,
+      "loss": 1.1258,
+      "step": 4093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.809593517668243e-06,
+      "loss": 1.1586,
+      "step": 4094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8072644744748224e-06,
+      "loss": 1.1462,
+      "step": 4095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.804936020198447e-06,
+      "loss": 1.3001,
+      "step": 4096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8026081554644846e-06,
+      "loss": 1.0478,
+      "step": 4097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.800280880898143e-06,
+      "loss": 1.0108,
+      "step": 4098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7979541971244773e-06,
+      "loss": 1.2795,
+      "step": 4099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.795628104768376e-06,
+      "loss": 0.7603,
+      "step": 4100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7933026044545773e-06,
+      "loss": 1.2205,
+      "step": 4101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.79097769680765e-06,
+      "loss": 1.4117,
+      "step": 4102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7886533824520146e-06,
+      "loss": 1.0226,
+      "step": 4103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7863296620119217e-06,
+      "loss": 1.124,
+      "step": 4104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7840065361114737e-06,
+      "loss": 1.0207,
+      "step": 4105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.781684005374604e-06,
+      "loss": 0.9985,
+      "step": 4106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7793620704250914e-06,
+      "loss": 1.0217,
+      "step": 4107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.777040731886549e-06,
+      "loss": 1.2012,
+      "step": 4108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.774719990382438e-06,
+      "loss": 1.2481,
+      "step": 4109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7723998465360537e-06,
+      "loss": 1.0743,
+      "step": 4110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.770080300970529e-06,
+      "loss": 1.185,
+      "step": 4111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7677613543088432e-06,
+      "loss": 1.1528,
+      "step": 4112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7654430071738104e-06,
+      "loss": 1.085,
+      "step": 4113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7631252601880816e-06,
+      "loss": 1.1623,
+      "step": 4114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7608081139741484e-06,
+      "loss": 0.8584,
+      "step": 4115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7584915691543444e-06,
+      "loss": 1.1045,
+      "step": 4116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7561756263508388e-06,
+      "loss": 0.8084,
+      "step": 4117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.753860286185637e-06,
+      "loss": 1.0977,
+      "step": 4118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.751545549280585e-06,
+      "loss": 1.1535,
+      "step": 4119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7492314162573687e-06,
+      "loss": 0.8186,
+      "step": 4120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.746917887737507e-06,
+      "loss": 0.9351,
+      "step": 4121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.744604964342364e-06,
+      "loss": 1.1492,
+      "step": 4122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.742292646693129e-06,
+      "loss": 0.6848,
+      "step": 4123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7399809354108415e-06,
+      "loss": 1.005,
+      "step": 4124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.737669831116369e-06,
+      "loss": 1.1933,
+      "step": 4125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.735359334430424e-06,
+      "loss": 0.8798,
+      "step": 4126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7330494459735484e-06,
+      "loss": 0.9844,
+      "step": 4127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7307401663661247e-06,
+      "loss": 1.0363,
+      "step": 4128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.728431496228369e-06,
+      "loss": 1.2618,
+      "step": 4129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7261234361803383e-06,
+      "loss": 1.2478,
+      "step": 4130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7238159868419205e-06,
+      "loss": 1.1741,
+      "step": 4131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.721509148832847e-06,
+      "loss": 0.8588,
+      "step": 4132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7192029227726736e-06,
+      "loss": 0.983,
+      "step": 4133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7168973092808025e-06,
+      "loss": 1.1416,
+      "step": 4134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.714592308976465e-06,
+      "loss": 1.1102,
+      "step": 4135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7122879224787315e-06,
+      "loss": 0.8027,
+      "step": 4136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7099841504065057e-06,
+      "loss": 1.0504,
+      "step": 4137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7076809933785254e-06,
+      "loss": 1.2428,
+      "step": 4138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.705378452013365e-06,
+      "loss": 1.1953,
+      "step": 4139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.70307652692943e-06,
+      "loss": 1.0821,
+      "step": 4140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7007752187449686e-06,
+      "loss": 1.0759,
+      "step": 4141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6984745280780524e-06,
+      "loss": 1.0984,
+      "step": 4142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.696174455546599e-06,
+      "loss": 1.1733,
+      "step": 4143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6938750017683457e-06,
+      "loss": 1.1579,
+      "step": 4144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6915761673608786e-06,
+      "loss": 1.165,
+      "step": 4145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6892779529416045e-06,
+      "loss": 1.1536,
+      "step": 4146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6869803591277756e-06,
+      "loss": 0.8665,
+      "step": 4147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6846833865364674e-06,
+      "loss": 0.9843,
+      "step": 4148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.682387035784594e-06,
+      "loss": 1.2087,
+      "step": 4149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6800913074888984e-06,
+      "loss": 1.1169,
+      "step": 4150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6777962022659633e-06,
+      "loss": 1.2129,
+      "step": 4151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6755017207321964e-06,
+      "loss": 0.9608,
+      "step": 4152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.673207863503846e-06,
+      "loss": 0.8679,
+      "step": 4153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6709146311969813e-06,
+      "loss": 1.0455,
+      "step": 4154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6686220244275153e-06,
+      "loss": 1.293,
+      "step": 4155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.666330043811185e-06,
+      "loss": 0.9798,
+      "step": 4156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6640386899635663e-06,
+      "loss": 1.0961,
+      "step": 4157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.66174796350006e-06,
+      "loss": 0.9587,
+      "step": 4158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.659457865035903e-06,
+      "loss": 1.0498,
+      "step": 4159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.657168395186157e-06,
+      "loss": 0.9808,
+      "step": 4160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6548795545657256e-06,
+      "loss": 1.1894,
+      "step": 4161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6525913437893346e-06,
+      "loss": 1.2505,
+      "step": 4162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6503037634715446e-06,
+      "loss": 1.08,
+      "step": 4163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.648016814226742e-06,
+      "loss": 1.1116,
+      "step": 4164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6457304966691527e-06,
+      "loss": 0.9631,
+      "step": 4165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6434448114128252e-06,
+      "loss": 0.8762,
+      "step": 4166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6411597590716387e-06,
+      "loss": 1.2992,
+      "step": 4167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6388753402593083e-06,
+      "loss": 1.3781,
+      "step": 4168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.636591555589373e-06,
+      "loss": 0.9134,
+      "step": 4169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6343084056752032e-06,
+      "loss": 1.1148,
+      "step": 4170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.632025891129998e-06,
+      "loss": 1.2018,
+      "step": 4171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6297440125667904e-06,
+      "loss": 0.7537,
+      "step": 4172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6274627705984352e-06,
+      "loss": 1.1579,
+      "step": 4173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6251821658376265e-06,
+      "loss": 0.909,
+      "step": 4174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6229021988968724e-06,
+      "loss": 1.2729,
+      "step": 4175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.620622870388524e-06,
+      "loss": 0.9041,
+      "step": 4176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.618344180924752e-06,
+      "loss": 0.9795,
+      "step": 4177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.616066131117563e-06,
+      "loss": 0.8716,
+      "step": 4178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.613788721578784e-06,
+      "loss": 1.2673,
+      "step": 4179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6115119529200748e-06,
+      "loss": 0.9396,
+      "step": 4180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6092358257529193e-06,
+      "loss": 1.3092,
+      "step": 4181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6069603406886347e-06,
+      "loss": 0.9614,
+      "step": 4182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.60468549833836e-06,
+      "loss": 0.9052,
+      "step": 4183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.60241129931307e-06,
+      "loss": 1.1397,
+      "step": 4184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.600137744223552e-06,
+      "loss": 1.1,
+      "step": 4185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.597864833680436e-06,
+      "loss": 0.991,
+      "step": 4186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5955925682941673e-06,
+      "loss": 1.1094,
+      "step": 4187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.593320948675029e-06,
+      "loss": 0.9303,
+      "step": 4188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5910499754331197e-06,
+      "loss": 1.1235,
+      "step": 4189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.588779649178371e-06,
+      "loss": 0.8491,
+      "step": 4190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5865099705205366e-06,
+      "loss": 1.1021,
+      "step": 4191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5842409400692026e-06,
+      "loss": 1.0294,
+      "step": 4192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.581972558433775e-06,
+      "loss": 0.8951,
+      "step": 4193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.579704826223488e-06,
+      "loss": 1.0532,
+      "step": 4194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5774377440473985e-06,
+      "loss": 1.0141,
+      "step": 4195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.575171312514395e-06,
+      "loss": 1.2208,
+      "step": 4196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.572905532233187e-06,
+      "loss": 0.7685,
+      "step": 4197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.570640403812306e-06,
+      "loss": 0.9418,
+      "step": 4198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5683759278601174e-06,
+      "loss": 0.8252,
+      "step": 4199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5661121049848026e-06,
+      "loss": 1.2681,
+      "step": 4200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5638489357943717e-06,
+      "loss": 0.8066,
+      "step": 4201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5615864208966573e-06,
+      "loss": 0.9779,
+      "step": 4202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5593245608993207e-06,
+      "loss": 1.2973,
+      "step": 4203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.55706335640984e-06,
+      "loss": 1.3907,
+      "step": 4204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.554802808035528e-06,
+      "loss": 1.0987,
+      "step": 4205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.552542916383507e-06,
+      "loss": 1.1513,
+      "step": 4206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.550283682060736e-06,
+      "loss": 1.2861,
+      "step": 4207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5480251056739874e-06,
+      "loss": 1.0808,
+      "step": 4208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.545767187829866e-06,
+      "loss": 1.0565,
+      "step": 4209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.543509929134794e-06,
+      "loss": 1.2176,
+      "step": 4210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5412533301950175e-06,
+      "loss": 0.9065,
+      "step": 4211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5389973916166037e-06,
+      "loss": 1.1499,
+      "step": 4212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5367421140054484e-06,
+      "loss": 1.0607,
+      "step": 4213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.534487497967262e-06,
+      "loss": 1.3821,
+      "step": 4214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.532233544107586e-06,
+      "loss": 1.0674,
+      "step": 4215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.529980253031774e-06,
+      "loss": 1.2602,
+      "step": 4216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5277276253450106e-06,
+      "loss": 1.0397,
+      "step": 4217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5254756616522953e-06,
+      "loss": 1.3368,
+      "step": 4218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5232243625584555e-06,
+      "loss": 1.069,
+      "step": 4219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5209737286681367e-06,
+      "loss": 0.907,
+      "step": 4220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.518723760585805e-06,
+      "loss": 1.0141,
+      "step": 4221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5164744589157488e-06,
+      "loss": 1.0975,
+      "step": 4222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5142258242620755e-06,
+      "loss": 1.1657,
+      "step": 4223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5119778572287195e-06,
+      "loss": 1.0677,
+      "step": 4224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5097305584194274e-06,
+      "loss": 1.0173,
+      "step": 4225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5074839284377774e-06,
+      "loss": 1.2166,
+      "step": 4226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.505237967887153e-06,
+      "loss": 0.9512,
+      "step": 4227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5029926773707713e-06,
+      "loss": 1.0303,
+      "step": 4228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500748057491662e-06,
+      "loss": 1.2513,
+      "step": 4229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.49850410885268e-06,
+      "loss": 1.1775,
+      "step": 4230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4962608320564953e-06,
+      "loss": 0.8771,
+      "step": 4231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4940182277055987e-06,
+      "loss": 1.029,
+      "step": 4232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.491776296402299e-06,
+      "loss": 1.0397,
+      "step": 4233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4895350387487304e-06,
+      "loss": 1.0655,
+      "step": 4234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.487294455346838e-06,
+      "loss": 1.1239,
+      "step": 4235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.485054546798395e-06,
+      "loss": 0.7906,
+      "step": 4236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.482815313704981e-06,
+      "loss": 0.9494,
+      "step": 4237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4805767566680057e-06,
+      "loss": 1.2699,
+      "step": 4238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.47833887628869e-06,
+      "loss": 0.9289,
+      "step": 4239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4761016731680792e-06,
+      "loss": 0.996,
+      "step": 4240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4738651479070313e-06,
+      "loss": 1.1181,
+      "step": 4241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4716293011062248e-06,
+      "loss": 1.0218,
+      "step": 4242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.469394133366153e-06,
+      "loss": 1.1021,
+      "step": 4243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.467159645287133e-06,
+      "loss": 1.0141,
+      "step": 4244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4649258374692943e-06,
+      "loss": 1.4865,
+      "step": 4245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4626927105125834e-06,
+      "loss": 0.9655,
+      "step": 4246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.460460265016766e-06,
+      "loss": 1.0791,
+      "step": 4247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4582285015814263e-06,
+      "loss": 1.0138,
+      "step": 4248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.455997420805962e-06,
+      "loss": 0.9735,
+      "step": 4249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4537670232895866e-06,
+      "loss": 1.0473,
+      "step": 4250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4515373096313367e-06,
+      "loss": 0.9986,
+      "step": 4251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4493082804300585e-06,
+      "loss": 1.075,
+      "step": 4252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4470799362844166e-06,
+      "loss": 1.0649,
+      "step": 4253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4448522777928903e-06,
+      "loss": 1.0876,
+      "step": 4254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.442625305553779e-06,
+      "loss": 1.0954,
+      "step": 4255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4403990201651915e-06,
+      "loss": 1.0293,
+      "step": 4256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.438173422225062e-06,
+      "loss": 0.952,
+      "step": 4257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.435948512331125e-06,
+      "loss": 1.1268,
+      "step": 4258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.433724291080945e-06,
+      "loss": 1.1951,
+      "step": 4259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4315007590718913e-06,
+      "loss": 1.1761,
+      "step": 4260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4292779169011564e-06,
+      "loss": 1.1008,
+      "step": 4261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.427055765165741e-06,
+      "loss": 1.1614,
+      "step": 4262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4248343044624626e-06,
+      "loss": 0.9379,
+      "step": 4263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4226135353879516e-06,
+      "loss": 0.7549,
+      "step": 4264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.420393458538658e-06,
+      "loss": 1.1159,
+      "step": 4265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4181740745108377e-06,
+      "loss": 1.1066,
+      "step": 4266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4159553839005712e-06,
+      "loss": 0.8741,
+      "step": 4267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.413737387303739e-06,
+      "loss": 0.983,
+      "step": 4268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4115200853160475e-06,
+      "loss": 0.8676,
+      "step": 4269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4093034785330087e-06,
+      "loss": 1.1182,
+      "step": 4270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4070875675499545e-06,
+      "loss": 1.1653,
+      "step": 4271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4048723529620246e-06,
+      "loss": 0.9577,
+      "step": 4272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.402657835364172e-06,
+      "loss": 0.9343,
+      "step": 4273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4004440153511642e-06,
+      "loss": 1.0617,
+      "step": 4274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.398230893517584e-06,
+      "loss": 0.9478,
+      "step": 4275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.396018470457821e-06,
+      "loss": 0.8481,
+      "step": 4276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3938067467660807e-06,
+      "loss": 0.8013,
+      "step": 4277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3915957230363783e-06,
+      "loss": 1.018,
+      "step": 4278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3893853998625455e-06,
+      "loss": 1.0979,
+      "step": 4279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3871757778382216e-06,
+      "loss": 1.1843,
+      "step": 4280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.384966857556857e-06,
+      "loss": 0.8723,
+      "step": 4281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3827586396117207e-06,
+      "loss": 1.1475,
+      "step": 4282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3805511245958816e-06,
+      "loss": 1.1817,
+      "step": 4283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.378344313102231e-06,
+      "loss": 1.1427,
+      "step": 4284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3761382057234627e-06,
+      "loss": 1.0925,
+      "step": 4285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.373932803052089e-06,
+      "loss": 1.3154,
+      "step": 4286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3717281056804253e-06,
+      "loss": 1.1164,
+      "step": 4287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.369524114200607e-06,
+      "loss": 1.1333,
+      "step": 4288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3673208292045668e-06,
+      "loss": 0.9469,
+      "step": 4289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3651182512840604e-06,
+      "loss": 1.0067,
+      "step": 4290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.362916381030645e-06,
+      "loss": 1.2051,
+      "step": 4291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.360715219035694e-06,
+      "loss": 1.0703,
+      "step": 4292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3585147658903863e-06,
+      "loss": 1.0867,
+      "step": 4293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.356315022185712e-06,
+      "loss": 1.2023,
+      "step": 4294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3541159885124672e-06,
+      "loss": 1.0064,
+      "step": 4295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3519176654612657e-06,
+      "loss": 1.0386,
+      "step": 4296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3497200536225204e-06,
+      "loss": 1.0855,
+      "step": 4297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3475231535864653e-06,
+      "loss": 0.8944,
+      "step": 4298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.345326965943126e-06,
+      "loss": 0.9636,
+      "step": 4299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3431314912823543e-06,
+      "loss": 1.1966,
+      "step": 4300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.340936730193798e-06,
+      "loss": 1.2596,
+      "step": 4301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.338742683266923e-06,
+      "loss": 1.054,
+      "step": 4302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3365493510909954e-06,
+      "loss": 0.8255,
+      "step": 4303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3343567342550933e-06,
+      "loss": 1.1187,
+      "step": 4304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.332164833348102e-06,
+      "loss": 1.0787,
+      "step": 4305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329973648958712e-06,
+      "loss": 0.7798,
+      "step": 4306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3277831816754277e-06,
+      "loss": 0.7784,
+      "step": 4307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3255934320865555e-06,
+      "loss": 1.247,
+      "step": 4308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3234044007802096e-06,
+      "loss": 1.0383,
+      "step": 4309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3212160883443107e-06,
+      "loss": 1.0287,
+      "step": 4310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.319028495366592e-06,
+      "loss": 1.0627,
+      "step": 4311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.316841622434586e-06,
+      "loss": 1.2121,
+      "step": 4312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.314655470135639e-06,
+      "loss": 0.8882,
+      "step": 4313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3124700390568945e-06,
+      "loss": 0.8829,
+      "step": 4314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3102853297853115e-06,
+      "loss": 1.1238,
+      "step": 4315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.30810134290765e-06,
+      "loss": 1.1144,
+      "step": 4316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.305918079010479e-06,
+      "loss": 1.029,
+      "step": 4317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3037355386801683e-06,
+      "loss": 1.009,
+      "step": 4318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.301553722502903e-06,
+      "loss": 1.2102,
+      "step": 4319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2993726310646603e-06,
+      "loss": 0.8617,
+      "step": 4320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.297192264951235e-06,
+      "loss": 1.4215,
+      "step": 4321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2950126247482178e-06,
+      "loss": 0.9036,
+      "step": 4322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2928337110410132e-06,
+      "loss": 0.9944,
+      "step": 4323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2906555244148233e-06,
+      "loss": 1.1976,
+      "step": 4324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2884780654546586e-06,
+      "loss": 1.1346,
+      "step": 4325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2863013347453305e-06,
+      "loss": 0.9581,
+      "step": 4326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.284125332871462e-06,
+      "loss": 1.1672,
+      "step": 4327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2819500604174733e-06,
+      "loss": 1.0167,
+      "step": 4328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.279775517967592e-06,
+      "loss": 0.958,
+      "step": 4329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.277601706105847e-06,
+      "loss": 0.9166,
+      "step": 4330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2754286254160755e-06,
+      "loss": 1.0794,
+      "step": 4331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2732562764819157e-06,
+      "loss": 1.0452,
+      "step": 4332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.271084659886806e-06,
+      "loss": 1.2342,
+      "step": 4333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2689137762139952e-06,
+      "loss": 1.0691,
+      "step": 4334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2667436260465304e-06,
+      "loss": 1.13,
+      "step": 4335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.264574209967262e-06,
+      "loss": 1.292,
+      "step": 4336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.262405528558842e-06,
+      "loss": 1.1676,
+      "step": 4337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260237582403732e-06,
+      "loss": 1.0489,
+      "step": 4338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.258070372084188e-06,
+      "loss": 1.217,
+      "step": 4339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2559038981822724e-06,
+      "loss": 1.0224,
+      "step": 4340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.253738161279847e-06,
+      "loss": 1.2278,
+      "step": 4341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2515731619585814e-06,
+      "loss": 0.7949,
+      "step": 4342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.249408900799939e-06,
+      "loss": 0.9852,
+      "step": 4343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.247245378385195e-06,
+      "loss": 1.2126,
+      "step": 4344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.245082595295414e-06,
+      "loss": 0.8398,
+      "step": 4345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242920552111473e-06,
+      "loss": 1.2047,
+      "step": 4346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2407592494140433e-06,
+      "loss": 0.8737,
+      "step": 4347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.238598687783603e-06,
+      "loss": 1.2985,
+      "step": 4348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2364388678004232e-06,
+      "loss": 0.9987,
+      "step": 4349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.234279790044588e-06,
+      "loss": 1.2171,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2321214550959656e-06,
+      "loss": 1.0183,
+      "step": 4351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.229963863534241e-06,
+      "loss": 1.1207,
+      "step": 4352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2278070159388875e-06,
+      "loss": 0.8034,
+      "step": 4353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.225650912889188e-06,
+      "loss": 1.025,
+      "step": 4354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2234955549642184e-06,
+      "loss": 0.9971,
+      "step": 4355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.221340942742858e-06,
+      "loss": 1.1016,
+      "step": 4356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2191870768037817e-06,
+      "loss": 1.0836,
+      "step": 4357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2170339577254714e-06,
+      "loss": 1.0093,
+      "step": 4358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2148815860862026e-06,
+      "loss": 1.2056,
+      "step": 4359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.212729962464051e-06,
+      "loss": 1.1335,
+      "step": 4360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2105790874368904e-06,
+      "loss": 1.1785,
+      "step": 4361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.208428961582399e-06,
+      "loss": 0.8062,
+      "step": 4362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.206279585478047e-06,
+      "loss": 1.1382,
+      "step": 4363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2041309597011057e-06,
+      "loss": 1.13,
+      "step": 4364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.20198308482865e-06,
+      "loss": 1.1051,
+      "step": 4365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1998359614375412e-06,
+      "loss": 1.2257,
+      "step": 4366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.197689590104452e-06,
+      "loss": 1.0407,
+      "step": 4367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1955439714058422e-06,
+      "loss": 1.1529,
+      "step": 4368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.19339910591798e-06,
+      "loss": 1.1132,
+      "step": 4369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.191254994216922e-06,
+      "loss": 0.8809,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1891116368785263e-06,
+      "loss": 1.1984,
+      "step": 4371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186969034478448e-06,
+      "loss": 0.9917,
+      "step": 4372
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1848271875921407e-06,
+      "loss": 1.0664,
+      "step": 4373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.182686096794852e-06,
+      "loss": 1.1712,
+      "step": 4374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.180545762661632e-06,
+      "loss": 1.3137,
+      "step": 4375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1784061857673217e-06,
+      "loss": 1.1485,
+      "step": 4376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1762673666865607e-06,
+      "loss": 1.054,
+      "step": 4377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.174129305993784e-06,
+      "loss": 0.8923,
+      "step": 4378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.171992004263228e-06,
+      "loss": 1.0023,
+      "step": 4379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1698554620689178e-06,
+      "loss": 1.2047,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1677196799846834e-06,
+      "loss": 0.8734,
+      "step": 4381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.165584658584138e-06,
+      "loss": 1.1697,
+      "step": 4382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1634503984407044e-06,
+      "loss": 1.0573,
+      "step": 4383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.16131690012759e-06,
+      "loss": 1.1063,
+      "step": 4384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159184164217807e-06,
+      "loss": 1.0895,
+      "step": 4385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.157052191284154e-06,
+      "loss": 1.0636,
+      "step": 4386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.154920981899231e-06,
+      "loss": 0.9784,
+      "step": 4387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1527905366354292e-06,
+      "loss": 1.1411,
+      "step": 4388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.150660856064934e-06,
+      "loss": 1.3033,
+      "step": 4389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1485319407597315e-06,
+      "loss": 0.9918,
+      "step": 4390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.146403791291597e-06,
+      "loss": 0.8066,
+      "step": 4391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1442764082321e-06,
+      "loss": 0.8147,
+      "step": 4392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.142149792152604e-06,
+      "loss": 0.9838,
+      "step": 4393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.140023943624272e-06,
+      "loss": 1.2222,
+      "step": 4394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1378988632180526e-06,
+      "loss": 1.2398,
+      "step": 4395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.135774551504698e-06,
+      "loss": 1.1661,
+      "step": 4396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.133651009054741e-06,
+      "loss": 1.106,
+      "step": 4397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1315282364385197e-06,
+      "loss": 0.9165,
+      "step": 4398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1294062342261585e-06,
+      "loss": 1.0666,
+      "step": 4399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1272850029875802e-06,
+      "loss": 0.8372,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.125164543292495e-06,
+      "loss": 1.0811,
+      "step": 4401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1230448557104087e-06,
+      "loss": 1.0982,
+      "step": 4402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1209259408106185e-06,
+      "loss": 1.1329,
+      "step": 4403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1188077991622174e-06,
+      "loss": 0.9926,
+      "step": 4404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.116690431334085e-06,
+      "loss": 0.982,
+      "step": 4405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1145738378949004e-06,
+      "loss": 1.1086,
+      "step": 4406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.112458019413129e-06,
+      "loss": 0.859,
+      "step": 4407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.110342976457029e-06,
+      "loss": 0.8179,
+      "step": 4408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108228709594649e-06,
+      "loss": 1.1214,
+      "step": 4409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1061152193938355e-06,
+      "loss": 1.3047,
+      "step": 4410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.104002506422221e-06,
+      "loss": 1.0889,
+      "step": 4411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1018905712472285e-06,
+      "loss": 1.0214,
+      "step": 4412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.099779414436074e-06,
+      "loss": 1.2601,
+      "step": 4413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0976690365557673e-06,
+      "loss": 1.0029,
+      "step": 4414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.095559438173104e-06,
+      "loss": 0.8854,
+      "step": 4415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.093450619854671e-06,
+      "loss": 0.9644,
+      "step": 4416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.091342582166851e-06,
+      "loss": 0.9786,
+      "step": 4417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0892353256758107e-06,
+      "loss": 1.0695,
+      "step": 4418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0871288509475103e-06,
+      "loss": 1.0095,
+      "step": 4419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0850231585476965e-06,
+      "loss": 1.1051,
+      "step": 4420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.082918249041913e-06,
+      "loss": 0.8645,
+      "step": 4421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0808141229954876e-06,
+      "loss": 1.0942,
+      "step": 4422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0787107809735376e-06,
+      "loss": 1.0961,
+      "step": 4423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0766082235409695e-06,
+      "loss": 1.0948,
+      "step": 4424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0745064512624845e-06,
+      "loss": 0.9056,
+      "step": 4425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.072405464702566e-06,
+      "loss": 0.9587,
+      "step": 4426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.070305264425494e-06,
+      "loss": 1.1215,
+      "step": 4427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.068205850995326e-06,
+      "loss": 1.0474,
+      "step": 4428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.066107224975921e-06,
+      "loss": 0.8769,
+      "step": 4429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.064009386930915e-06,
+      "loss": 1.4138,
+      "step": 4430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0619123374237425e-06,
+      "loss": 1.0735,
+      "step": 4431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0598160770176208e-06,
+      "loss": 1.0257,
+      "step": 4432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.057720606275554e-06,
+      "loss": 0.8694,
+      "step": 4433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0556259257603355e-06,
+      "loss": 1.281,
+      "step": 4434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0535320360345505e-06,
+      "loss": 1.174,
+      "step": 4435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0514389376605646e-06,
+      "loss": 0.9897,
+      "step": 4436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0493466312005387e-06,
+      "loss": 0.8977,
+      "step": 4437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0472551172164152e-06,
+      "loss": 0.9482,
+      "step": 4438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.045164396269924e-06,
+      "loss": 0.9,
+      "step": 4439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0430744689225833e-06,
+      "loss": 1.2408,
+      "step": 4440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0409853357357018e-06,
+      "loss": 1.0726,
+      "step": 4441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0388969972703688e-06,
+      "loss": 1.0615,
+      "step": 4442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.036809454087463e-06,
+      "loss": 1.182,
+      "step": 4443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0347227067476478e-06,
+      "loss": 0.9746,
+      "step": 4444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.032636755811377e-06,
+      "loss": 1.1312,
+      "step": 4445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.030551601838887e-06,
+      "loss": 0.8748,
+      "step": 4446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0284672453901995e-06,
+      "loss": 1.0079,
+      "step": 4447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0263836870251277e-06,
+      "loss": 0.8232,
+      "step": 4448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0243009273032593e-06,
+      "loss": 1.0519,
+      "step": 4449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0222189667839805e-06,
+      "loss": 1.1178,
+      "step": 4450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0201378060264525e-06,
+      "loss": 1.3776,
+      "step": 4451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01805744558963e-06,
+      "loss": 1.0164,
+      "step": 4452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0159778860322473e-06,
+      "loss": 0.7821,
+      "step": 4453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.013899127912824e-06,
+      "loss": 1.0159,
+      "step": 4454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0118211717896644e-06,
+      "loss": 1.1479,
+      "step": 4455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.009744018220863e-06,
+      "loss": 1.2326,
+      "step": 4456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.007667667764288e-06,
+      "loss": 1.2126,
+      "step": 4457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.005592120977606e-06,
+      "loss": 1.0568,
+      "step": 4458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0035173784182527e-06,
+      "loss": 0.8976,
+      "step": 4459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0014434406434584e-06,
+      "loss": 1.0554,
+      "step": 4460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9993703082102325e-06,
+      "loss": 0.9058,
+      "step": 4461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9972979816753717e-06,
+      "loss": 0.9715,
+      "step": 4462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9952264615954524e-06,
+      "loss": 1.0115,
+      "step": 4463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9931557485268365e-06,
+      "loss": 0.944,
+      "step": 4464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9910858430256667e-06,
+      "loss": 1.0575,
+      "step": 4465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890167456478748e-06,
+      "loss": 0.8882,
+      "step": 4466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9869484569491686e-06,
+      "loss": 0.9641,
+      "step": 4467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.984880977485041e-06,
+      "loss": 1.1121,
+      "step": 4468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.982814307810771e-06,
+      "loss": 1.0304,
+      "step": 4469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.980748448481416e-06,
+      "loss": 0.9821,
+      "step": 4470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9786834000518174e-06,
+      "loss": 1.32,
+      "step": 4471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9766191630765964e-06,
+      "loss": 0.8866,
+      "step": 4472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9745557381101616e-06,
+      "loss": 0.8358,
+      "step": 4473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9724931257066988e-06,
+      "loss": 1.1741,
+      "step": 4474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9704313264201768e-06,
+      "loss": 1.0236,
+      "step": 4475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9683703408043447e-06,
+      "loss": 1.1397,
+      "step": 4476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9663101694127385e-06,
+      "loss": 1.0066,
+      "step": 4477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9642508127986676e-06,
+      "loss": 0.9852,
+      "step": 4478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.962192271515232e-06,
+      "loss": 0.9037,
+      "step": 4479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9601345461153005e-06,
+      "loss": 1.0829,
+      "step": 4480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9580776371515346e-06,
+      "loss": 0.8297,
+      "step": 4481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9560215451763685e-06,
+      "loss": 0.9206,
+      "step": 4482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.953966270742023e-06,
+      "loss": 1.3596,
+      "step": 4483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.951911814400495e-06,
+      "loss": 0.9527,
+      "step": 4484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.949858176703563e-06,
+      "loss": 0.961,
+      "step": 4485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9478053582027826e-06,
+      "loss": 1.0602,
+      "step": 4486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9457533594494983e-06,
+      "loss": 1.0378,
+      "step": 4487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9437021809948232e-06,
+      "loss": 1.4318,
+      "step": 4488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9416518233896612e-06,
+      "loss": 1.0574,
+      "step": 4489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9396022871846836e-06,
+      "loss": 1.1183,
+      "step": 4490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.937553572930352e-06,
+      "loss": 1.1248,
+      "step": 4491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.935505681176899e-06,
+      "loss": 1.1991,
+      "step": 4492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933458612474345e-06,
+      "loss": 1.1701,
+      "step": 4493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9314123673724805e-06,
+      "loss": 0.9635,
+      "step": 4494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9293669464208808e-06,
+      "loss": 0.8345,
+      "step": 4495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9273223501688943e-06,
+      "loss": 0.8774,
+      "step": 4496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.925278579165656e-06,
+      "loss": 1.0599,
+      "step": 4497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9232356339600717e-06,
+      "loss": 1.1859,
+      "step": 4498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9211935151008276e-06,
+      "loss": 0.8148,
+      "step": 4499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.919152223136391e-06,
+      "loss": 0.7975,
+      "step": 4500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.917111758615005e-06,
+      "loss": 0.9609,
+      "step": 4501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9150721220846884e-06,
+      "loss": 1.2144,
+      "step": 4502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9130333140932385e-06,
+      "loss": 1.1886,
+      "step": 4503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.910995335188234e-06,
+      "loss": 1.2658,
+      "step": 4504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9089581859170268e-06,
+      "loss": 0.8098,
+      "step": 4505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.906921866826747e-06,
+      "loss": 1.1249,
+      "step": 4506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9048863784643002e-06,
+      "loss": 0.7549,
+      "step": 4507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9028517213763737e-06,
+      "loss": 1.169,
+      "step": 4508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9008178961094248e-06,
+      "loss": 1.1541,
+      "step": 4509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8987849032096973e-06,
+      "loss": 1.1896,
+      "step": 4510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.896752743223197e-06,
+      "loss": 1.2279,
+      "step": 4511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.89472141669572e-06,
+      "loss": 1.2884,
+      "step": 4512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8926909241728286e-06,
+      "loss": 1.0259,
+      "step": 4513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8906612661998698e-06,
+      "loss": 1.1011,
+      "step": 4514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8886324433219588e-06,
+      "loss": 1.1822,
+      "step": 4515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8866044560839902e-06,
+      "loss": 1.2239,
+      "step": 4516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8845773050306322e-06,
+      "loss": 0.9326,
+      "step": 4517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8825509907063328e-06,
+      "loss": 1.0744,
+      "step": 4518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8805255136553085e-06,
+      "loss": 1.1117,
+      "step": 4519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8785008744215606e-06,
+      "loss": 0.887,
+      "step": 4520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8764770735488518e-06,
+      "loss": 1.2603,
+      "step": 4521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.874454111580733e-06,
+      "loss": 0.8935,
+      "step": 4522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8724319890605202e-06,
+      "loss": 0.9365,
+      "step": 4523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8704107065313116e-06,
+      "loss": 1.0227,
+      "step": 4524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8683902645359737e-06,
+      "loss": 0.8602,
+      "step": 4525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8663706636171503e-06,
+      "loss": 1.1907,
+      "step": 4526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8643519043172565e-06,
+      "loss": 1.2523,
+      "step": 4527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8623339871784869e-06,
+      "loss": 1.0711,
+      "step": 4528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.860316912742805e-06,
+      "loss": 1.0541,
+      "step": 4529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8583006815519473e-06,
+      "loss": 1.1,
+      "step": 4530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8562852941474307e-06,
+      "loss": 1.049,
+      "step": 4531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8542707510705355e-06,
+      "loss": 1.074,
+      "step": 4532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8522570528623235e-06,
+      "loss": 1.1838,
+      "step": 4533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8502442000636246e-06,
+      "loss": 1.2229,
+      "step": 4534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8482321932150466e-06,
+      "loss": 1.16,
+      "step": 4535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.846221032856965e-06,
+      "loss": 1.1137,
+      "step": 4536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8442107195295312e-06,
+      "loss": 0.9726,
+      "step": 4537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8422012537726646e-06,
+      "loss": 1.3064,
+      "step": 4538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8401926361260647e-06,
+      "loss": 1.1625,
+      "step": 4539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8381848671291953e-06,
+      "loss": 1.1878,
+      "step": 4540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8361779473213004e-06,
+      "loss": 1.0934,
+      "step": 4541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8341718772413852e-06,
+      "loss": 1.064,
+      "step": 4542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8321666574282381e-06,
+      "loss": 0.9469,
+      "step": 4543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8301622884204096e-06,
+      "loss": 0.896,
+      "step": 4544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8281587707562298e-06,
+      "loss": 0.8084,
+      "step": 4545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8261561049737946e-06,
+      "loss": 1.0788,
+      "step": 4546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8241542916109728e-06,
+      "loss": 0.7335,
+      "step": 4547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8221533312054024e-06,
+      "loss": 1.1861,
+      "step": 4548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.820153224294498e-06,
+      "loss": 0.9724,
+      "step": 4549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.818153971415439e-06,
+      "loss": 1.1583,
+      "step": 4550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.816155573105177e-06,
+      "loss": 1.21,
+      "step": 4551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8141580299004342e-06,
+      "loss": 1.387,
+      "step": 4552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8121613423377065e-06,
+      "loss": 1.0229,
+      "step": 4553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8101655109532552e-06,
+      "loss": 1.0073,
+      "step": 4554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.808170536283112e-06,
+      "loss": 1.0202,
+      "step": 4555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8061764188630831e-06,
+      "loss": 1.3893,
+      "step": 4556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.80418315922874e-06,
+      "loss": 0.8858,
+      "step": 4557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8021907579154257e-06,
+      "loss": 0.9789,
+      "step": 4558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8001992154582499e-06,
+      "loss": 1.2043,
+      "step": 4559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7982085323920973e-06,
+      "loss": 1.2067,
+      "step": 4560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7962187092516154e-06,
+      "loss": 0.8933,
+      "step": 4561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7942297465712282e-06,
+      "loss": 1.0295,
+      "step": 4562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7922416448851181e-06,
+      "loss": 0.9844,
+      "step": 4563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7902544047272468e-06,
+      "loss": 1.145,
+      "step": 4564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7882680266313374e-06,
+      "loss": 1.0544,
+      "step": 4565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7862825111308873e-06,
+      "loss": 0.9129,
+      "step": 4566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7842978587591576e-06,
+      "loss": 0.9312,
+      "step": 4567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7823140700491786e-06,
+      "loss": 1.1207,
+      "step": 4568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.780331145533748e-06,
+      "loss": 1.173,
+      "step": 4569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7783490857454354e-06,
+      "loss": 1.1282,
+      "step": 4570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7763678912165727e-06,
+      "loss": 0.8624,
+      "step": 4571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7743875624792662e-06,
+      "loss": 0.921,
+      "step": 4572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724081000653792e-06,
+      "loss": 1.2043,
+      "step": 4573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770429504506554e-06,
+      "loss": 1.0602,
+      "step": 4574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684517763341908e-06,
+      "loss": 1.183,
+      "step": 4575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7664749160794642e-06,
+      "loss": 0.9645,
+      "step": 4576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7644989242733112e-06,
+      "loss": 1.3198,
+      "step": 4577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7625238014464358e-06,
+      "loss": 1.0544,
+      "step": 4578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7605495481293077e-06,
+      "loss": 1.0053,
+      "step": 4579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7585761648521688e-06,
+      "loss": 1.228,
+      "step": 4580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7566036521450209e-06,
+      "loss": 0.9882,
+      "step": 4581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7546320105376346e-06,
+      "loss": 0.8628,
+      "step": 4582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.752661240559545e-06,
+      "loss": 1.0184,
+      "step": 4583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.750691342740058e-06,
+      "loss": 1.0298,
+      "step": 4584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7487223176082392e-06,
+      "loss": 1.2419,
+      "step": 4585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.746754165692921e-06,
+      "loss": 0.935,
+      "step": 4586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7447868875227063e-06,
+      "loss": 0.9365,
+      "step": 4587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.742820483625957e-06,
+      "loss": 1.2474,
+      "step": 4588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.740854954530804e-06,
+      "loss": 1.2185,
+      "step": 4589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7388903007651398e-06,
+      "loss": 0.9723,
+      "step": 4590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7369265228566274e-06,
+      "loss": 0.9675,
+      "step": 4591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7349636213326876e-06,
+      "loss": 0.927,
+      "step": 4592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7330015967205149e-06,
+      "loss": 1.0133,
+      "step": 4593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7310404495470557e-06,
+      "loss": 1.0134,
+      "step": 4594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7290801803390334e-06,
+      "loss": 0.7427,
+      "step": 4595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.727120789622927e-06,
+      "loss": 0.8423,
+      "step": 4596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7251622779249848e-06,
+      "loss": 0.9113,
+      "step": 4597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7232046457712164e-06,
+      "loss": 1.1533,
+      "step": 4598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7212478936873949e-06,
+      "loss": 1.0063,
+      "step": 4599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7192920221990566e-06,
+      "loss": 0.907,
+      "step": 4600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7173370318315052e-06,
+      "loss": 1.1386,
+      "step": 4601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7153829231098018e-06,
+      "loss": 1.3089,
+      "step": 4602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7134296965587793e-06,
+      "loss": 1.106,
+      "step": 4603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7114773527030215e-06,
+      "loss": 0.9343,
+      "step": 4604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7095258920668866e-06,
+      "loss": 1.1716,
+      "step": 4605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7075753151744885e-06,
+      "loss": 0.7925,
+      "step": 4606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7056256225497075e-06,
+      "loss": 0.8416,
+      "step": 4607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7036768147161853e-06,
+      "loss": 0.9717,
+      "step": 4608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7017288921973251e-06,
+      "loss": 0.8075,
+      "step": 4609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6997818555162915e-06,
+      "loss": 1.007,
+      "step": 4610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6978357051960154e-06,
+      "loss": 1.0915,
+      "step": 4611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6958904417591853e-06,
+      "loss": 0.8169,
+      "step": 4612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6939460657282536e-06,
+      "loss": 1.053,
+      "step": 4613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6920025776254334e-06,
+      "loss": 0.8644,
+      "step": 4614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6900599779726984e-06,
+      "loss": 0.7715,
+      "step": 4615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6881182672917879e-06,
+      "loss": 1.5446,
+      "step": 4616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6861774461041969e-06,
+      "loss": 0.7704,
+      "step": 4617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6842375149311868e-06,
+      "loss": 0.7789,
+      "step": 4618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6822984742937764e-06,
+      "loss": 0.9984,
+      "step": 4619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.680360324712746e-06,
+      "loss": 0.8011,
+      "step": 4620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6784230667086355e-06,
+      "loss": 0.8926,
+      "step": 4621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6764867008017493e-06,
+      "loss": 1.2976,
+      "step": 4622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6745512275121466e-06,
+      "loss": 0.9938,
+      "step": 4623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.672616647359655e-06,
+      "loss": 1.1858,
+      "step": 4624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6706829608638508e-06,
+      "loss": 1.0422,
+      "step": 4625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.668750168544081e-06,
+      "loss": 1.0094,
+      "step": 4626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6668182709194448e-06,
+      "loss": 1.0852,
+      "step": 4627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.664887268508808e-06,
+      "loss": 1.126,
+      "step": 4628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.66295716183079e-06,
+      "loss": 1.1213,
+      "step": 4629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6610279514037725e-06,
+      "loss": 0.896,
+      "step": 4630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.659099637745894e-06,
+      "loss": 1.0292,
+      "step": 4631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.657172221375058e-06,
+      "loss": 1.1824,
+      "step": 4632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6552457028089203e-06,
+      "loss": 1.3334,
+      "step": 4633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6533200825648993e-06,
+      "loss": 1.0811,
+      "step": 4634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.651395361160169e-06,
+      "loss": 1.0998,
+      "step": 4635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6494715391116671e-06,
+      "loss": 1.0989,
+      "step": 4636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6475486169360867e-06,
+      "loss": 1.0946,
+      "step": 4637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6456265951498763e-06,
+      "loss": 0.8644,
+      "step": 4638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6437054742692493e-06,
+      "loss": 1.2227,
+      "step": 4639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.641785254810172e-06,
+      "loss": 1.0586,
+      "step": 4640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6398659372883696e-06,
+      "loss": 1.0983,
+      "step": 4641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6379475222193248e-06,
+      "loss": 1.1605,
+      "step": 4642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6360300101182808e-06,
+      "loss": 1.0387,
+      "step": 4643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6341134015002352e-06,
+      "loss": 1.1984,
+      "step": 4644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6321976968799435e-06,
+      "loss": 0.8331,
+      "step": 4645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6302828967719175e-06,
+      "loss": 1.309,
+      "step": 4646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6283690016904302e-06,
+      "loss": 1.3154,
+      "step": 4647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.626456012149506e-06,
+      "loss": 1.0345,
+      "step": 4648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6245439286629312e-06,
+      "loss": 1.0682,
+      "step": 4649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6226327517442453e-06,
+      "loss": 1.0849,
+      "step": 4650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.620722481906745e-06,
+      "loss": 0.8466,
+      "step": 4651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6188131196634827e-06,
+      "loss": 1.0049,
+      "step": 4652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6169046655272708e-06,
+      "loss": 0.9731,
+      "step": 4653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6149971200106723e-06,
+      "loss": 1.1445,
+      "step": 4654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.613090483626013e-06,
+      "loss": 1.193,
+      "step": 4655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6111847568853645e-06,
+      "loss": 1.0985,
+      "step": 4656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6092799403005648e-06,
+      "loss": 1.0315,
+      "step": 4657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6073760343831996e-06,
+      "loss": 1.2456,
+      "step": 4658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.605473039644616e-06,
+      "loss": 0.982,
+      "step": 4659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.603570956595913e-06,
+      "loss": 1.2182,
+      "step": 4660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6016697857479435e-06,
+      "loss": 1.1982,
+      "step": 4661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5997695276113168e-06,
+      "loss": 0.8407,
+      "step": 4662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5978701826964e-06,
+      "loss": 0.9441,
+      "step": 4663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595971751513311e-06,
+      "loss": 1.1534,
+      "step": 4664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.594074234571924e-06,
+      "loss": 1.2933,
+      "step": 4665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5921776323818655e-06,
+      "loss": 1.3954,
+      "step": 4666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5902819454525209e-06,
+      "loss": 1.1286,
+      "step": 4667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5883871742930257e-06,
+      "loss": 1.2082,
+      "step": 4668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586493319412269e-06,
+      "loss": 0.924,
+      "step": 4669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5846003813188993e-06,
+      "loss": 1.1677,
+      "step": 4670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5827083605213123e-06,
+      "loss": 0.8689,
+      "step": 4671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5808172575276615e-06,
+      "loss": 0.9857,
+      "step": 4672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5789270728458505e-06,
+      "loss": 0.9605,
+      "step": 4673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5770378069835412e-06,
+      "loss": 1.0761,
+      "step": 4674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5751494604481427e-06,
+      "loss": 1.1194,
+      "step": 4675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5732620337468258e-06,
+      "loss": 1.209,
+      "step": 4676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5713755273865017e-06,
+      "loss": 0.8477,
+      "step": 4677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5694899418738462e-06,
+      "loss": 0.793,
+      "step": 4678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5676052777152806e-06,
+      "loss": 0.9179,
+      "step": 4679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5657215354169841e-06,
+      "loss": 1.0427,
+      "step": 4680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563838715484884e-06,
+      "loss": 0.9473,
+      "step": 4681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.561956818424661e-06,
+      "loss": 1.0338,
+      "step": 4682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.560075844741747e-06,
+      "loss": 0.979,
+      "step": 4683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5581957949413295e-06,
+      "loss": 1.0631,
+      "step": 4684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5563166695283438e-06,
+      "loss": 0.9718,
+      "step": 4685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.554438469007482e-06,
+      "loss": 1.2285,
+      "step": 4686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5525611938831791e-06,
+      "loss": 1.1272,
+      "step": 4687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5506848446596317e-06,
+      "loss": 1.0194,
+      "step": 4688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548809421840779e-06,
+      "loss": 0.9593,
+      "step": 4689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546934925930319e-06,
+      "loss": 0.9557,
+      "step": 4690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5450613574316958e-06,
+      "loss": 1.2345,
+      "step": 4691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5431887168481051e-06,
+      "loss": 0.9351,
+      "step": 4692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5413170046824926e-06,
+      "loss": 1.249,
+      "step": 4693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5394462214375593e-06,
+      "loss": 0.94,
+      "step": 4694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5375763676157517e-06,
+      "loss": 1.1999,
+      "step": 4695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5357074437192688e-06,
+      "loss": 0.8426,
+      "step": 4696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5338394502500592e-06,
+      "loss": 0.9771,
+      "step": 4697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5319723877098202e-06,
+      "loss": 1.2279,
+      "step": 4698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5301062566000047e-06,
+      "loss": 0.901,
+      "step": 4699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5282410574218072e-06,
+      "loss": 1.2416,
+      "step": 4700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5263767906761817e-06,
+      "loss": 0.9892,
+      "step": 4701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5245134568638197e-06,
+      "loss": 1.2283,
+      "step": 4702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5226510564851732e-06,
+      "loss": 1.3031,
+      "step": 4703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5207895900404363e-06,
+      "loss": 1.2139,
+      "step": 4704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5189290580295584e-06,
+      "loss": 0.9628,
+      "step": 4705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5170694609522306e-06,
+      "loss": 1.1049,
+      "step": 4706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5152107993079018e-06,
+      "loss": 0.7284,
+      "step": 4707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5133530735957586e-06,
+      "loss": 0.8545,
+      "step": 4708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5114962843147468e-06,
+      "loss": 1.3491,
+      "step": 4709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5096404319635533e-06,
+      "loss": 1.221,
+      "step": 4710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5077855170406198e-06,
+      "loss": 1.2195,
+      "step": 4711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.50593154004413e-06,
+      "loss": 0.893,
+      "step": 4712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5040785014720188e-06,
+      "loss": 1.1227,
+      "step": 4713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.502226401821968e-06,
+      "loss": 1.2965,
+      "step": 4714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5003752415914098e-06,
+      "loss": 1.0722,
+      "step": 4715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.498525021277521e-06,
+      "loss": 1.1067,
+      "step": 4716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4966757413772276e-06,
+      "loss": 1.0577,
+      "step": 4717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4948274023872005e-06,
+      "loss": 1.1522,
+      "step": 4718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4929800048038633e-06,
+      "loss": 1.0392,
+      "step": 4719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4911335491233818e-06,
+      "loss": 1.1612,
+      "step": 4720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4892880358416689e-06,
+      "loss": 0.9165,
+      "step": 4721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.487443465454389e-06,
+      "loss": 1.1937,
+      "step": 4722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4855998384569491e-06,
+      "loss": 1.0686,
+      "step": 4723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.483757155344503e-06,
+      "loss": 1.3311,
+      "step": 4724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.481915416611951e-06,
+      "loss": 1.0828,
+      "step": 4725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4800746227539437e-06,
+      "loss": 1.0516,
+      "step": 4726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4782347742648734e-06,
+      "loss": 0.9543,
+      "step": 4727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4763958716388798e-06,
+      "loss": 0.9657,
+      "step": 4728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.474557915369848e-06,
+      "loss": 0.9451,
+      "step": 4729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4727209059514114e-06,
+      "loss": 1.1614,
+      "step": 4730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4708848438769456e-06,
+      "loss": 1.1002,
+      "step": 4731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4690497296395773e-06,
+      "loss": 0.9284,
+      "step": 4732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4672155637321695e-06,
+      "loss": 1.2056,
+      "step": 4733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.46538234664734e-06,
+      "loss": 1.1296,
+      "step": 4734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4635500788774448e-06,
+      "loss": 1.0049,
+      "step": 4735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4617187609145906e-06,
+      "loss": 0.7855,
+      "step": 4736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4598883932506236e-06,
+      "loss": 0.8547,
+      "step": 4737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4580589763771413e-06,
+      "loss": 1.031,
+      "step": 4738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4562305107854768e-06,
+      "loss": 1.0025,
+      "step": 4739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4544029969667167e-06,
+      "loss": 1.1791,
+      "step": 4740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.452576435411685e-06,
+      "loss": 0.8106,
+      "step": 4741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4507508266109565e-06,
+      "loss": 1.0794,
+      "step": 4742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4489261710548451e-06,
+      "loss": 0.814,
+      "step": 4743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4471024692334101e-06,
+      "loss": 1.1713,
+      "step": 4744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.445279721636453e-06,
+      "loss": 1.3861,
+      "step": 4745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4434579287535244e-06,
+      "loss": 0.9353,
+      "step": 4746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.441637091073913e-06,
+      "loss": 1.1707,
+      "step": 4747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.439817209086653e-06,
+      "loss": 0.8701,
+      "step": 4748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4379982832805206e-06,
+      "loss": 1.106,
+      "step": 4749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4361803141440384e-06,
+      "loss": 0.9026,
+      "step": 4750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4343633021654696e-06,
+      "loss": 0.7454,
+      "step": 4751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.432547247832819e-06,
+      "loss": 0.941,
+      "step": 4752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4307321516338385e-06,
+      "loss": 0.9555,
+      "step": 4753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4289180140560189e-06,
+      "loss": 0.9912,
+      "step": 4754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4271048355865946e-06,
+      "loss": 1.1888,
+      "step": 4755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4252926167125413e-06,
+      "loss": 0.9869,
+      "step": 4756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4234813579205815e-06,
+      "loss": 0.9507,
+      "step": 4757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.421671059697175e-06,
+      "loss": 1.0106,
+      "step": 4758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4198617225285244e-06,
+      "loss": 1.2044,
+      "step": 4759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.418053346900574e-06,
+      "loss": 1.2789,
+      "step": 4760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4162459332990147e-06,
+      "loss": 1.2391,
+      "step": 4761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4144394822092712e-06,
+      "loss": 0.8773,
+      "step": 4762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.412633994116518e-06,
+      "loss": 0.9553,
+      "step": 4763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4108294695056606e-06,
+      "loss": 1.3718,
+      "step": 4764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4090259088613573e-06,
+      "loss": 1.06,
+      "step": 4765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4072233126679985e-06,
+      "loss": 0.8163,
+      "step": 4766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.405421681409721e-06,
+      "loss": 1.2747,
+      "step": 4767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4036210155703989e-06,
+      "loss": 0.9483,
+      "step": 4768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4018213156336518e-06,
+      "loss": 0.8822,
+      "step": 4769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4000225820828317e-06,
+      "loss": 1.0532,
+      "step": 4770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3982248154010402e-06,
+      "loss": 0.9014,
+      "step": 4771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3964280160711119e-06,
+      "loss": 1.4194,
+      "step": 4772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3946321845756277e-06,
+      "loss": 0.8783,
+      "step": 4773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3928373213969038e-06,
+      "loss": 1.0468,
+      "step": 4774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3910434270169987e-06,
+      "loss": 0.835,
+      "step": 4775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.38925050191771e-06,
+      "loss": 0.9589,
+      "step": 4776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3874585465805734e-06,
+      "loss": 0.8376,
+      "step": 4777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3856675614868687e-06,
+      "loss": 1.0494,
+      "step": 4778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3838775471176113e-06,
+      "loss": 1.0165,
+      "step": 4779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3820885039535564e-06,
+      "loss": 0.9294,
+      "step": 4780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3803004324751974e-06,
+      "loss": 0.9894,
+      "step": 4781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.378513333162771e-06,
+      "loss": 0.8708,
+      "step": 4782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3767272064962466e-06,
+      "loss": 1.1326,
+      "step": 4783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3749420529553414e-06,
+      "loss": 0.7354,
+      "step": 4784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3731578730194983e-06,
+      "loss": 1.2693,
+      "step": 4785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3713746671679112e-06,
+      "loss": 1.0961,
+      "step": 4786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3695924358795037e-06,
+      "loss": 0.6317,
+      "step": 4787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3678111796329446e-06,
+      "loss": 1.1887,
+      "step": 4788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.366030898906635e-06,
+      "loss": 0.8604,
+      "step": 4789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3642515941787171e-06,
+      "loss": 0.7541,
+      "step": 4790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3624732659270684e-06,
+      "loss": 0.7827,
+      "step": 4791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3606959146293086e-06,
+      "loss": 1.0219,
+      "step": 4792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3589195407627898e-06,
+      "loss": 1.0603,
+      "step": 4793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3571441448046086e-06,
+      "loss": 0.9947,
+      "step": 4794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.355369727231588e-06,
+      "loss": 1.1177,
+      "step": 4795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3535962885202997e-06,
+      "loss": 1.1972,
+      "step": 4796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3518238291470448e-06,
+      "loss": 1.0853,
+      "step": 4797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.350052349587866e-06,
+      "loss": 1.1251,
+      "step": 4798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3482818503185402e-06,
+      "loss": 1.1449,
+      "step": 4799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3465123318145817e-06,
+      "loss": 0.9919,
+      "step": 4800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3447437945512398e-06,
+      "loss": 1.0595,
+      "step": 4801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.342976239003505e-06,
+      "loss": 0.9914,
+      "step": 4802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3412096656461e-06,
+      "loss": 0.9347,
+      "step": 4803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339444074953482e-06,
+      "loss": 1.3728,
+      "step": 4804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3376794673998504e-06,
+      "loss": 1.151,
+      "step": 4805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.335915843459137e-06,
+      "loss": 1.0668,
+      "step": 4806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3341532036050081e-06,
+      "loss": 1.1345,
+      "step": 4807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3323915483108662e-06,
+      "loss": 1.0771,
+      "step": 4808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3306308780498534e-06,
+      "loss": 1.0136,
+      "step": 4809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3288711932948427e-06,
+      "loss": 1.0263,
+      "step": 4810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3271124945184439e-06,
+      "loss": 1.4203,
+      "step": 4811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3253547821930002e-06,
+      "loss": 0.7562,
+      "step": 4812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3235980567905954e-06,
+      "loss": 1.0096,
+      "step": 4813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3218423187830409e-06,
+      "loss": 1.0162,
+      "step": 4814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3200875686418908e-06,
+      "loss": 1.4983,
+      "step": 4815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3183338068384243e-06,
+      "loss": 1.1835,
+      "step": 4816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3165810338436647e-06,
+      "loss": 0.8973,
+      "step": 4817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3148292501283627e-06,
+      "loss": 1.3385,
+      "step": 4818
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3130784561630083e-06,
+      "loss": 1.1987,
+      "step": 4819
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3113286524178232e-06,
+      "loss": 1.4256,
+      "step": 4820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3095798393627617e-06,
+      "loss": 1.1941,
+      "step": 4821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3078320174675141e-06,
+      "loss": 0.9447,
+      "step": 4822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3060851872015058e-06,
+      "loss": 1.0355,
+      "step": 4823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3043393490338918e-06,
+      "loss": 1.0413,
+      "step": 4824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3025945034335664e-06,
+      "loss": 1.0153,
+      "step": 4825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3008506508691516e-06,
+      "loss": 1.2494,
+      "step": 4826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2991077918090056e-06,
+      "loss": 1.0172,
+      "step": 4827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2973659267212173e-06,
+      "loss": 1.1897,
+      "step": 4828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2956250560736143e-06,
+      "loss": 1.2541,
+      "step": 4829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2938851803337516e-06,
+      "loss": 1.1815,
+      "step": 4830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2921462999689182e-06,
+      "loss": 1.1793,
+      "step": 4831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.290408415446136e-06,
+      "loss": 0.9005,
+      "step": 4832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2886715272321616e-06,
+      "loss": 0.9439,
+      "step": 4833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2869356357934815e-06,
+      "loss": 1.0068,
+      "step": 4834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2852007415963136e-06,
+      "loss": 1.0724,
+      "step": 4835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2834668451066118e-06,
+      "loss": 0.9992,
+      "step": 4836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2817339467900597e-06,
+      "loss": 1.0162,
+      "step": 4837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2800020471120717e-06,
+      "loss": 1.0767,
+      "step": 4838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.278271146537794e-06,
+      "loss": 0.7836,
+      "step": 4839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.276541245532109e-06,
+      "loss": 1.0715,
+      "step": 4840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2748123445596255e-06,
+      "loss": 1.3068,
+      "step": 4841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2730844440846862e-06,
+      "loss": 1.209,
+      "step": 4842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2713575445713616e-06,
+      "loss": 0.9778,
+      "step": 4843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2696316464834607e-06,
+      "loss": 1.2487,
+      "step": 4844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2679067502845144e-06,
+      "loss": 0.963,
+      "step": 4845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2661828564377948e-06,
+      "loss": 1.4915,
+      "step": 4846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2644599654062934e-06,
+      "loss": 1.2717,
+      "step": 4847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2627380776527415e-06,
+      "loss": 1.1984,
+      "step": 4848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2610171936395949e-06,
+      "loss": 0.8577,
+      "step": 4849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.259297313829046e-06,
+      "loss": 1.1547,
+      "step": 4850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.257578438683012e-06,
+      "loss": 0.8131,
+      "step": 4851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.255860568663142e-06,
+      "loss": 1.0253,
+      "step": 4852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.254143704230814e-06,
+      "loss": 0.9722,
+      "step": 4853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2524278458471411e-06,
+      "loss": 1.0301,
+      "step": 4854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2507129939729583e-06,
+      "loss": 1.0946,
+      "step": 4855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.248999149068838e-06,
+      "loss": 1.1481,
+      "step": 4856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2472863115950768e-06,
+      "loss": 1.0875,
+      "step": 4857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2455744820117028e-06,
+      "loss": 1.364,
+      "step": 4858
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2438636607784722e-06,
+      "loss": 1.0891,
+      "step": 4859
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2421538483548706e-06,
+      "loss": 0.7168,
+      "step": 4860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2404450452001154e-06,
+      "loss": 1.0488,
+      "step": 4861
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2387372517731505e-06,
+      "loss": 1.1525,
+      "step": 4862
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.237030468532648e-06,
+      "loss": 0.9124,
+      "step": 4863
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2353246959370086e-06,
+      "loss": 0.9189,
+      "step": 4864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2336199344443656e-06,
+      "loss": 1.1384,
+      "step": 4865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2319161845125744e-06,
+      "loss": 0.8892,
+      "step": 4866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.230213446599226e-06,
+      "loss": 1.1183,
+      "step": 4867
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.228511721161631e-06,
+      "loss": 1.0918,
+      "step": 4868
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.226811008656837e-06,
+      "loss": 0.9774,
+      "step": 4869
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2251113095416113e-06,
+      "loss": 0.9684,
+      "step": 4870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2234126242724565e-06,
+      "loss": 1.0594,
+      "step": 4871
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2217149533055976e-06,
+      "loss": 1.1466,
+      "step": 4872
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2200182970969899e-06,
+      "loss": 1.129,
+      "step": 4873
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2183226561023132e-06,
+      "loss": 0.9693,
+      "step": 4874
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.216628030776979e-06,
+      "loss": 1.0372,
+      "step": 4875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2149344215761216e-06,
+      "loss": 1.0008,
+      "step": 4876
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2132418289546089e-06,
+      "loss": 1.2212,
+      "step": 4877
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2115502533670253e-06,
+      "loss": 0.9454,
+      "step": 4878
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2098596952676921e-06,
+      "loss": 0.9416,
+      "step": 4879
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2081701551106506e-06,
+      "loss": 0.9737,
+      "step": 4880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2064816333496749e-06,
+      "loss": 1.2445,
+      "step": 4881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.20479413043826e-06,
+      "loss": 1.0358,
+      "step": 4882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.20310764682963e-06,
+      "loss": 0.9615,
+      "step": 4883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.201422182976732e-06,
+      "loss": 0.7799,
+      "step": 4884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1997377393322463e-06,
+      "loss": 0.841,
+      "step": 4885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1980543163485726e-06,
+      "loss": 0.8414,
+      "step": 4886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1963719144778363e-06,
+      "loss": 0.8408,
+      "step": 4887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1946905341718951e-06,
+      "loss": 1.112,
+      "step": 4888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1930101758823259e-06,
+      "loss": 0.8527,
+      "step": 4889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1913308400604339e-06,
+      "loss": 0.9557,
+      "step": 4890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1896525271572469e-06,
+      "loss": 1.4533,
+      "step": 4891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1879752376235231e-06,
+      "loss": 1.3497,
+      "step": 4892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1862989719097423e-06,
+      "loss": 1.1625,
+      "step": 4893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1846237304661095e-06,
+      "loss": 1.2579,
+      "step": 4894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1829495137425534e-06,
+      "loss": 1.1064,
+      "step": 4895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.181276322188732e-06,
+      "loss": 1.0397,
+      "step": 4896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1796041562540217e-06,
+      "loss": 0.8734,
+      "step": 4897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1779330163875325e-06,
+      "loss": 0.9812,
+      "step": 4898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1762629030380867e-06,
+      "loss": 1.3881,
+      "step": 4899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1745938166542414e-06,
+      "loss": 0.8899,
+      "step": 4900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.172925757684271e-06,
+      "loss": 0.8385,
+      "step": 4901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712587265761799e-06,
+      "loss": 0.9198,
+      "step": 4902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1695927237776922e-06,
+      "loss": 1.2821,
+      "step": 4903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1679277497362563e-06,
+      "loss": 0.8323,
+      "step": 4904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1662638048990443e-06,
+      "loss": 1.1166,
+      "step": 4905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1646008897129546e-06,
+      "loss": 1.0523,
+      "step": 4906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1629390046246042e-06,
+      "loss": 1.1061,
+      "step": 4907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.161278150080341e-06,
+      "loss": 0.8867,
+      "step": 4908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1596183265262256e-06,
+      "loss": 0.8495,
+      "step": 4909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.157959534408052e-06,
+      "loss": 1.5488,
+      "step": 4910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1563017741713289e-06,
+      "loss": 0.9813,
+      "step": 4911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1546450462612951e-06,
+      "loss": 1.1766,
+      "step": 4912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1529893511229068e-06,
+      "loss": 0.8371,
+      "step": 4913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.151334689200845e-06,
+      "loss": 0.8684,
+      "step": 4914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1496810609395115e-06,
+      "loss": 0.8209,
+      "step": 4915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1480284667830343e-06,
+      "loss": 1.1174,
+      "step": 4916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1463769071752606e-06,
+      "loss": 0.9788,
+      "step": 4917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1447263825597577e-06,
+      "loss": 1.1551,
+      "step": 4918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1430768933798215e-06,
+      "loss": 1.0487,
+      "step": 4919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1414284400784643e-06,
+      "loss": 1.0228,
+      "step": 4920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1397810230984207e-06,
+      "loss": 1.1702,
+      "step": 4921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1381346428821482e-06,
+      "loss": 1.2297,
+      "step": 4922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.136489299871828e-06,
+      "loss": 1.0329,
+      "step": 4923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.134844994509358e-06,
+      "loss": 1.1709,
+      "step": 4924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.133201727236361e-06,
+      "loss": 1.2174,
+      "step": 4925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1315594984941786e-06,
+      "loss": 1.2314,
+      "step": 4926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1299183087238764e-06,
+      "loss": 0.9408,
+      "step": 4927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1282781583662372e-06,
+      "loss": 1.2457,
+      "step": 4928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1266390478617711e-06,
+      "loss": 1.0561,
+      "step": 4929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1250009776506982e-06,
+      "loss": 1.066,
+      "step": 4930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1233639481729703e-06,
+      "loss": 1.0988,
+      "step": 4931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1217279598682518e-06,
+      "loss": 0.856,
+      "step": 4932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1200930131759337e-06,
+      "loss": 1.0681,
+      "step": 4933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.118459108535122e-06,
+      "loss": 0.7947,
+      "step": 4934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1168262463846457e-06,
+      "loss": 0.9094,
+      "step": 4935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1151944271630517e-06,
+      "loss": 0.9224,
+      "step": 4936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1135636513086107e-06,
+      "loss": 1.0245,
+      "step": 4937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1119339192593077e-06,
+      "loss": 1.1326,
+      "step": 4938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1103052314528545e-06,
+      "loss": 1.1959,
+      "step": 4939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1086775883266725e-06,
+      "loss": 1.0329,
+      "step": 4940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1070509903179127e-06,
+      "loss": 1.0258,
+      "step": 4941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1054254378634399e-06,
+      "loss": 0.8963,
+      "step": 4942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.103800931399837e-06,
+      "loss": 1.1042,
+      "step": 4943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.102177471363412e-06,
+      "loss": 1.099,
+      "step": 4944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1005550581901853e-06,
+      "loss": 0.9894,
+      "step": 4945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0989336923158999e-06,
+      "loss": 1.2632,
+      "step": 4946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.097313374176014e-06,
+      "loss": 1.0593,
+      "step": 4947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0956941042057106e-06,
+      "loss": 0.9696,
+      "step": 4948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.094075882839884e-06,
+      "loss": 1.4002,
+      "step": 4949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0924587105131546e-06,
+      "loss": 1.0192,
+      "step": 4950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0908425876598512e-06,
+      "loss": 1.1336,
+      "step": 4951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0892275147140307e-06,
+      "loss": 1.1337,
+      "step": 4952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0876134921094606e-06,
+      "loss": 0.8264,
+      "step": 4953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.086000520279632e-06,
+      "loss": 1.2135,
+      "step": 4954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0843885996577503e-06,
+      "loss": 0.9619,
+      "step": 4955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0827777306767384e-06,
+      "loss": 1.117,
+      "step": 4956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0811679137692367e-06,
+      "loss": 1.2456,
+      "step": 4957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0795591493676072e-06,
+      "loss": 1.0099,
+      "step": 4958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.077951437903923e-06,
+      "loss": 1.0946,
+      "step": 4959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0763447798099813e-06,
+      "loss": 0.9423,
+      "step": 4960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0747391755172877e-06,
+      "loss": 0.866,
+      "step": 4961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0731346254570735e-06,
+      "loss": 1.1151,
+      "step": 4962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0715311300602805e-06,
+      "loss": 0.94,
+      "step": 4963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0699286897575718e-06,
+      "loss": 0.8686,
+      "step": 4964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0683273049793242e-06,
+      "loss": 0.9166,
+      "step": 4965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.066726976155632e-06,
+      "loss": 0.8575,
+      "step": 4966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0651277037163038e-06,
+      "loss": 1.1442,
+      "step": 4967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0635294880908702e-06,
+      "loss": 1.1201,
+      "step": 4968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.061932329708572e-06,
+      "loss": 1.0834,
+      "step": 4969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0603362289983687e-06,
+      "loss": 1.3317,
+      "step": 4970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0587411863889336e-06,
+      "loss": 1.1824,
+      "step": 4971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0571472023086604e-06,
+      "loss": 1.0024,
+      "step": 4972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0555542771856541e-06,
+      "loss": 1.1193,
+      "step": 4973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.053962411447736e-06,
+      "loss": 1.3651,
+      "step": 4974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0523716055224458e-06,
+      "loss": 1.1669,
+      "step": 4975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0507818598370355e-06,
+      "loss": 1.3306,
+      "step": 4976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0491931748184736e-06,
+      "loss": 1.109,
+      "step": 4977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0476055508934408e-06,
+      "loss": 0.8911,
+      "step": 4978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0460189884883393e-06,
+      "loss": 0.9793,
+      "step": 4979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0444334880292794e-06,
+      "loss": 0.8409,
+      "step": 4980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0428490499420935e-06,
+      "loss": 1.2276,
+      "step": 4981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0412656746523182e-06,
+      "loss": 1.056,
+      "step": 4982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0396833625852148e-06,
+      "loss": 0.9194,
+      "step": 4983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0381021141657526e-06,
+      "loss": 1.2208,
+      "step": 4984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0365219298186207e-06,
+      "loss": 0.9974,
+      "step": 4985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0349428099682173e-06,
+      "loss": 1.117,
+      "step": 4986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0333647550386572e-06,
+      "loss": 1.4076,
+      "step": 4987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0317877654537672e-06,
+      "loss": 1.0419,
+      "step": 4988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0302118416370926e-06,
+      "loss": 1.3794,
+      "step": 4989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0286369840118859e-06,
+      "loss": 0.7521,
+      "step": 4990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0270631930011216e-06,
+      "loss": 1.1748,
+      "step": 4991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025490469027477e-06,
+      "loss": 1.1264,
+      "step": 4992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0239188125133525e-06,
+      "loss": 1.0867,
+      "step": 4993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0223482238808557e-06,
+      "loss": 0.8522,
+      "step": 4994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0207787035518119e-06,
+      "loss": 1.0846,
+      "step": 4995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0192102519477565e-06,
+      "loss": 0.9919,
+      "step": 4996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0176428694899382e-06,
+      "loss": 1.1929,
+      "step": 4997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016076556599318e-06,
+      "loss": 1.1758,
+      "step": 4998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0145113136965728e-06,
+      "loss": 1.0872,
+      "step": 4999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0129471412020886e-06,
+      "loss": 1.2245,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0113840395359658e-06,
+      "loss": 0.8824,
+      "step": 5001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0098220091180145e-06,
+      "loss": 1.0325,
+      "step": 5002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0082610503677625e-06,
+      "loss": 1.2299,
+      "step": 5003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.006701163704445e-06,
+      "loss": 0.8885,
+      "step": 5004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0051423495470098e-06,
+      "loss": 1.1012,
+      "step": 5005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0035846083141193e-06,
+      "loss": 1.0848,
+      "step": 5006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0020279404241456e-06,
+      "loss": 0.7578,
+      "step": 5007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0004723462951732e-06,
+      "loss": 1.0458,
+      "step": 5008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.989178263449956e-07,
+      "loss": 1.1849,
+      "step": 5009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.973643809911238e-07,
+      "loss": 0.9067,
+      "step": 5010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.958120106507729e-07,
+      "loss": 0.9718,
+      "step": 5011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.942607157408784e-07,
+      "loss": 0.9549,
+      "step": 5012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.927104966780749e-07,
+      "loss": 0.9735,
+      "step": 5013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.911613538787196e-07,
+      "loss": 1.0065,
+      "step": 5014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.896132877588721e-07,
+      "loss": 1.165,
+      "step": 5015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.880662987343103e-07,
+      "loss": 1.0132,
+      "step": 5016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.865203872205165e-07,
+      "loss": 0.9683,
+      "step": 5017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.849755536326866e-07,
+      "loss": 1.1187,
+      "step": 5018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834317983857256e-07,
+      "loss": 0.87,
+      "step": 5019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.818891218942511e-07,
+      "loss": 1.1801,
+      "step": 5020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.803475245725885e-07,
+      "loss": 1.1671,
+      "step": 5021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78807006834777e-07,
+      "loss": 1.2275,
+      "step": 5022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.772675690945593e-07,
+      "loss": 1.189,
+      "step": 5023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.757292117653955e-07,
+      "loss": 0.9301,
+      "step": 5024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.741919352604506e-07,
+      "loss": 0.9392,
+      "step": 5025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.726557399925995e-07,
+      "loss": 0.9138,
+      "step": 5026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.71120626374431e-07,
+      "loss": 1.2175,
+      "step": 5027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.695865948182392e-07,
+      "loss": 1.0011,
+      "step": 5028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.680536457360284e-07,
+      "loss": 0.8381,
+      "step": 5029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66521779539511e-07,
+      "loss": 1.2436,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.649909966401134e-07,
+      "loss": 0.9662,
+      "step": 5031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63461297448966e-07,
+      "loss": 1.0122,
+      "step": 5032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.619326823769104e-07,
+      "loss": 1.3076,
+      "step": 5033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.604051518344948e-07,
+      "loss": 0.8921,
+      "step": 5034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.588787062319804e-07,
+      "loss": 1.0921,
+      "step": 5035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57353345979332e-07,
+      "loss": 0.8834,
+      "step": 5036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.55829071486229e-07,
+      "loss": 0.8841,
+      "step": 5037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.543058831620528e-07,
+      "loss": 1.1068,
+      "step": 5038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.527837814158963e-07,
+      "loss": 1.456,
+      "step": 5039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.512627666565588e-07,
+      "loss": 0.9676,
+      "step": 5040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.497428392925523e-07,
+      "loss": 1.1532,
+      "step": 5041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.482239997320903e-07,
+      "loss": 1.1454,
+      "step": 5042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.467062483831008e-07,
+      "loss": 1.1958,
+      "step": 5043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.451895856532117e-07,
+      "loss": 0.7108,
+      "step": 5044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.436740119497661e-07,
+      "loss": 1.2757,
+      "step": 5045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.421595276798084e-07,
+      "loss": 1.054,
+      "step": 5046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.406461332500972e-07,
+      "loss": 1.0477,
+      "step": 5047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.39133829067092e-07,
+      "loss": 0.9057,
+      "step": 5048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.376226155369628e-07,
+      "loss": 0.9437,
+      "step": 5049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.361124930655841e-07,
+      "loss": 0.9313,
+      "step": 5050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.346034620585426e-07,
+      "loss": 0.8278,
+      "step": 5051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.330955229211259e-07,
+      "loss": 1.1452,
+      "step": 5052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31588676058332e-07,
+      "loss": 1.0093,
+      "step": 5053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.300829218748625e-07,
+      "loss": 1.1703,
+      "step": 5054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.285782607751298e-07,
+      "loss": 0.942,
+      "step": 5055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.270746931632501e-07,
+      "loss": 1.1489,
+      "step": 5056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.255722194430444e-07,
+      "loss": 1.1116,
+      "step": 5057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.240708400180437e-07,
+      "loss": 0.973,
+      "step": 5058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.225705552914827e-07,
+      "loss": 0.9461,
+      "step": 5059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.210713656663023e-07,
+      "loss": 1.1876,
+      "step": 5060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.195732715451477e-07,
+      "loss": 1.2176,
+      "step": 5061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.180762733303745e-07,
+      "loss": 0.8684,
+      "step": 5062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.165803714240401e-07,
+      "loss": 0.9983,
+      "step": 5063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.150855662279079e-07,
+      "loss": 0.9491,
+      "step": 5064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.135918581434467e-07,
+      "loss": 1.1338,
+      "step": 5065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.120992475718333e-07,
+      "loss": 1.109,
+      "step": 5066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.106077349139453e-07,
+      "loss": 0.9949,
+      "step": 5067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.091173205703708e-07,
+      "loss": 1.0173,
+      "step": 5068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.076280049413977e-07,
+      "loss": 1.0219,
+      "step": 5069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.061397884270217e-07,
+      "loss": 1.2241,
+      "step": 5070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.046526714269416e-07,
+      "loss": 0.8522,
+      "step": 5071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.031666543405637e-07,
+      "loss": 1.088,
+      "step": 5072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.016817375669956e-07,
+      "loss": 1.0336,
+      "step": 5073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.001979215050544e-07,
+      "loss": 1.1514,
+      "step": 5074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.987152065532522e-07,
+      "loss": 1.2673,
+      "step": 5075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.972335931098159e-07,
+      "loss": 0.9349,
+      "step": 5076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.957530815726695e-07,
+      "loss": 1.1699,
+      "step": 5077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.942736723394458e-07,
+      "loss": 0.9805,
+      "step": 5078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.927953658074773e-07,
+      "loss": 1.1766,
+      "step": 5079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.913181623738032e-07,
+      "loss": 1.0554,
+      "step": 5080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.898420624351639e-07,
+      "loss": 0.8997,
+      "step": 5081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.883670663880078e-07,
+      "loss": 1.1196,
+      "step": 5082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.86893174628482e-07,
+      "loss": 0.9476,
+      "step": 5083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.854203875524403e-07,
+      "loss": 1.2104,
+      "step": 5084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.839487055554374e-07,
+      "loss": 1.2515,
+      "step": 5085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.824781290327317e-07,
+      "loss": 1.1498,
+      "step": 5086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.810086583792877e-07,
+      "loss": 1.0859,
+      "step": 5087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.795402939897679e-07,
+      "loss": 0.922,
+      "step": 5088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.780730362585443e-07,
+      "loss": 1.1207,
+      "step": 5089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.766068855796833e-07,
+      "loss": 1.1554,
+      "step": 5090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.751418423469609e-07,
+      "loss": 1.0068,
+      "step": 5091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.736779069538521e-07,
+      "loss": 0.9471,
+      "step": 5092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.722150797935364e-07,
+      "loss": 1.2855,
+      "step": 5093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.707533612588948e-07,
+      "loss": 1.093,
+      "step": 5094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.692927517425092e-07,
+      "loss": 1.1395,
+      "step": 5095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.67833251636665e-07,
+      "loss": 0.9874,
+      "step": 5096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.663748613333506e-07,
+      "loss": 1.1155,
+      "step": 5097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.649175812242532e-07,
+      "loss": 1.1761,
+      "step": 5098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.634614117007673e-07,
+      "loss": 0.9637,
+      "step": 5099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.62006353153983e-07,
+      "loss": 1.019,
+      "step": 5100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.605524059746955e-07,
+      "loss": 1.0188,
+      "step": 5101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.590995705533994e-07,
+      "loss": 0.9211,
+      "step": 5102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.576478472802946e-07,
+      "loss": 0.9709,
+      "step": 5103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.561972365452775e-07,
+      "loss": 1.0825,
+      "step": 5104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.547477387379505e-07,
+      "loss": 1.0352,
+      "step": 5105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.532993542476108e-07,
+      "loss": 1.2566,
+      "step": 5106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.518520834632632e-07,
+      "loss": 1.0815,
+      "step": 5107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.504059267736097e-07,
+      "loss": 1.0525,
+      "step": 5108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.489608845670527e-07,
+      "loss": 1.0141,
+      "step": 5109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.475169572316988e-07,
+      "loss": 1.0904,
+      "step": 5110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.460741451553517e-07,
+      "loss": 0.828,
+      "step": 5111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.446324487255164e-07,
+      "loss": 1.2668,
+      "step": 5112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.431918683293977e-07,
+      "loss": 1.0568,
+      "step": 5113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.417524043539038e-07,
+      "loss": 1.1205,
+      "step": 5114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.403140571856389e-07,
+      "loss": 1.4513,
+      "step": 5115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.388768272109105e-07,
+      "loss": 1.0578,
+      "step": 5116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.374407148157221e-07,
+      "loss": 0.8594,
+      "step": 5117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.36005720385783e-07,
+      "loss": 1.2161,
+      "step": 5118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.34571844306496e-07,
+      "loss": 0.9275,
+      "step": 5119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.331390869629702e-07,
+      "loss": 1.0794,
+      "step": 5120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.317074487400056e-07,
+      "loss": 1.1028,
+      "step": 5121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.302769300221098e-07,
+      "loss": 0.7937,
+      "step": 5122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.288475311934841e-07,
+      "loss": 0.9879,
+      "step": 5123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.274192526380337e-07,
+      "loss": 1.1401,
+      "step": 5124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.259920947393585e-07,
+      "loss": 0.9272,
+      "step": 5125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.24566057880763e-07,
+      "loss": 0.841,
+      "step": 5126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.231411424452418e-07,
+      "loss": 1.1826,
+      "step": 5127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.217173488154972e-07,
+      "loss": 0.9938,
+      "step": 5128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.202946773739246e-07,
+      "loss": 1.0642,
+      "step": 5129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.188731285026219e-07,
+      "loss": 0.9183,
+      "step": 5130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.174527025833828e-07,
+      "loss": 0.9376,
+      "step": 5131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.160333999977004e-07,
+      "loss": 0.9644,
+      "step": 5132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.146152211267644e-07,
+      "loss": 1.1039,
+      "step": 5133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131981663514665e-07,
+      "loss": 0.9687,
+      "step": 5134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.117822360523931e-07,
+      "loss": 1.3929,
+      "step": 5135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.103674306098291e-07,
+      "loss": 0.9994,
+      "step": 5136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.089537504037581e-07,
+      "loss": 1.3045,
+      "step": 5137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.075411958138623e-07,
+      "loss": 1.1036,
+      "step": 5138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.061297672195201e-07,
+      "loss": 0.9207,
+      "step": 5139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.047194649998063e-07,
+      "loss": 1.0543,
+      "step": 5140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.033102895334971e-07,
+      "loss": 1.0276,
+      "step": 5141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.019022411990634e-07,
+      "loss": 0.8566,
+      "step": 5142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.004953203746729e-07,
+      "loss": 0.9649,
+      "step": 5143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.99089527438191e-07,
+      "loss": 1.0506,
+      "step": 5144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.976848627671818e-07,
+      "loss": 1.1584,
+      "step": 5145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.962813267389052e-07,
+      "loss": 0.8654,
+      "step": 5146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.94878919730317e-07,
+      "loss": 0.7591,
+      "step": 5147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.93477642118069e-07,
+      "loss": 1.0678,
+      "step": 5148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.920774942785153e-07,
+      "loss": 0.9497,
+      "step": 5149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.906784765876985e-07,
+      "loss": 0.9429,
+      "step": 5150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.892805894213668e-07,
+      "loss": 0.769,
+      "step": 5151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.878838331549538e-07,
+      "loss": 1.1447,
+      "step": 5152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.864882081635993e-07,
+      "loss": 1.0445,
+      "step": 5153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.850937148221332e-07,
+      "loss": 0.8444,
+      "step": 5154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.837003535050852e-07,
+      "loss": 1.2632,
+      "step": 5155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.823081245866776e-07,
+      "loss": 0.9691,
+      "step": 5156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.809170284408335e-07,
+      "loss": 0.9877,
+      "step": 5157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.795270654411635e-07,
+      "loss": 1.0683,
+      "step": 5158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.781382359609829e-07,
+      "loss": 1.354,
+      "step": 5159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.767505403732961e-07,
+      "loss": 0.9734,
+      "step": 5160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.753639790508077e-07,
+      "loss": 0.6062,
+      "step": 5161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.739785523659144e-07,
+      "loss": 0.7232,
+      "step": 5162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.725942606907083e-07,
+      "loss": 1.0653,
+      "step": 5163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.712111043969772e-07,
+      "loss": 1.0104,
+      "step": 5164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.698290838562062e-07,
+      "loss": 1.1013,
+      "step": 5165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.684481994395726e-07,
+      "loss": 1.0127,
+      "step": 5166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.670684515179488e-07,
+      "loss": 1.0195,
+      "step": 5167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.656898404619029e-07,
+      "loss": 0.9183,
+      "step": 5168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.643123666416963e-07,
+      "loss": 1.0455,
+      "step": 5169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.629360304272882e-07,
+      "loss": 1.1944,
+      "step": 5170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.615608321883278e-07,
+      "loss": 0.9912,
+      "step": 5171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.601867722941642e-07,
+      "loss": 0.9507,
+      "step": 5172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.588138511138326e-07,
+      "loss": 0.863,
+      "step": 5173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.57442069016071e-07,
+      "loss": 1.0497,
+      "step": 5174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.560714263693059e-07,
+      "loss": 0.8462,
+      "step": 5175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.547019235416609e-07,
+      "loss": 0.9918,
+      "step": 5176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.533335609009518e-07,
+      "loss": 1.0247,
+      "step": 5177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.519663388146886e-07,
+      "loss": 0.9543,
+      "step": 5178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.506002576500732e-07,
+      "loss": 1.1421,
+      "step": 5179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.492353177740047e-07,
+      "loss": 1.0587,
+      "step": 5180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.478715195530728e-07,
+      "loss": 0.9968,
+      "step": 5181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.465088633535639e-07,
+      "loss": 1.3883,
+      "step": 5182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.451473495414513e-07,
+      "loss": 0.947,
+      "step": 5183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.437869784824086e-07,
+      "loss": 0.9399,
+      "step": 5184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.424277505417971e-07,
+      "loss": 1.0712,
+      "step": 5185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.410696660846761e-07,
+      "loss": 1.2537,
+      "step": 5186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.397127254757918e-07,
+      "loss": 0.9294,
+      "step": 5187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.383569290795911e-07,
+      "loss": 1.1498,
+      "step": 5188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.370022772602036e-07,
+      "loss": 0.8904,
+      "step": 5189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.356487703814602e-07,
+      "loss": 1.3192,
+      "step": 5190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.342964088068799e-07,
+      "loss": 1.0514,
+      "step": 5191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.329451928996745e-07,
+      "loss": 1.1545,
+      "step": 5192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.315951230227502e-07,
+      "loss": 0.8751,
+      "step": 5193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.302461995387033e-07,
+      "loss": 0.8147,
+      "step": 5194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.288984228098228e-07,
+      "loss": 1.2418,
+      "step": 5195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.275517931980886e-07,
+      "loss": 1.1786,
+      "step": 5196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.262063110651757e-07,
+      "loss": 0.9521,
+      "step": 5197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.24861976772448e-07,
+      "loss": 0.9283,
+      "step": 5198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.235187906809616e-07,
+      "loss": 1.0494,
+      "step": 5199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.22176753151464e-07,
+      "loss": 1.2711,
+      "step": 5200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.208358645443964e-07,
+      "loss": 1.124,
+      "step": 5201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.194961252198885e-07,
+      "loss": 1.1062,
+      "step": 5202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.181575355377651e-07,
+      "loss": 1.0727,
+      "step": 5203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.168200958575361e-07,
+      "loss": 1.0615,
+      "step": 5204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.15483806538409e-07,
+      "loss": 1.1331,
+      "step": 5205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.141486679392778e-07,
+      "loss": 0.9179,
+      "step": 5206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.128146804187313e-07,
+      "loss": 0.8363,
+      "step": 5207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.114818443350463e-07,
+      "loss": 0.9791,
+      "step": 5208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.101501600461908e-07,
+      "loss": 1.0036,
+      "step": 5209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.088196279098225e-07,
+      "loss": 1.2314,
+      "step": 5210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.074902482832934e-07,
+      "loss": 1.1788,
+      "step": 5211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.061620215236415e-07,
+      "loss": 0.8241,
+      "step": 5212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.048349479875999e-07,
+      "loss": 0.8153,
+      "step": 5213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.035090280315854e-07,
+      "loss": 1.0662,
+      "step": 5214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.021842620117119e-07,
+      "loss": 1.0844,
+      "step": 5215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.008606502837784e-07,
+      "loss": 0.7626,
+      "step": 5216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.995381932032774e-07,
+      "loss": 0.9228,
+      "step": 5217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.982168911253895e-07,
+      "loss": 0.7952,
+      "step": 5218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.968967444049846e-07,
+      "loss": 0.9503,
+      "step": 5219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.955777533966212e-07,
+      "loss": 1.0137,
+      "step": 5220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.942599184545529e-07,
+      "loss": 0.9467,
+      "step": 5221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.929432399327174e-07,
+      "loss": 0.9522,
+      "step": 5222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.91627718184742e-07,
+      "loss": 0.846,
+      "step": 5223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.903133535639467e-07,
+      "loss": 1.0409,
+      "step": 5224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.890001464233392e-07,
+      "loss": 1.1143,
+      "step": 5225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.876880971156147e-07,
+      "loss": 1.2225,
+      "step": 5226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.863772059931578e-07,
+      "loss": 0.7569,
+      "step": 5227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.850674734080454e-07,
+      "loss": 1.1768,
+      "step": 5228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.837588997120392e-07,
+      "loss": 1.0908,
+      "step": 5229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.824514852565922e-07,
+      "loss": 0.9907,
+      "step": 5230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.811452303928429e-07,
+      "loss": 1.0901,
+      "step": 5231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.798401354716233e-07,
+      "loss": 1.0104,
+      "step": 5232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.785362008434493e-07,
+      "loss": 1.2447,
+      "step": 5233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.772334268585296e-07,
+      "loss": 0.919,
+      "step": 5234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.759318138667543e-07,
+      "loss": 0.8721,
+      "step": 5235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.746313622177097e-07,
+      "loss": 1.3784,
+      "step": 5236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.733320722606634e-07,
+      "loss": 0.9416,
+      "step": 5237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.720339443445772e-07,
+      "loss": 1.073,
+      "step": 5238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.707369788180956e-07,
+      "loss": 0.8036,
+      "step": 5239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.694411760295538e-07,
+      "loss": 1.0317,
+      "step": 5240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.681465363269723e-07,
+      "loss": 0.9775,
+      "step": 5241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.66853060058063e-07,
+      "loss": 0.9103,
+      "step": 5242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.655607475702214e-07,
+      "loss": 0.9385,
+      "step": 5243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642695992105347e-07,
+      "loss": 0.8945,
+      "step": 5244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.629796153257706e-07,
+      "loss": 1.2986,
+      "step": 5245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.61690796262392e-07,
+      "loss": 1.048,
+      "step": 5246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.604031423665436e-07,
+      "loss": 1.0263,
+      "step": 5247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.591166539840599e-07,
+      "loss": 1.2955,
+      "step": 5248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.578313314604612e-07,
+      "loss": 1.1445,
+      "step": 5249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.565471751409541e-07,
+      "loss": 1.1175,
+      "step": 5250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.552641853704328e-07,
+      "loss": 1.1447,
+      "step": 5251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.539823624934777e-07,
+      "loss": 0.9868,
+      "step": 5252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.527017068543578e-07,
+      "loss": 1.051,
+      "step": 5253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.514222187970248e-07,
+      "loss": 0.999,
+      "step": 5254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.501438986651226e-07,
+      "loss": 0.8405,
+      "step": 5255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.488667468019727e-07,
+      "loss": 1.03,
+      "step": 5256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.475907635505924e-07,
+      "loss": 1.0356,
+      "step": 5257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.46315949253678e-07,
+      "loss": 1.3794,
+      "step": 5258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.450423042536164e-07,
+      "loss": 0.8716,
+      "step": 5259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.437698288924777e-07,
+      "loss": 1.1327,
+      "step": 5260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.424985235120191e-07,
+      "loss": 1.2836,
+      "step": 5261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.412283884536818e-07,
+      "loss": 1.1562,
+      "step": 5262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.399594240585966e-07,
+      "loss": 1.0147,
+      "step": 5263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.38691630667575e-07,
+      "loss": 0.8958,
+      "step": 5264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.374250086211192e-07,
+      "loss": 0.9735,
+      "step": 5265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.36159558259411e-07,
+      "loss": 1.0737,
+      "step": 5266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.348952799223224e-07,
+      "loss": 1.2442,
+      "step": 5267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.336321739494072e-07,
+      "loss": 0.8242,
+      "step": 5268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.323702406799082e-07,
+      "loss": 0.879,
+      "step": 5269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.31109480452749e-07,
+      "loss": 0.7445,
+      "step": 5270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.298498936065412e-07,
+      "loss": 0.9641,
+      "step": 5271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.285914804795784e-07,
+      "loss": 1.2881,
+      "step": 5272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.273342414098426e-07,
+      "loss": 0.9967,
+      "step": 5273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.260781767349983e-07,
+      "loss": 1.3643,
+      "step": 5274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.248232867923931e-07,
+      "loss": 1.0921,
+      "step": 5275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235695719190632e-07,
+      "loss": 1.1352,
+      "step": 5276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.223170324517259e-07,
+      "loss": 1.1546,
+      "step": 5277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.210656687267835e-07,
+      "loss": 1.296,
+      "step": 5278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.198154810803214e-07,
+      "loss": 1.1543,
+      "step": 5279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.185664698481137e-07,
+      "loss": 0.994,
+      "step": 5280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.173186353656136e-07,
+      "loss": 1.0902,
+      "step": 5281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.160719779679597e-07,
+      "loss": 0.7058,
+      "step": 5282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.148264979899743e-07,
+      "loss": 0.9568,
+      "step": 5283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.135821957661658e-07,
+      "loss": 0.8735,
+      "step": 5284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.12339071630722e-07,
+      "loss": 1.1074,
+      "step": 5285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.110971259175208e-07,
+      "loss": 0.9559,
+      "step": 5286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.098563589601148e-07,
+      "loss": 0.9521,
+      "step": 5287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.086167710917479e-07,
+      "loss": 0.8907,
+      "step": 5288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.073783626453417e-07,
+      "loss": 1.2103,
+      "step": 5289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.061411339535062e-07,
+      "loss": 0.8811,
+      "step": 5290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.049050853485305e-07,
+      "loss": 1.1516,
+      "step": 5291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.036702171623876e-07,
+      "loss": 1.3159,
+      "step": 5292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.024365297267343e-07,
+      "loss": 1.1407,
+      "step": 5293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.012040233729105e-07,
+      "loss": 0.8653,
+      "step": 5294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.999726984319376e-07,
+      "loss": 1.0663,
+      "step": 5295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.987425552345222e-07,
+      "loss": 0.9953,
+      "step": 5296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.975135941110488e-07,
+      "loss": 1.1417,
+      "step": 5297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.962858153915896e-07,
+      "loss": 0.8713,
+      "step": 5298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.950592194058958e-07,
+      "loss": 1.2274,
+      "step": 5299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.938338064834037e-07,
+      "loss": 0.8805,
+      "step": 5300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.926095769532286e-07,
+      "loss": 0.9693,
+      "step": 5301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.913865311441714e-07,
+      "loss": 1.1713,
+      "step": 5302
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.901646693847108e-07,
+      "loss": 1.1062,
+      "step": 5303
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.889439920030127e-07,
+      "loss": 0.9696,
+      "step": 5304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.87724499326921e-07,
+      "loss": 1.1432,
+      "step": 5305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.865061916839615e-07,
+      "loss": 1.2378,
+      "step": 5306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.852890694013458e-07,
+      "loss": 1.0273,
+      "step": 5307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.840731328059629e-07,
+      "loss": 1.2192,
+      "step": 5308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.828583822243839e-07,
+      "loss": 0.9758,
+      "step": 5309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.816448179828616e-07,
+      "loss": 1.0811,
+      "step": 5310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.804324404073331e-07,
+      "loss": 0.9231,
+      "step": 5311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.792212498234134e-07,
+      "loss": 1.1003,
+      "step": 5312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.780112465563992e-07,
+      "loss": 1.0321,
+      "step": 5313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.768024309312681e-07,
+      "loss": 0.883,
+      "step": 5314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.755948032726815e-07,
+      "loss": 1.1697,
+      "step": 5315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.74388363904978e-07,
+      "loss": 0.9118,
+      "step": 5316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.731831131521815e-07,
+      "loss": 0.7959,
+      "step": 5317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.719790513379891e-07,
+      "loss": 1.3775,
+      "step": 5318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.707761787857879e-07,
+      "loss": 1.1364,
+      "step": 5319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.695744958186383e-07,
+      "loss": 0.8526,
+      "step": 5320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.68374002759286e-07,
+      "loss": 1.1335,
+      "step": 5321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.671746999301542e-07,
+      "loss": 0.9022,
+      "step": 5322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.659765876533474e-07,
+      "loss": 0.7956,
+      "step": 5323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.647796662506493e-07,
+      "loss": 1.1433,
+      "step": 5324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.635839360435263e-07,
+      "loss": 1.2639,
+      "step": 5325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.623893973531225e-07,
+      "loss": 1.0542,
+      "step": 5326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.611960505002645e-07,
+      "loss": 1.0572,
+      "step": 5327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.600038958054538e-07,
+      "loss": 1.0203,
+      "step": 5328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.588129335888781e-07,
+      "loss": 1.0891,
+      "step": 5329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.576231641703994e-07,
+      "loss": 1.2174,
+      "step": 5330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.564345878695643e-07,
+      "loss": 1.2326,
+      "step": 5331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.552472050055946e-07,
+      "loss": 1.113,
+      "step": 5332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.540610158973936e-07,
+      "loss": 0.9498,
+      "step": 5333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.528760208635436e-07,
+      "loss": 0.8514,
+      "step": 5334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.516922202223063e-07,
+      "loss": 1.0898,
+      "step": 5335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.505096142916233e-07,
+      "loss": 0.7903,
+      "step": 5336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.493282033891129e-07,
+      "loss": 1.4462,
+      "step": 5337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.481479878320784e-07,
+      "loss": 1.2298,
+      "step": 5338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.469689679374924e-07,
+      "loss": 0.8901,
+      "step": 5339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.457911440220154e-07,
+      "loss": 0.9874,
+      "step": 5340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.446145164019812e-07,
+      "loss": 0.9907,
+      "step": 5341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.434390853934063e-07,
+      "loss": 0.928,
+      "step": 5342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.422648513119827e-07,
+      "loss": 1.1619,
+      "step": 5343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.410918144730815e-07,
+      "loss": 0.9916,
+      "step": 5344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.399199751917522e-07,
+      "loss": 1.1758,
+      "step": 5345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.387493337827254e-07,
+      "loss": 0.775,
+      "step": 5346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.375798905604052e-07,
+      "loss": 1.135,
+      "step": 5347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.364116458388802e-07,
+      "loss": 1.0575,
+      "step": 5348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.352445999319084e-07,
+      "loss": 1.0312,
+      "step": 5349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.340787531529346e-07,
+      "loss": 1.1036,
+      "step": 5350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.329141058150755e-07,
+      "loss": 1.1386,
+      "step": 5351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3175065823113e-07,
+      "loss": 1.2001,
+      "step": 5352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.305884107135711e-07,
+      "loss": 0.8391,
+      "step": 5353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.294273635745517e-07,
+      "loss": 1.1605,
+      "step": 5354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.282675171259005e-07,
+      "loss": 0.8995,
+      "step": 5355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.271088716791273e-07,
+      "loss": 1.2205,
+      "step": 5356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.259514275454153e-07,
+      "loss": 0.8927,
+      "step": 5357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.24795185035627e-07,
+      "loss": 1.1365,
+      "step": 5358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.236401444603007e-07,
+      "loss": 1.0881,
+      "step": 5359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.224863061296553e-07,
+      "loss": 1.1741,
+      "step": 5360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.213336703535837e-07,
+      "loss": 1.076,
+      "step": 5361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.201822374416549e-07,
+      "loss": 1.2485,
+      "step": 5362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.190320077031197e-07,
+      "loss": 0.7152,
+      "step": 5363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.178829814469006e-07,
+      "loss": 1.1966,
+      "step": 5364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.167351589815994e-07,
+      "loss": 1.1223,
+      "step": 5365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.155885406154937e-07,
+      "loss": 1.2622,
+      "step": 5366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.144431266565397e-07,
+      "loss": 1.2307,
+      "step": 5367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.132989174123659e-07,
+      "loss": 0.7868,
+      "step": 5368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121559131902843e-07,
+      "loss": 0.8919,
+      "step": 5369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.110141142972735e-07,
+      "loss": 0.9792,
+      "step": 5370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.09873521039998e-07,
+      "loss": 0.9898,
+      "step": 5371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.087341337247914e-07,
+      "loss": 1.032,
+      "step": 5372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.075959526576685e-07,
+      "loss": 1.097,
+      "step": 5373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.064589781443163e-07,
+      "loss": 1.0999,
+      "step": 5374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.053232104901006e-07,
+      "loss": 1.0171,
+      "step": 5375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041886500000603e-07,
+      "loss": 0.991,
+      "step": 5376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.030552969789126e-07,
+      "loss": 0.9985,
+      "step": 5377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.019231517310491e-07,
+      "loss": 0.7369,
+      "step": 5378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.007922145605387e-07,
+      "loss": 0.805,
+      "step": 5379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.996624857711219e-07,
+      "loss": 1.0185,
+      "step": 5380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.985339656662186e-07,
+      "loss": 1.1263,
+      "step": 5381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.97406654548922e-07,
+      "loss": 0.9362,
+      "step": 5382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.962805527220027e-07,
+      "loss": 0.9544,
+      "step": 5383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.951556604879049e-07,
+      "loss": 1.0055,
+      "step": 5384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.940319781487468e-07,
+      "loss": 0.9933,
+      "step": 5385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.929095060063227e-07,
+      "loss": 1.3398,
+      "step": 5386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.917882443621042e-07,
+      "loss": 1.2779,
+      "step": 5387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.906681935172342e-07,
+      "loss": 0.8698,
+      "step": 5388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.895493537725326e-07,
+      "loss": 1.0167,
+      "step": 5389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.88431725428492e-07,
+      "loss": 1.0097,
+      "step": 5390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.873153087852833e-07,
+      "loss": 1.3072,
+      "step": 5391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.862001041427488e-07,
+      "loss": 0.8921,
+      "step": 5392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.850861118004047e-07,
+      "loss": 0.9932,
+      "step": 5393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.839733320574457e-07,
+      "loss": 0.9329,
+      "step": 5394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.828617652127355e-07,
+      "loss": 0.8585,
+      "step": 5395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.817514115648164e-07,
+      "loss": 1.0037,
+      "step": 5396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.806422714119019e-07,
+      "loss": 0.9274,
+      "step": 5397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795343450518825e-07,
+      "loss": 1.0057,
+      "step": 5398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.784276327823195e-07,
+      "loss": 1.2038,
+      "step": 5399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773221349004531e-07,
+      "loss": 0.8599,
+      "step": 5400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7621785170318843e-07,
+      "loss": 1.1331,
+      "step": 5401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7511478348711447e-07,
+      "loss": 1.1257,
+      "step": 5402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7401293054848696e-07,
+      "loss": 0.9612,
+      "step": 5403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729122931832392e-07,
+      "loss": 1.1775,
+      "step": 5404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.718128716869752e-07,
+      "loss": 1.1147,
+      "step": 5405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.707146663549744e-07,
+      "loss": 1.2355,
+      "step": 5406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.696176774821881e-07,
+      "loss": 0.8758,
+      "step": 5407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.685219053632423e-07,
+      "loss": 0.8668,
+      "step": 5408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.674273502924348e-07,
+      "loss": 1.0546,
+      "step": 5409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.663340125637389e-07,
+      "loss": 0.8856,
+      "step": 5410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.652418924707969e-07,
+      "loss": 1.0552,
+      "step": 5411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6415099030692914e-07,
+      "loss": 0.7868,
+      "step": 5412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.630613063651235e-07,
+      "loss": 0.9404,
+      "step": 5413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.619728409380453e-07,
+      "loss": 0.901,
+      "step": 5414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.608855943180307e-07,
+      "loss": 0.9143,
+      "step": 5415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.597995667970878e-07,
+      "loss": 1.188,
+      "step": 5416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5871475866689827e-07,
+      "loss": 0.9264,
+      "step": 5417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5763117021881467e-07,
+      "loss": 0.9299,
+      "step": 5418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5654880174386517e-07,
+      "loss": 0.9271,
+      "step": 5419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5546765353274846e-07,
+      "loss": 1.1145,
+      "step": 5420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.543877258758339e-07,
+      "loss": 1.2826,
+      "step": 5421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5330901906316506e-07,
+      "loss": 0.9331,
+      "step": 5422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.522315333844579e-07,
+      "loss": 1.2913,
+      "step": 5423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.511552691290988e-07,
+      "loss": 0.8625,
+      "step": 5424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.500802265861481e-07,
+      "loss": 1.0358,
+      "step": 5425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.490064060443361e-07,
+      "loss": 1.2399,
+      "step": 5426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4793380779206586e-07,
+      "loss": 1.1058,
+      "step": 5427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.468624321174109e-07,
+      "loss": 0.7023,
+      "step": 5428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.457922793081193e-07,
+      "loss": 0.9851,
+      "step": 5429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4472334965160736e-07,
+      "loss": 0.9571,
+      "step": 5430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4365564343496647e-07,
+      "loss": 1.0855,
+      "step": 5431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4258916094495394e-07,
+      "loss": 1.0369,
+      "step": 5432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4152390246800457e-07,
+      "loss": 0.6937,
+      "step": 5433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4045986829022e-07,
+      "loss": 1.2277,
+      "step": 5434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.393970586973767e-07,
+      "loss": 0.9127,
+      "step": 5435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.38335473974919e-07,
+      "loss": 1.3006,
+      "step": 5436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3727511440796343e-07,
+      "loss": 1.2573,
+      "step": 5437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.362159802812971e-07,
+      "loss": 0.9809,
+      "step": 5438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.351580718793802e-07,
+      "loss": 0.8986,
+      "step": 5439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.341013894863405e-07,
+      "loss": 0.7118,
+      "step": 5440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3304593338597824e-07,
+      "loss": 1.0685,
+      "step": 5441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3199170386176325e-07,
+      "loss": 1.2498,
+      "step": 5442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3093870119683856e-07,
+      "loss": 0.8255,
+      "step": 5443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2988692567401515e-07,
+      "loss": 1.051,
+      "step": 5444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2883637757577367e-07,
+      "loss": 1.2907,
+      "step": 5445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2778705718426907e-07,
+      "loss": 1.1127,
+      "step": 5446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.267389647813225e-07,
+      "loss": 1.0858,
+      "step": 5447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2569210064842716e-07,
+      "loss": 0.8874,
+      "step": 5448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.246464650667448e-07,
+      "loss": 1.0523,
+      "step": 5449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.236020583171108e-07,
+      "loss": 0.8876,
+      "step": 5450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.225588806800268e-07,
+      "loss": 1.2194,
+      "step": 5451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.215169324356666e-07,
+      "loss": 1.058,
+      "step": 5452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2047621386387074e-07,
+      "loss": 0.9011,
+      "step": 5453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.194367252441545e-07,
+      "loss": 1.0648,
+      "step": 5454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1839846685569786e-07,
+      "loss": 1.1814,
+      "step": 5455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1736143897735394e-07,
+      "loss": 0.8735,
+      "step": 5456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1632564188764377e-07,
+      "loss": 1.0637,
+      "step": 5457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.152910758647577e-07,
+      "loss": 1.0523,
+      "step": 5458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1425774118655505e-07,
+      "loss": 0.763,
+      "step": 5459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1322563813056606e-07,
+      "loss": 0.9119,
+      "step": 5460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1219476697398896e-07,
+      "loss": 1.2727,
+      "step": 5461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.111651279936929e-07,
+      "loss": 1.1216,
+      "step": 5462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1013672146621165e-07,
+      "loss": 1.2449,
+      "step": 5463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.091095476677531e-07,
+      "loss": 1.1194,
+      "step": 5464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0808360687419056e-07,
+      "loss": 1.0741,
+      "step": 5465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.070588993610697e-07,
+      "loss": 1.0712,
+      "step": 5466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0603542540360085e-07,
+      "loss": 0.9047,
+      "step": 5467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.050131852766659e-07,
+      "loss": 0.8856,
+      "step": 5468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0399217925481294e-07,
+      "loss": 1.1188,
+      "step": 5469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.029724076122621e-07,
+      "loss": 1.0606,
+      "step": 5470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0195387062289927e-07,
+      "loss": 1.0553,
+      "step": 5471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009365685602795e-07,
+      "loss": 0.9545,
+      "step": 5472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9992050169762487e-07,
+      "loss": 1.0031,
+      "step": 5473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.989056703078292e-07,
+      "loss": 0.9807,
+      "step": 5474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.978920746634507e-07,
+      "loss": 1.0514,
+      "step": 5475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.968797150367171e-07,
+      "loss": 0.9776,
+      "step": 5476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9586859169952696e-07,
+      "loss": 1.2016,
+      "step": 5477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.948587049234398e-07,
+      "loss": 0.794,
+      "step": 5478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.938500549796909e-07,
+      "loss": 0.9762,
+      "step": 5479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.928426421391773e-07,
+      "loss": 1.1568,
+      "step": 5480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9183646667246887e-07,
+      "loss": 1.2583,
+      "step": 5481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9083152884979935e-07,
+      "loss": 1.0073,
+      "step": 5482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8982782894107176e-07,
+      "loss": 0.8012,
+      "step": 5483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8882536721585486e-07,
+      "loss": 0.9237,
+      "step": 5484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8782414394338774e-07,
+      "loss": 1.1886,
+      "step": 5485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.868241593925742e-07,
+      "loss": 1.2073,
+      "step": 5486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.858254138319878e-07,
+      "loss": 1.1064,
+      "step": 5487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.848279075298678e-07,
+      "loss": 0.8599,
+      "step": 5488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.838316407541204e-07,
+      "loss": 0.9605,
+      "step": 5489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.828366137723183e-07,
+      "loss": 0.7928,
+      "step": 5490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8184282685170435e-07,
+      "loss": 1.2692,
+      "step": 5491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.80850280259184e-07,
+      "loss": 0.9318,
+      "step": 5492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7985897426133566e-07,
+      "loss": 1.0411,
+      "step": 5493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7886890912439633e-07,
+      "loss": 1.0582,
+      "step": 5494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7788008511427664e-07,
+      "loss": 0.7721,
+      "step": 5495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.768925024965503e-07,
+      "loss": 0.7621,
+      "step": 5496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.759061615364595e-07,
+      "loss": 1.0788,
+      "step": 5497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.749210624989125e-07,
+      "loss": 0.9407,
+      "step": 5498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.73937205648483e-07,
+      "loss": 1.0565,
+      "step": 5499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.729545912494115e-07,
+      "loss": 0.9812,
+      "step": 5500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.719732195656051e-07,
+      "loss": 1.0161,
+      "step": 5501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7099309086063794e-07,
+      "loss": 0.8658,
+      "step": 5502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.700142053977485e-07,
+      "loss": 0.9462,
+      "step": 5503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6903656343984293e-07,
+      "loss": 0.8395,
+      "step": 5504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6806016524949196e-07,
+      "loss": 0.9499,
+      "step": 5505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.670850110889346e-07,
+      "loss": 1.4655,
+      "step": 5506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.661111012200724e-07,
+      "loss": 1.1385,
+      "step": 5507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.651384359044774e-07,
+      "loss": 1.0361,
+      "step": 5508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6416701540338105e-07,
+      "loss": 1.226,
+      "step": 5509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.631968399776864e-07,
+      "loss": 1.0642,
+      "step": 5510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6222790988795773e-07,
+      "loss": 0.8145,
+      "step": 5511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6126022539442975e-07,
+      "loss": 0.9992,
+      "step": 5512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6029378675699724e-07,
+      "loss": 1.0604,
+      "step": 5513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.593285942352237e-07,
+      "loss": 1.2804,
+      "step": 5514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.583646480883363e-07,
+      "loss": 1.0517,
+      "step": 5515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5740194857523e-07,
+      "loss": 0.986,
+      "step": 5516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5644049595446095e-07,
+      "loss": 1.2127,
+      "step": 5517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.554802904842547e-07,
+      "loss": 0.9046,
+      "step": 5518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5452133242249886e-07,
+      "loss": 0.937,
+      "step": 5519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5356362202674687e-07,
+      "loss": 0.9354,
+      "step": 5520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.526071595542163e-07,
+      "loss": 0.7525,
+      "step": 5521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.516519452617922e-07,
+      "loss": 0.8254,
+      "step": 5522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.506979794060217e-07,
+      "loss": 0.853,
+      "step": 5523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4974526224311744e-07,
+      "loss": 0.8784,
+      "step": 5524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.487937940289565e-07,
+      "loss": 0.9919,
+      "step": 5525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.478435750190817e-07,
+      "loss": 0.998,
+      "step": 5526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4689460546869947e-07,
+      "loss": 0.6444,
+      "step": 5527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.459468856326792e-07,
+      "loss": 0.7914,
+      "step": 5528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4500041576555733e-07,
+      "loss": 1.0703,
+      "step": 5529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4405519612153326e-07,
+      "loss": 1.1742,
+      "step": 5530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.431112269544706e-07,
+      "loss": 0.9312,
+      "step": 5531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4216850851789663e-07,
+      "loss": 0.8413,
+      "step": 5532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.412270410650043e-07,
+      "loss": 0.8856,
+      "step": 5533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.402868248486485e-07,
+      "loss": 1.0358,
+      "step": 5534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.393478601213501e-07,
+      "loss": 1.0608,
+      "step": 5535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3841014713529184e-07,
+      "loss": 0.8592,
+      "step": 5536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3747368614232223e-07,
+      "loss": 0.9221,
+      "step": 5537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3653847739395174e-07,
+      "loss": 0.9436,
+      "step": 5538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.356045211413572e-07,
+      "loss": 0.827,
+      "step": 5539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.346718176353747e-07,
+      "loss": 0.8823,
+      "step": 5540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3374036712650827e-07,
+      "loss": 1.1044,
+      "step": 5541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3281016986492165e-07,
+      "loss": 1.0274,
+      "step": 5542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.318812261004467e-07,
+      "loss": 0.9246,
+      "step": 5543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3095353608257385e-07,
+      "loss": 0.7612,
+      "step": 5544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3002710006045937e-07,
+      "loss": 0.9535,
+      "step": 5545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2910191828292083e-07,
+      "loss": 0.8805,
+      "step": 5546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.281779909984428e-07,
+      "loss": 1.1699,
+      "step": 5547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2725531845516744e-07,
+      "loss": 1.2691,
+      "step": 5548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2633390090090535e-07,
+      "loss": 1.0605,
+      "step": 5549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.254137385831263e-07,
+      "loss": 0.8363,
+      "step": 5550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2449483174896434e-07,
+      "loss": 1.2129,
+      "step": 5551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2357718064521594e-07,
+      "loss": 0.7987,
+      "step": 5552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.226607855183411e-07,
+      "loss": 1.0295,
+      "step": 5553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.217456466144614e-07,
+      "loss": 0.9297,
+      "step": 5554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2083176417936234e-07,
+      "loss": 1.1714,
+      "step": 5555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.199191384584893e-07,
+      "loss": 1.0373,
+      "step": 5556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.190077696969546e-07,
+      "loss": 0.9326,
+      "step": 5557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.180976581395295e-07,
+      "loss": 1.0278,
+      "step": 5558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1718880403064703e-07,
+      "loss": 1.4643,
+      "step": 5559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1628120761440616e-07,
+      "loss": 1.0692,
+      "step": 5560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.153748691345637e-07,
+      "loss": 1.1364,
+      "step": 5561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.144697888345427e-07,
+      "loss": 1.1826,
+      "step": 5562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.135659669574248e-07,
+      "loss": 1.1767,
+      "step": 5563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1266340374595693e-07,
+      "loss": 1.1336,
+      "step": 5564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.117620994425458e-07,
+      "loss": 0.9851,
+      "step": 5565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.108620542892593e-07,
+      "loss": 1.1787,
+      "step": 5566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0996326852782863e-07,
+      "loss": 0.8542,
+      "step": 5567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0906574239964795e-07,
+      "loss": 0.9154,
+      "step": 5568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0816947614576933e-07,
+      "loss": 0.8444,
+      "step": 5569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.072744700069119e-07,
+      "loss": 1.05,
+      "step": 5570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0638072422344957e-07,
+      "loss": 0.9329,
+      "step": 5571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.054882390354241e-07,
+      "loss": 0.7599,
+      "step": 5572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.045970146825339e-07,
+      "loss": 1.0898,
+      "step": 5573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0370705140414293e-07,
+      "loss": 1.1765,
+      "step": 5574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0281834943927224e-07,
+      "loss": 1.3525,
+      "step": 5575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.019309090266087e-07,
+      "loss": 1.1028,
+      "step": 5576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0104473040449556e-07,
+      "loss": 1.161,
+      "step": 5577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0015981381094073e-07,
+      "loss": 0.9819,
+      "step": 5578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9927615948361077e-07,
+      "loss": 1.1309,
+      "step": 5579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9839376765983583e-07,
+      "loss": 0.98,
+      "step": 5580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.975126385766053e-07,
+      "loss": 1.245,
+      "step": 5581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9663277247056923e-07,
+      "loss": 1.2377,
+      "step": 5582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.957541695780386e-07,
+      "loss": 0.9223,
+      "step": 5583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9487683013498523e-07,
+      "loss": 0.9028,
+      "step": 5584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.940007543770429e-07,
+      "loss": 0.7664,
+      "step": 5585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.93125942539505e-07,
+      "loss": 1.021,
+      "step": 5586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.922523948573242e-07,
+      "loss": 1.0281,
+      "step": 5587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.913801115651144e-07,
+      "loss": 1.1277,
+      "step": 5588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.905090928971527e-07,
+      "loss": 1.1795,
+      "step": 5589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.896393390873714e-07,
+      "loss": 0.9242,
+      "step": 5590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.887708503693687e-07,
+      "loss": 0.7841,
+      "step": 5591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8790362697639685e-07,
+      "loss": 0.8908,
+      "step": 5592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8703766914137453e-07,
+      "loss": 1.1032,
+      "step": 5593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8617297709687577e-07,
+      "loss": 1.3303,
+      "step": 5594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8530955107513803e-07,
+      "loss": 1.0459,
+      "step": 5595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8444739130805587e-07,
+      "loss": 1.083,
+      "step": 5596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8358649802718575e-07,
+      "loss": 1.0474,
+      "step": 5597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.827268714637421e-07,
+      "loss": 1.1001,
+      "step": 5598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.818685118486025e-07,
+      "loss": 1.1166,
+      "step": 5599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.810114194122998e-07,
+      "loss": 1.3032,
+      "step": 5600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8015559438503094e-07,
+      "loss": 1.3772,
+      "step": 5601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.793010369966487e-07,
+      "loss": 1.0234,
+      "step": 5602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7844774747666783e-07,
+      "loss": 0.9056,
+      "step": 5603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7759572605426057e-07,
+      "loss": 1.1106,
+      "step": 5604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.767449729582622e-07,
+      "loss": 1.2242,
+      "step": 5605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7589548841716274e-07,
+      "loss": 0.9792,
+      "step": 5606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.750472726591152e-07,
+      "loss": 0.9937,
+      "step": 5607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7420032591192856e-07,
+      "loss": 1.0768,
+      "step": 5608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.733546484030747e-07,
+      "loss": 0.925,
+      "step": 5609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7251024035968134e-07,
+      "loss": 0.7813,
+      "step": 5610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.716671020085365e-07,
+      "loss": 1.0244,
+      "step": 5611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7082523357608856e-07,
+      "loss": 0.7991,
+      "step": 5612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.699846352884422e-07,
+      "loss": 1.081,
+      "step": 5613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6914530737136346e-07,
+      "loss": 0.9316,
+      "step": 5614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6830725005027424e-07,
+      "loss": 0.8653,
+      "step": 5615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.674704635502584e-07,
+      "loss": 0.9095,
+      "step": 5616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.666349480960573e-07,
+      "loss": 0.7599,
+      "step": 5617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.658007039120697e-07,
+      "loss": 1.0619,
+      "step": 5618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.649677312223542e-07,
+      "loss": 0.9831,
+      "step": 5619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.64136030250628e-07,
+      "loss": 0.9187,
+      "step": 5620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.633056012202656e-07,
+      "loss": 0.9603,
+      "step": 5621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6247644435430263e-07,
+      "loss": 0.8953,
+      "step": 5622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.616485598754276e-07,
+      "loss": 1.0613,
+      "step": 5623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6082194800599424e-07,
+      "loss": 0.9117,
+      "step": 5624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.599966089680084e-07,
+      "loss": 0.8051,
+      "step": 5625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.591725429831382e-07,
+      "loss": 1.0357,
+      "step": 5626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5834975027270837e-07,
+      "loss": 1.0924,
+      "step": 5627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.57528231057701e-07,
+      "loss": 1.0955,
+      "step": 5628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5670798555875674e-07,
+      "loss": 0.9963,
+      "step": 5629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.558890139961745e-07,
+      "loss": 1.06,
+      "step": 5630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5507131658990993e-07,
+      "loss": 0.8549,
+      "step": 5631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5425489355957956e-07,
+      "loss": 0.9136,
+      "step": 5632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534397451244525e-07,
+      "loss": 0.8674,
+      "step": 5633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526258715034602e-07,
+      "loss": 1.0485,
+      "step": 5634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5181327291518843e-07,
+      "loss": 0.9067,
+      "step": 5635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.510019495778837e-07,
+      "loss": 1.1826,
+      "step": 5636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5019190170944786e-07,
+      "loss": 1.1326,
+      "step": 5637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4938312952744016e-07,
+      "loss": 1.114,
+      "step": 5638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.485756332490774e-07,
+      "loss": 1.0238,
+      "step": 5639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.477694130912356e-07,
+      "loss": 1.1401,
+      "step": 5640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.469644692704448e-07,
+      "loss": 0.8339,
+      "step": 5641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.461608020028944e-07,
+      "loss": 1.2127,
+      "step": 5642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.453584115044316e-07,
+      "loss": 0.9502,
+      "step": 5643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.445572979905575e-07,
+      "loss": 1.2369,
+      "step": 5644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.437574616764338e-07,
+      "loss": 1.0155,
+      "step": 5645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4295890277687695e-07,
+      "loss": 0.9989,
+      "step": 5646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.421616215063621e-07,
+      "loss": 0.7823,
+      "step": 5647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4136561807901916e-07,
+      "loss": 1.2516,
+      "step": 5648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4057089270863644e-07,
+      "loss": 1.2419,
+      "step": 5649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.397774456086577e-07,
+      "loss": 1.1146,
+      "step": 5650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3898527699218466e-07,
+      "loss": 1.1562,
+      "step": 5651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3819438707197495e-07,
+      "loss": 1.2055,
+      "step": 5652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3740477606044364e-07,
+      "loss": 1.0622,
+      "step": 5653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3661644416966057e-07,
+      "loss": 1.115,
+      "step": 5654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3582939161135355e-07,
+      "loss": 1.1948,
+      "step": 5655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3504361859690628e-07,
+      "loss": 1.0308,
+      "step": 5656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.342591253373594e-07,
+      "loss": 1.0569,
+      "step": 5657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3347591204340881e-07,
+      "loss": 1.0303,
+      "step": 5658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.326939789254068e-07,
+      "loss": 0.9238,
+      "step": 5659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3191332619336204e-07,
+      "loss": 1.2113,
+      "step": 5660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3113395405694073e-07,
+      "loss": 1.1046,
+      "step": 5661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3035586272546207e-07,
+      "loss": 0.929,
+      "step": 5662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2957905240790557e-07,
+      "loss": 1.0643,
+      "step": 5663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2880352331290102e-07,
+      "loss": 1.0412,
+      "step": 5664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2802927564873954e-07,
+      "loss": 1.0615,
+      "step": 5665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2725630962336542e-07,
+      "loss": 1.1547,
+      "step": 5666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2648462544437754e-07,
+      "loss": 0.6206,
+      "step": 5667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2571422331903458e-07,
+      "loss": 1.0298,
+      "step": 5668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2494510345424657e-07,
+      "loss": 0.8031,
+      "step": 5669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2417726605658164e-07,
+      "loss": 0.8015,
+      "step": 5670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2341071133226199e-07,
+      "loss": 0.8505,
+      "step": 5671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.226454394871669e-07,
+      "loss": 1.0584,
+      "step": 5672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.218814507268302e-07,
+      "loss": 0.9998,
+      "step": 5673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2111874525644228e-07,
+      "loss": 0.9731,
+      "step": 5674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2035732328084592e-07,
+      "loss": 1.1687,
+      "step": 5675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1959718500454196e-07,
+      "loss": 0.7673,
+      "step": 5676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1883833063168546e-07,
+      "loss": 0.9199,
+      "step": 5677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1808076036608783e-07,
+      "loss": 1.0324,
+      "step": 5678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1732447441121353e-07,
+      "loss": 1.2217,
+      "step": 5679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.165694729701834e-07,
+      "loss": 1.0849,
+      "step": 5680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1581575624577245e-07,
+      "loss": 1.0139,
+      "step": 5681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1506332444041212e-07,
+      "loss": 0.9627,
+      "step": 5682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1431217775618684e-07,
+      "loss": 0.9404,
+      "step": 5683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1356231639483917e-07,
+      "loss": 1.0527,
+      "step": 5684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.128137405577607e-07,
+      "loss": 0.9835,
+      "step": 5685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1206645044600404e-07,
+      "loss": 1.1853,
+      "step": 5686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.113204462602725e-07,
+      "loss": 0.9708,
+      "step": 5687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1057572820092576e-07,
+      "loss": 1.0106,
+      "step": 5688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0983229646797666e-07,
+      "loss": 0.8576,
+      "step": 5689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0909015126109488e-07,
+      "loss": 1.1967,
+      "step": 5690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0834929277960103e-07,
+      "loss": 1.0275,
+      "step": 5691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0760972122247425e-07,
+      "loss": 0.8147,
+      "step": 5692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.068714367883451e-07,
+      "loss": 1.2051,
+      "step": 5693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.061344396754994e-07,
+      "loss": 1.1077,
+      "step": 5694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.053987300818766e-07,
+      "loss": 0.8684,
+      "step": 5695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.04664308205072e-07,
+      "loss": 0.8741,
+      "step": 5696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0393117424233387e-07,
+      "loss": 1.0304,
+      "step": 5697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0319932839056365e-07,
+      "loss": 1.1195,
+      "step": 5698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.024687708463191e-07,
+      "loss": 1.1415,
+      "step": 5699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0173950180581047e-07,
+      "loss": 1.1882,
+      "step": 5700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0101152146490222e-07,
+      "loss": 0.8744,
+      "step": 5701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.002848300191118e-07,
+      "loss": 0.835,
+      "step": 5702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9955942766361258e-07,
+      "loss": 1.0529,
+      "step": 5703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.988353145932298e-07,
+      "loss": 1.1427,
+      "step": 5704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.98112491002444e-07,
+      "loss": 1.094,
+      "step": 5705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9739095708538714e-07,
+      "loss": 1.4376,
+      "step": 5706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.966707130358475e-07,
+      "loss": 0.7689,
+      "step": 5707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9595175904726481e-07,
+      "loss": 1.0133,
+      "step": 5708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9523409531273352e-07,
+      "loss": 1.0477,
+      "step": 5709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9451772202500163e-07,
+      "loss": 1.0182,
+      "step": 5710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9380263937646972e-07,
+      "loss": 0.7773,
+      "step": 5711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9308884755919132e-07,
+      "loss": 0.9792,
+      "step": 5712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9237634676487537e-07,
+      "loss": 1.1094,
+      "step": 5713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9166513718488155e-07,
+      "loss": 1.4043,
+      "step": 5714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9095521901022597e-07,
+      "loss": 0.9015,
+      "step": 5715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.902465924315733e-07,
+      "loss": 1.1344,
+      "step": 5716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8953925763924575e-07,
+      "loss": 1.2215,
+      "step": 5717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8883321482321583e-07,
+      "loss": 1.0915,
+      "step": 5718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.881284641731107e-07,
+      "loss": 0.8942,
+      "step": 5719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8742500587820955e-07,
+      "loss": 0.9874,
+      "step": 5720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8672284012744513e-07,
+      "loss": 1.3238,
+      "step": 5721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.86021967109401e-07,
+      "loss": 0.8887,
+      "step": 5722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8532238701231665e-07,
+      "loss": 0.9009,
+      "step": 5723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8462410002408228e-07,
+      "loss": 1.0785,
+      "step": 5724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8392710633224064e-07,
+      "loss": 0.8082,
+      "step": 5725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8323140612399038e-07,
+      "loss": 1.2407,
+      "step": 5726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8253699958617643e-07,
+      "loss": 1.004,
+      "step": 5727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8184388690530242e-07,
+      "loss": 1.1387,
+      "step": 5728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8115206826752163e-07,
+      "loss": 1.4325,
+      "step": 5729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.804615438586399e-07,
+      "loss": 1.0966,
+      "step": 5730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7977231386411665e-07,
+      "loss": 0.7825,
+      "step": 5731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7908437846906158e-07,
+      "loss": 1.3491,
+      "step": 5732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7839773785823856e-07,
+      "loss": 1.0573,
+      "step": 5733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7771239221606285e-07,
+      "loss": 0.9716,
+      "step": 5734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770283417266022e-07,
+      "loss": 1.2237,
+      "step": 5735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7634558657357748e-07,
+      "loss": 1.4557,
+      "step": 5736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7566412694035862e-07,
+      "loss": 1.0213,
+      "step": 5737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7498396300997146e-07,
+      "loss": 1.1872,
+      "step": 5738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7430509496508985e-07,
+      "loss": 1.0113,
+      "step": 5739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.736275229880441e-07,
+      "loss": 1.1947,
+      "step": 5740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7295124726081303e-07,
+      "loss": 1.1322,
+      "step": 5741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7227626796502807e-07,
+      "loss": 0.9474,
+      "step": 5742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7160258528197304e-07,
+      "loss": 0.9599,
+      "step": 5743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7093019939258327e-07,
+      "loss": 1.0088,
+      "step": 5744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.702591104774448e-07,
+      "loss": 1.1883,
+      "step": 5745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6958931871679908e-07,
+      "loss": 1.4104,
+      "step": 5746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.689208242905327e-07,
+      "loss": 0.8679,
+      "step": 5747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6825362737818985e-07,
+      "loss": 1.1978,
+      "step": 5748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6758772815896384e-07,
+      "loss": 1.0486,
+      "step": 5749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6692312681169775e-07,
+      "loss": 0.992,
+      "step": 5750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.662598235148899e-07,
+      "loss": 1.1239,
+      "step": 5751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6559781844668666e-07,
+      "loss": 0.9528,
+      "step": 5752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6493711178488746e-07,
+      "loss": 1.3101,
+      "step": 5753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6427770370694208e-07,
+      "loss": 1.3751,
+      "step": 5754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6361959438995268e-07,
+      "loss": 1.1037,
+      "step": 5755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6296278401067122e-07,
+      "loss": 1.1253,
+      "step": 5756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6230727274550274e-07,
+      "loss": 0.926,
+      "step": 5757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6165306077049969e-07,
+      "loss": 0.9189,
+      "step": 5758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6100014826137045e-07,
+      "loss": 0.8271,
+      "step": 5759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.603485353934703e-07,
+      "loss": 0.8104,
+      "step": 5760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.596982223418081e-07,
+      "loss": 1.0657,
+      "step": 5761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5904920928104196e-07,
+      "loss": 1.0875,
+      "step": 5762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5840149638548185e-07,
+      "loss": 0.9705,
+      "step": 5763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.577550838290881e-07,
+      "loss": 0.8524,
+      "step": 5764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5710997178547182e-07,
+      "loss": 0.9322,
+      "step": 5765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.564661604278944e-07,
+      "loss": 0.9578,
+      "step": 5766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.558236499292698e-07,
+      "loss": 0.8461,
+      "step": 5767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5518244046215936e-07,
+      "loss": 1.2919,
+      "step": 5768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5454253219877812e-07,
+      "loss": 1.1369,
+      "step": 5769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.539039253109892e-07,
+      "loss": 1.2484,
+      "step": 5770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5326661997030924e-07,
+      "loss": 0.8503,
+      "step": 5771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.526306163479019e-07,
+      "loss": 0.8331,
+      "step": 5772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5199591461458274e-07,
+      "loss": 0.9691,
+      "step": 5773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5136251494081822e-07,
+      "loss": 0.9736,
+      "step": 5774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5073041749672445e-07,
+      "loss": 0.9698,
+      "step": 5775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5009962245206845e-07,
+      "loss": 1.2164,
+      "step": 5776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4947012997626576e-07,
+      "loss": 1.1648,
+      "step": 5777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.488419402383834e-07,
+      "loss": 1.1413,
+      "step": 5778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.482150534071386e-07,
+      "loss": 0.9975,
+      "step": 5779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4758946965089894e-07,
+      "loss": 1.1345,
+      "step": 5780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4696518913768054e-07,
+      "loss": 0.7595,
+      "step": 5781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4634221203515097e-07,
+      "loss": 0.8599,
+      "step": 5782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4572053851062752e-07,
+      "loss": 1.2167,
+      "step": 5783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4510016873107657e-07,
+      "loss": 0.7672,
+      "step": 5784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.444811028631138e-07,
+      "loss": 0.9937,
+      "step": 5785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4386334107300727e-07,
+      "loss": 1.0498,
+      "step": 5786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.432468835266726e-07,
+      "loss": 1.2262,
+      "step": 5787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4263173038967627e-07,
+      "loss": 0.9221,
+      "step": 5788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4201788182723274e-07,
+      "loss": 0.844,
+      "step": 5789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4140533800420853e-07,
+      "loss": 0.9536,
+      "step": 5790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4079409908511755e-07,
+      "loss": 1.1465,
+      "step": 5791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.401841652341246e-07,
+      "loss": 1.105,
+      "step": 5792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3957553661504364e-07,
+      "loss": 1.0419,
+      "step": 5793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.389682133913378e-07,
+      "loss": 1.0706,
+      "step": 5794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3836219572611886e-07,
+      "loss": 0.8532,
+      "step": 5795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3775748378215047e-07,
+      "loss": 0.9329,
+      "step": 5796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3715407772184275e-07,
+      "loss": 1.1548,
+      "step": 5797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3655197770725826e-07,
+      "loss": 1.2132,
+      "step": 5798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3595118390010374e-07,
+      "loss": 1.0019,
+      "step": 5799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3535169646174073e-07,
+      "loss": 0.9566,
+      "step": 5800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.347535155531765e-07,
+      "loss": 0.9192,
+      "step": 5801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3415664133506812e-07,
+      "loss": 1.175,
+      "step": 5802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3356107396772287e-07,
+      "loss": 1.0159,
+      "step": 5803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3296681361109564e-07,
+      "loss": 0.978,
+      "step": 5804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3237386042478984e-07,
+      "loss": 1.0971,
+      "step": 5805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3178221456806028e-07,
+      "loss": 0.9463,
+      "step": 5806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3119187619980766e-07,
+      "loss": 1.0691,
+      "step": 5807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3060284547858403e-07,
+      "loss": 0.8072,
+      "step": 5808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3001512256258841e-07,
+      "loss": 0.9546,
+      "step": 5809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2942870760966952e-07,
+      "loss": 0.9397,
+      "step": 5810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2884360077732472e-07,
+      "loss": 1.0642,
+      "step": 5811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.282598022226994e-07,
+      "loss": 1.031,
+      "step": 5812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2767731210258983e-07,
+      "loss": 0.964,
+      "step": 5813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.270961305734364e-07,
+      "loss": 0.9691,
+      "step": 5814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2651625779133314e-07,
+      "loss": 0.9062,
+      "step": 5815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2593769391201827e-07,
+      "loss": 1.0206,
+      "step": 5816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.253604390908819e-07,
+      "loss": 1.1557,
+      "step": 5817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.247844934829606e-07,
+      "loss": 1.0817,
+      "step": 5818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.242098572429401e-07,
+      "loss": 1.1916,
+      "step": 5819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2363653052515302e-07,
+      "loss": 1.0371,
+      "step": 5820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2306451348358228e-07,
+      "loss": 0.9892,
+      "step": 5821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2249380627185781e-07,
+      "loss": 0.9598,
+      "step": 5822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2192440904325864e-07,
+      "loss": 1.0425,
+      "step": 5823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2135632195071133e-07,
+      "loss": 0.861,
+      "step": 5824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2078954514678996e-07,
+      "loss": 0.8608,
+      "step": 5825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.202240787837178e-07,
+      "loss": 1.0355,
+      "step": 5826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.196599230133666e-07,
+      "loss": 0.9619,
+      "step": 5827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1909707798725412e-07,
+      "loss": 1.0278,
+      "step": 5828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1853554385654775e-07,
+      "loss": 1.1461,
+      "step": 5829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1797532077206187e-07,
+      "loss": 1.2074,
+      "step": 5830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1741640888425997e-07,
+      "loss": 0.9184,
+      "step": 5831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1685880834325203e-07,
+      "loss": 0.9438,
+      "step": 5832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1630251929879654e-07,
+      "loss": 0.8509,
+      "step": 5833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1574754190030014e-07,
+      "loss": 0.9632,
+      "step": 5834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1519387629681578e-07,
+      "loss": 0.6182,
+      "step": 5835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1464152263704565e-07,
+      "loss": 0.8957,
+      "step": 5836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1409048106933828e-07,
+      "loss": 1.0822,
+      "step": 5837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1354075174169088e-07,
+      "loss": 0.9574,
+      "step": 5838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.129923348017481e-07,
+      "loss": 0.9757,
+      "step": 5839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.12445230396801e-07,
+      "loss": 0.9925,
+      "step": 5840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1189943867378927e-07,
+      "loss": 1.1013,
+      "step": 5841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1135495977930011e-07,
+      "loss": 0.8676,
+      "step": 5842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1081179385956708e-07,
+      "loss": 0.755,
+      "step": 5843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1026994106047296e-07,
+      "loss": 0.9932,
+      "step": 5844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0972940152754463e-07,
+      "loss": 0.9772,
+      "step": 5845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0919017540595933e-07,
+      "loss": 1.1659,
+      "step": 5846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0865226284054065e-07,
+      "loss": 1.2653,
+      "step": 5847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0811566397575912e-07,
+      "loss": 1.1179,
+      "step": 5848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0758037895573226e-07,
+      "loss": 1.0382,
+      "step": 5849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0704640792422616e-07,
+      "loss": 1.1326,
+      "step": 5850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0651375102465056e-07,
+      "loss": 0.9203,
+      "step": 5851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0598240840006658e-07,
+      "loss": 0.8392,
+      "step": 5852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0545238019317894e-07,
+      "loss": 0.8946,
+      "step": 5853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0492366654634211e-07,
+      "loss": 0.905,
+      "step": 5854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.043962676015553e-07,
+      "loss": 0.9712,
+      "step": 5855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0387018350046519e-07,
+      "loss": 1.088,
+      "step": 5856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0334541438436596e-07,
+      "loss": 0.9058,
+      "step": 5857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0282196039419823e-07,
+      "loss": 0.8914,
+      "step": 5858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0229982167054897e-07,
+      "loss": 1.1028,
+      "step": 5859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0177899835365323e-07,
+      "loss": 0.9135,
+      "step": 5860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0125949058339023e-07,
+      "loss": 1.1281,
+      "step": 5861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0074129849928948e-07,
+      "loss": 0.8949,
+      "step": 5862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0022442224052409e-07,
+      "loss": 0.9947,
+      "step": 5863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.970886194591467e-08,
+      "loss": 0.9862,
+      "step": 5864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.919461775392936e-08,
+      "loss": 1.2714,
+      "step": 5865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.8681689802681e-08,
+      "loss": 1.189,
+      "step": 5866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.817007822993108e-08,
+      "loss": 1.1321,
+      "step": 5867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.765978317308522e-08,
+      "loss": 0.9357,
+      "step": 5868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.715080476919714e-08,
+      "loss": 1.1538,
+      "step": 5869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.664314315496692e-08,
+      "loss": 1.0873,
+      "step": 5870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.613679846673996e-08,
+      "loss": 0.9045,
+      "step": 5871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.5631770840508e-08,
+      "loss": 1.1879,
+      "step": 5872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.512806041191091e-08,
+      "loss": 0.9854,
+      "step": 5873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.462566731623213e-08,
+      "loss": 1.0303,
+      "step": 5874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.412459168840372e-08,
+      "loss": 1.0194,
+      "step": 5875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.36248336630019e-08,
+      "loss": 1.0287,
+      "step": 5876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.312639337425045e-08,
+      "loss": 0.9845,
+      "step": 5877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.262927095601782e-08,
+      "loss": 0.9711,
+      "step": 5878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.213346654182054e-08,
+      "loss": 0.9049,
+      "step": 5879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.163898026481876e-08,
+      "loss": 1.0255,
+      "step": 5880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.114581225782126e-08,
+      "loss": 0.9144,
+      "step": 5881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.065396265327986e-08,
+      "loss": 0.9689,
+      "step": 5882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.016343158329498e-08,
+      "loss": 1.3303,
+      "step": 5883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.967421917961072e-08,
+      "loss": 0.7269,
+      "step": 5884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.918632557361862e-08,
+      "loss": 1.0029,
+      "step": 5885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.869975089635552e-08,
+      "loss": 0.9787,
+      "step": 5886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.821449527850357e-08,
+      "loss": 0.8144,
+      "step": 5887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.773055885039072e-08,
+      "loss": 0.9509,
+      "step": 5888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.724794174199136e-08,
+      "loss": 0.9669,
+      "step": 5889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.676664408292457e-08,
+      "loss": 1.1325,
+      "step": 5890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.628666600245639e-08,
+      "loss": 1.0032,
+      "step": 5891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.580800762949704e-08,
+      "loss": 1.0095,
+      "step": 5892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.533066909260201e-08,
+      "loss": 0.8497,
+      "step": 5893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.485465051997488e-08,
+      "loss": 0.9583,
+      "step": 5894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.437995203946169e-08,
+      "loss": 1.074,
+      "step": 5895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.39065737785566e-08,
+      "loss": 1.2412,
+      "step": 5896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.343451586439622e-08,
+      "loss": 1.1242,
+      "step": 5897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.296377842376524e-08,
+      "loss": 1.3834,
+      "step": 5898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.249436158309199e-08,
+      "loss": 0.8812,
+      "step": 5899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.202626546845172e-08,
+      "loss": 0.9163,
+      "step": 5900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.15594902055633e-08,
+      "loss": 0.8535,
+      "step": 5901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.109403591979148e-08,
+      "loss": 0.9953,
+      "step": 5902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.062990273614624e-08,
+      "loss": 1.159,
+      "step": 5903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.016709077928397e-08,
+      "loss": 1.1624,
+      "step": 5904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.970560017350304e-08,
+      "loss": 1.0283,
+      "step": 5905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.924543104275095e-08,
+      "loss": 0.9482,
+      "step": 5906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.878658351061663e-08,
+      "loss": 0.9885,
+      "step": 5907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.832905770033705e-08,
+      "loss": 1.2087,
+      "step": 5908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.787285373479169e-08,
+      "loss": 1.0415,
+      "step": 5909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.7417971736507e-08,
+      "loss": 1.2946,
+      "step": 5910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.696441182765357e-08,
+      "loss": 1.2588,
+      "step": 5911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.651217413004674e-08,
+      "loss": 1.2094,
+      "step": 5912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.606125876514603e-08,
+      "loss": 0.8961,
+      "step": 5913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.561166585405789e-08,
+      "loss": 1.2521,
+      "step": 5914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.516339551753182e-08,
+      "loss": 0.8652,
+      "step": 5915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.47164478759621e-08,
+      "loss": 1.136,
+      "step": 5916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.427082304938993e-08,
+      "loss": 1.1211,
+      "step": 5917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.382652115749789e-08,
+      "loss": 0.9667,
+      "step": 5918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.338354231961664e-08,
+      "loss": 0.9821,
+      "step": 5919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.294188665471769e-08,
+      "loss": 1.0036,
+      "step": 5920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.250155428142169e-08,
+      "loss": 1.0722,
+      "step": 5921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.206254531799018e-08,
+      "loss": 0.8516,
+      "step": 5922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.162485988233048e-08,
+      "loss": 0.9931,
+      "step": 5923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.118849809199524e-08,
+      "loss": 1.0854,
+      "step": 5924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.075346006418072e-08,
+      "loss": 1.2043,
+      "step": 5925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.031974591572732e-08,
+      "loss": 1.0038,
+      "step": 5926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.988735576312189e-08,
+      "loss": 1.0622,
+      "step": 5927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.945628972249208e-08,
+      "loss": 1.2329,
+      "step": 5928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.902654790961416e-08,
+      "loss": 1.1194,
+      "step": 5929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.859813043990526e-08,
+      "loss": 1.3554,
+      "step": 5930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.81710374284289e-08,
+      "loss": 1.0915,
+      "step": 5931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.77452689898922e-08,
+      "loss": 0.7978,
+      "step": 5932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.732082523864592e-08,
+      "loss": 0.658,
+      "step": 5933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.689770628868609e-08,
+      "loss": 1.0688,
+      "step": 5934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.647591225365236e-08,
+      "loss": 0.9761,
+      "step": 5935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.605544324682855e-08,
+      "loss": 0.7618,
+      "step": 5936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.563629938114325e-08,
+      "loss": 0.9623,
+      "step": 5937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.521848076916859e-08,
+      "loss": 0.8562,
+      "step": 5938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.480198752311984e-08,
+      "loss": 1.0682,
+      "step": 5939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.438681975485805e-08,
+      "loss": 1.339,
+      "step": 5940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.397297757588739e-08,
+      "loss": 1.0947,
+      "step": 5941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.356046109735614e-08,
+      "loss": 0.9595,
+      "step": 5942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.314927043005625e-08,
+      "loss": 0.747,
+      "step": 5943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.273940568442327e-08,
+      "loss": 0.9359,
+      "step": 5944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.233086697053858e-08,
+      "loss": 0.8602,
+      "step": 5945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.192365439812553e-08,
+      "loss": 0.9558,
+      "step": 5946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.151776807655108e-08,
+      "loss": 1.1384,
+      "step": 5947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.111320811482802e-08,
+      "loss": 1.1661,
+      "step": 5948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.070997462161055e-08,
+      "loss": 0.9851,
+      "step": 5949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.030806770519815e-08,
+      "loss": 1.166,
+      "step": 5950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.99074874735328e-08,
+      "loss": 1.1873,
+      "step": 5951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9508234034202364e-08,
+      "loss": 1.1515,
+      "step": 5952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.911030749443547e-08,
+      "loss": 1.0613,
+      "step": 5953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.871370796110665e-08,
+      "loss": 1.0767,
+      "step": 5954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8318435540732357e-08,
+      "loss": 0.9376,
+      "step": 5955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7924490339474335e-08,
+      "loss": 0.8292,
+      "step": 5956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.753187246313685e-08,
+      "loss": 1.274,
+      "step": 5957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7140582017167764e-08,
+      "loss": 0.959,
+      "step": 5958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6750619106657466e-08,
+      "loss": 1.0719,
+      "step": 5959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.636198383634217e-08,
+      "loss": 0.98,
+      "step": 5960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5974676310599497e-08,
+      "loss": 0.7495,
+      "step": 5961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.558869663345123e-08,
+      "loss": 0.7863,
+      "step": 5962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.520404490856224e-08,
+      "loss": 1.0727,
+      "step": 5963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.482072123924098e-08,
+      "loss": 0.9039,
+      "step": 5964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.443872572843956e-08,
+      "loss": 1.0524,
+      "step": 5965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.405805847875256e-08,
+      "loss": 0.8804,
+      "step": 5966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.367871959241766e-08,
+      "loss": 1.1656,
+      "step": 5967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.330070917131724e-08,
+      "loss": 0.9397,
+      "step": 5968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.292402731697621e-08,
+      "loss": 0.9025,
+      "step": 5969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2548674130561974e-08,
+      "loss": 1.2081,
+      "step": 5970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.217464971288444e-08,
+      "loss": 1.0228,
+      "step": 5971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1801954164399925e-08,
+      "loss": 0.9957,
+      "step": 5972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1430587585203916e-08,
+      "loss": 1.3242,
+      "step": 5973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.106055007503774e-08,
+      "loss": 1.0755,
+      "step": 5974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.069184173328412e-08,
+      "loss": 0.9196,
+      "step": 5975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0324462658969395e-08,
+      "loss": 0.9772,
+      "step": 5976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9958412950763536e-08,
+      "loss": 1.0539,
+      "step": 5977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.959369270697789e-08,
+      "loss": 1.2496,
+      "step": 5978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.923030202556911e-08,
+      "loss": 0.8351,
+      "step": 5979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.886824100413412e-08,
+      "loss": 0.8031,
+      "step": 5980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.850750973991458e-08,
+      "loss": 1.0009,
+      "step": 5981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.814810832979411e-08,
+      "loss": 1.2363,
+      "step": 5982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.779003687029993e-08,
+      "loss": 1.0774,
+      "step": 5983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.743329545760122e-08,
+      "loss": 1.2565,
+      "step": 5984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.707788418750969e-08,
+      "loss": 1.1022,
+      "step": 5985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.67238031554812e-08,
+      "loss": 1.2069,
+      "step": 5986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.637105245661411e-08,
+      "loss": 1.2191,
+      "step": 5987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6019632185647645e-08,
+      "loss": 1.0377,
+      "step": 5988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.566954243696631e-08,
+      "loss": 1.0852,
+      "step": 5989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532078330459433e-08,
+      "loss": 0.9794,
+      "step": 5990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4973354882201784e-08,
+      "loss": 0.9743,
+      "step": 5991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4627257263098465e-08,
+      "loss": 0.8697,
+      "step": 5992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.428249054023892e-08,
+      "loss": 1.132,
+      "step": 5993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.393905480621907e-08,
+      "loss": 1.3374,
+      "step": 5994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.359695015327792e-08,
+      "loss": 1.1668,
+      "step": 5995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3256176673295846e-08,
+      "loss": 1.1101,
+      "step": 5996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.291673445779798e-08,
+      "loss": 1.1406,
+      "step": 5997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2578623597949174e-08,
+      "loss": 0.7874,
+      "step": 5998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.224184418455846e-08,
+      "loss": 0.9538,
+      "step": 5999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1906396308077356e-08,
+      "loss": 0.9881,
+      "step": 6000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.157228005859881e-08,
+      "loss": 1.3446,
+      "step": 6001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.123949552585826e-08,
+      "loss": 1.0005,
+      "step": 6002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0908042799234215e-08,
+      "loss": 1.2062,
+      "step": 6003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0577921967747126e-08,
+      "loss": 1.1592,
+      "step": 6004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.02491331200594e-08,
+      "loss": 1.1083,
+      "step": 6005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9921676344475966e-08,
+      "loss": 1.2179,
+      "step": 6006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9595551728943695e-08,
+      "loss": 1.1639,
+      "step": 6007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.927075936105307e-08,
+      "loss": 1.2509,
+      "step": 6008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.894729932803432e-08,
+      "loss": 1.0033,
+      "step": 6009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8625171716762385e-08,
+      "loss": 0.9882,
+      "step": 6010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.830437661375253e-08,
+      "loss": 1.0517,
+      "step": 6011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7984914105162474e-08,
+      "loss": 1.0009,
+      "step": 6012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7666784276792487e-08,
+      "loss": 1.0505,
+      "step": 6013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7349987214084784e-08,
+      "loss": 1.0467,
+      "step": 6014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7034523002124065e-08,
+      "loss": 0.8865,
+      "step": 6015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.672039172563646e-08,
+      "loss": 1.0,
+      "step": 6016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6407593468989474e-08,
+      "loss": 1.0843,
+      "step": 6017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.609612831619369e-08,
+      "loss": 1.0801,
+      "step": 6018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.578599635090163e-08,
+      "loss": 0.8658,
+      "step": 6019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.547719765640778e-08,
+      "loss": 1.1871,
+      "step": 6020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5169732315646355e-08,
+      "loss": 1.3909,
+      "step": 6021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4863600411197404e-08,
+      "loss": 1.2022,
+      "step": 6022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4558802025279614e-08,
+      "loss": 1.1158,
+      "step": 6023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.425533723975527e-08,
+      "loss": 0.9164,
+      "step": 6024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395320613612752e-08,
+      "loss": 1.1265,
+      "step": 6025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.365240879554144e-08,
+      "loss": 1.2401,
+      "step": 6026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3352945298784076e-08,
+      "loss": 1.1245,
+      "step": 6027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3054815726285e-08,
+      "loss": 1.2069,
+      "step": 6028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2758020158114027e-08,
+      "loss": 1.0107,
+      "step": 6029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2462558673983516e-08,
+      "loss": 1.1911,
+      "step": 6030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.216843135324777e-08,
+      "loss": 1.4365,
+      "step": 6031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1875638274902476e-08,
+      "loss": 1.045,
+      "step": 6032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1584179517584744e-08,
+      "loss": 1.1361,
+      "step": 6033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.129405515957307e-08,
+      "loss": 1.0174,
+      "step": 6034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1005265278789e-08,
+      "loss": 1.1772,
+      "step": 6035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.071780995279439e-08,
+      "loss": 0.809,
+      "step": 6036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0431689258792475e-08,
+      "loss": 1.0056,
+      "step": 6037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.014690327362846e-08,
+      "loss": 1.019,
+      "step": 6038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.986345207379004e-08,
+      "loss": 1.2919,
+      "step": 6039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9581335735404672e-08,
+      "loss": 0.8241,
+      "step": 6040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9300554334242836e-08,
+      "loss": 0.9816,
+      "step": 6041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9021107945714777e-08,
+      "loss": 1.1788,
+      "step": 6042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8742996644874343e-08,
+      "loss": 1.2598,
+      "step": 6043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8466220506414565e-08,
+      "loss": 1.1823,
+      "step": 6044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.819077960467209e-08,
+      "loss": 1.0925,
+      "step": 6045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.79166740136233e-08,
+      "loss": 1.1673,
+      "step": 6046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.764390380688653e-08,
+      "loss": 1.0281,
+      "step": 6047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7372469057721506e-08,
+      "loss": 1.0302,
+      "step": 6048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.710236983902881e-08,
+      "loss": 0.8291,
+      "step": 6049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6833606223351515e-08,
+      "loss": 1.011,
+      "step": 6050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6566178282872445e-08,
+      "loss": 0.8507,
+      "step": 6051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6300086089416366e-08,
+      "loss": 1.1705,
+      "step": 6052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6035329714450553e-08,
+      "loss": 0.9846,
+      "step": 6053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.577190922908035e-08,
+      "loss": 0.9978,
+      "step": 6054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5509824704056385e-08,
+      "loss": 1.1179,
+      "step": 6055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5249076209767353e-08,
+      "loss": 0.9445,
+      "step": 6056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4989663816243904e-08,
+      "loss": 0.9862,
+      "step": 6057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.473158759315808e-08,
+      "loss": 0.9253,
+      "step": 6058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4474847609823882e-08,
+      "loss": 1.1781,
+      "step": 6059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.421944393519504e-08,
+      "loss": 1.0802,
+      "step": 6060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3965376637866688e-08,
+      "loss": 1.0414,
+      "step": 6061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3712645786075905e-08,
+      "loss": 0.9889,
+      "step": 6062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3461251447699507e-08,
+      "loss": 1.0962,
+      "step": 6063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3211193690257373e-08,
+      "loss": 0.7063,
+      "step": 6064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2962472580907446e-08,
+      "loss": 1.09,
+      "step": 6065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.271508818645185e-08,
+      "loss": 1.0329,
+      "step": 6066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2469040573331324e-08,
+      "loss": 0.9361,
+      "step": 6067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.222432980762912e-08,
+      "loss": 1.0775,
+      "step": 6068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.198095595506766e-08,
+      "loss": 1.3929,
+      "step": 6069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1738919081012446e-08,
+      "loss": 1.0233,
+      "step": 6070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1498219250468134e-08,
+      "loss": 1.0295,
+      "step": 6071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1258856528081906e-08,
+      "loss": 1.1014,
+      "step": 6072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1020830978140117e-08,
+      "loss": 1.1005,
+      "step": 6073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0784142664571626e-08,
+      "loss": 0.9106,
+      "step": 6074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0548791650944476e-08,
+      "loss": 1.3674,
+      "step": 6075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.031477800046866e-08,
+      "loss": 0.971,
+      "step": 6076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.008210177599501e-08,
+      "loss": 1.0763,
+      "step": 6077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9850763040014654e-08,
+      "loss": 0.8603,
+      "step": 6078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.962076185466011e-08,
+      "loss": 0.9439,
+      "step": 6079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.939209828170363e-08,
+      "loss": 1.182,
+      "step": 6080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9164772382559426e-08,
+      "loss": 1.1099,
+      "step": 6081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8938784218281435e-08,
+      "loss": 0.9733,
+      "step": 6082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.871413384956444e-08,
+      "loss": 0.9965,
+      "step": 6083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.849082133674518e-08,
+      "loss": 1.1091,
+      "step": 6084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8268846739799562e-08,
+      "loss": 1.3415,
+      "step": 6085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.80482101183449e-08,
+      "loss": 0.9082,
+      "step": 6086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7828911531639348e-08,
+      "loss": 0.9508,
+      "step": 6087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.761095103858024e-08,
+      "loss": 1.299,
+      "step": 6088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7394328697707407e-08,
+      "loss": 0.9381,
+      "step": 6089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.717904456720043e-08,
+      "loss": 0.9972,
+      "step": 6090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6965098704879724e-08,
+      "loss": 0.9427,
+      "step": 6091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.675249116820543e-08,
+      "loss": 1.0209,
+      "step": 6092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.654122201428021e-08,
+      "loss": 1.0728,
+      "step": 6093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6331291299844233e-08,
+      "loss": 1.0923,
+      "step": 6094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612269908128128e-08,
+      "loss": 1.167,
+      "step": 6095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5915445414613208e-08,
+      "loss": 0.9822,
+      "step": 6096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570953035550493e-08,
+      "loss": 1.323,
+      "step": 6097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.550495395925944e-08,
+      "loss": 1.145,
+      "step": 6098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.530171628082111e-08,
+      "loss": 0.8194,
+      "step": 6099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5099817374774615e-08,
+      "loss": 1.101,
+      "step": 6100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4899257295346026e-08,
+      "loss": 1.2648,
+      "step": 6101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4700036096400028e-08,
+      "loss": 0.9179,
+      "step": 6102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4502153831443821e-08,
+      "loss": 0.678,
+      "step": 6103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4305610553623228e-08,
+      "loss": 1.3386,
+      "step": 6104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4110406315724911e-08,
+      "loss": 1.1283,
+      "step": 6105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3916541170176934e-08,
+      "loss": 0.9451,
+      "step": 6106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3724015169046533e-08,
+      "loss": 1.0414,
+      "step": 6107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3532828364041239e-08,
+      "loss": 1.1548,
+      "step": 6108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3342980806509976e-08,
+      "loss": 0.7973,
+      "step": 6109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3154472547440289e-08,
+      "loss": 1.2072,
+      "step": 6110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2967303637462792e-08,
+      "loss": 0.9674,
+      "step": 6111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2781474126845051e-08,
+      "loss": 1.2392,
+      "step": 6112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2596984065497142e-08,
+      "loss": 1.121,
+      "step": 6113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.241383350296832e-08,
+      "loss": 1.156,
+      "step": 6114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2232022488449236e-08,
+      "loss": 0.9796,
+      "step": 6115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2051551070769719e-08,
+      "loss": 1.2399,
+      "step": 6116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1872419298399884e-08,
+      "loss": 0.9877,
+      "step": 6117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1694627219450694e-08,
+      "loss": 1.1761,
+      "step": 6118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1518174881672284e-08,
+      "loss": 1.2002,
+      "step": 6119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.134306233245619e-08,
+      "loss": 1.2834,
+      "step": 6120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1169289618833679e-08,
+      "loss": 1.2729,
+      "step": 6121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0996856787475197e-08,
+      "loss": 1.0603,
+      "step": 6122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0825763884692586e-08,
+      "loss": 0.8027,
+      "step": 6123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0656010956437979e-08,
+      "loss": 1.3576,
+      "step": 6124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0487598048301573e-08,
+      "loss": 1.1739,
+      "step": 6125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0320525205516629e-08,
+      "loss": 1.2361,
+      "step": 6126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0154792472953367e-08,
+      "loss": 0.9122,
+      "step": 6127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.990399895125624e-09,
+      "loss": 1.0411,
+      "step": 6128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.827347516183639e-09,
+      "loss": 1.2688,
+      "step": 6129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.665635379920157e-09,
+      "loss": 1.0206,
+      "step": 6130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.50526352976766e-09,
+      "loss": 0.9327,
+      "step": 6131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.346232008797252e-09,
+      "loss": 1.0974,
+      "step": 6132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.188540859721983e-09,
+      "loss": 0.9779,
+      "step": 6133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.032190124893536e-09,
+      "loss": 0.9496,
+      "step": 6134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.87717984630443e-09,
+      "loss": 1.3457,
+      "step": 6135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.723510065585806e-09,
+      "loss": 0.8761,
+      "step": 6136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.571180824010761e-09,
+      "loss": 1.0825,
+      "step": 6137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.42019216249046e-09,
+      "loss": 0.9757,
+      "step": 6138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.270544121578017e-09,
+      "loss": 1.0911,
+      "step": 6139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.122236741464618e-09,
+      "loss": 1.0992,
+      "step": 6140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.975270061981732e-09,
+      "loss": 1.0352,
+      "step": 6141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.82964412260223e-09,
+      "loss": 0.9897,
+      "step": 6142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.685358962436495e-09,
+      "loss": 1.0888,
+      "step": 6143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.542414620237414e-09,
+      "loss": 1.0429,
+      "step": 6144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.400811134395946e-09,
+      "loss": 1.2134,
+      "step": 6145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260548542943335e-09,
+      "loss": 0.7132,
+      "step": 6146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.121626883550558e-09,
+      "loss": 0.992,
+      "step": 6147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.984046193528881e-09,
+      "loss": 1.0893,
+      "step": 6148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.847806509829858e-09,
+      "loss": 0.8812,
+      "step": 6149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.712907869043661e-09,
+      "loss": 1.0877,
+      "step": 6150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.579350307400756e-09,
+      "loss": 1.1975,
+      "step": 6151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.447133860771893e-09,
+      "loss": 0.8254,
+      "step": 6152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.316258564667555e-09,
+      "loss": 0.8784,
+      "step": 6153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.186724454236847e-09,
+      "loss": 1.1618,
+      "step": 6154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.058531564270276e-09,
+      "loss": 0.8854,
+      "step": 6155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9316799291969654e-09,
+      "loss": 1.0106,
+      "step": 6156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.806169583086885e-09,
+      "loss": 1.1667,
+      "step": 6157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.682000559649181e-09,
+      "loss": 1.1925,
+      "step": 6158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5591728922316235e-09,
+      "loss": 1.3126,
+      "step": 6159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.437686613823934e-09,
+      "loss": 1.1795,
+      "step": 6160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.317541757054456e-09,
+      "loss": 0.8875,
+      "step": 6161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.198738354190158e-09,
+      "loss": 0.6164,
+      "step": 6162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.081276437140514e-09,
+      "loss": 1.0617,
+      "step": 6163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9651560374514015e-09,
+      "loss": 0.8768,
+      "step": 6164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8503771863106504e-09,
+      "loss": 0.7431,
+      "step": 6165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.736939914545824e-09,
+      "loss": 1.117,
+      "step": 6166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.624844252622551e-09,
+      "loss": 1.1741,
+      "step": 6167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.514090230647305e-09,
+      "loss": 1.1223,
+      "step": 6168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4046778783662924e-09,
+      "loss": 1.1542,
+      "step": 6169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.296607225164895e-09,
+      "loss": 0.9774,
+      "step": 6170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.189878300068228e-09,
+      "loss": 1.0604,
+      "step": 6171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.084491131741697e-09,
+      "loss": 1.1012,
+      "step": 6172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9804457484887705e-09,
+      "loss": 0.8017,
+      "step": 6173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.877742178254873e-09,
+      "loss": 1.0768,
+      "step": 6174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.77638044862294e-09,
+      "loss": 0.9044,
+      "step": 6175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6763605868167516e-09,
+      "loss": 1.2225,
+      "step": 6176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.577682619698708e-09,
+      "loss": 1.0936,
+      "step": 6177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4803465737714983e-09,
+      "loss": 1.0713,
+      "step": 6178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3843524751780988e-09,
+      "loss": 1.0131,
+      "step": 6179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.289700349698999e-09,
+      "loss": 1.0748,
+      "step": 6180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1963902227566403e-09,
+      "loss": 1.1248,
+      "step": 6181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.104422119411532e-09,
+      "loss": 1.2812,
+      "step": 6182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0137960643639165e-09,
+      "loss": 1.2072,
+      "step": 6183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9245120819543226e-09,
+      "loss": 1.3435,
+      "step": 6184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.836570196161903e-09,
+      "loss": 1.1165,
+      "step": 6185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.749970430605542e-09,
+      "loss": 0.9193,
+      "step": 6186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.664712808544412e-09,
+      "loss": 1.1939,
+      "step": 6187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5807973528768626e-09,
+      "loss": 0.9076,
+      "step": 6188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4982240861398664e-09,
+      "loss": 1.2752,
+      "step": 6189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.416993030511239e-09,
+      "loss": 1.1974,
+      "step": 6190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.337104207807417e-09,
+      "loss": 1.0849,
+      "step": 6191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.258557639484571e-09,
+      "loss": 0.9571,
+      "step": 6192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.181353346638604e-09,
+      "loss": 0.9358,
+      "step": 6193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1054913500051512e-09,
+      "loss": 1.0539,
+      "step": 6194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0309716699584704e-09,
+      "loss": 1.0591,
+      "step": 6195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.957794326513107e-09,
+      "loss": 1.1803,
+      "step": 6196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8859593393222296e-09,
+      "loss": 1.0162,
+      "step": 6197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8154667276798488e-09,
+      "loss": 0.9795,
+      "step": 6198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7463165105180423e-09,
+      "loss": 0.9477,
+      "step": 6199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6785087064086213e-09,
+      "loss": 0.9997,
+      "step": 6200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6120433335642395e-09,
+      "loss": 0.8656,
+      "step": 6201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.546920409834507e-09,
+      "loss": 1.078,
+      "step": 6202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.483139952710988e-09,
+      "loss": 0.938,
+      "step": 6203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4207019793238686e-09,
+      "loss": 1.2189,
+      "step": 6204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3596065064414022e-09,
+      "loss": 1.1115,
+      "step": 6205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.299853550472685e-09,
+      "loss": 0.9886,
+      "step": 6206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2414431274665463e-09,
+      "loss": 0.8804,
+      "step": 6207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1843752531104368e-09,
+      "loss": 0.8867,
+      "step": 6208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.128649942730986e-09,
+      "loss": 0.8889,
+      "step": 6209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0742672112951103e-09,
+      "loss": 1.0879,
+      "step": 6210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0212270734089035e-09,
+      "loss": 1.064,
+      "step": 6211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.695295433170826e-10,
+      "loss": 0.8998,
+      "step": 6212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.191746349052067e-10,
+      "loss": 1.0595,
+      "step": 6213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.701623616963472e-10,
+      "loss": 0.9048,
+      "step": 6214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.22492736854974e-10,
+      "loss": 1.2133,
+      "step": 6215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.761657731836236e-10,
+      "loss": 1.3538,
+      "step": 6216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.311814831245656e-10,
+      "loss": 0.9084,
+      "step": 6217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.87539878759802e-10,
+      "loss": 1.1365,
+      "step": 6218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.452409718105124e-10,
+      "loss": 1.0603,
+      "step": 6219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.042847736364987e-10,
+      "loss": 0.8445,
+      "step": 6220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.646712952384059e-10,
+      "loss": 1.0495,
+      "step": 6221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.264005472549461e-10,
+      "loss": 1.1927,
+      "step": 6222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.89472539965119e-10,
+      "loss": 1.0318,
+      "step": 6223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.538872832865471e-10,
+      "loss": 0.8531,
+      "step": 6224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.196447867771403e-10,
+      "loss": 1.1337,
+      "step": 6225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.867450596328759e-10,
+      "loss": 0.9191,
+      "step": 6226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.551881106905741e-10,
+      "loss": 1.0659,
+      "step": 6227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2497394842512244e-10,
+      "loss": 1.1631,
+      "step": 6228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9610258095169596e-10,
+      "loss": 1.0267,
+      "step": 6229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.685740160240924e-10,
+      "loss": 0.9769,
+      "step": 6230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.42388261036397e-10,
+      "loss": 1.0502,
+      "step": 6231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1754532302076247e-10,
+      "loss": 1.0385,
+      "step": 6232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9404520865018427e-10,
+      "loss": 1.0126,
+      "step": 6233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.718879242357252e-10,
+      "loss": 0.8799,
+      "step": 6234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5107347572873575e-10,
+      "loss": 0.9628,
+      "step": 6235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.316018687191889e-10,
+      "loss": 1.1608,
+      "step": 6236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1347310843623504e-10,
+      "loss": 0.7681,
+      "step": 6237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.668719974986751e-11,
+      "loss": 0.9883,
+      "step": 6238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.124414716759177e-11,
+      "loss": 1.2551,
+      "step": 6239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.714395483720105e-11,
+      "loss": 0.8053,
+      "step": 6240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.438662654566606e-11,
+      "loss": 0.915,
+      "step": 6241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.297216571969021e-11,
+      "loss": 1.1273,
+      "step": 6242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2900575424599234e-11,
+      "loss": 1.1171,
+      "step": 6243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.417185836545155e-11,
+      "loss": 1.2417,
+      "step": 6244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.678601688648307e-11,
+      "loss": 1.1539,
+      "step": 6245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0743052971107225e-11,
+      "loss": 1.1737,
+      "step": 6246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.042968243025193e-12,
+      "loss": 1.0003,
+      "step": 6247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6857639640054387e-12,
+      "loss": 0.8394,
+      "step": 6248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.714410361041701e-13,
+      "loss": 1.114,
+      "step": 6249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.1109,
+      "step": 6250
+    },
+    {
+      "epoch": 1.0,
+      "step": 6250,
+      "total_flos": 2.181614827040932e+17,
+      "train_loss": 1.165738568868637,
+      "train_runtime": 55605.2108,
+      "train_samples_per_second": 1.798,
+      "train_steps_per_second": 0.112
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6250,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.181614827040932e+17,
+  "trial_name": null,
+  "trial_params": null
+}