diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16225 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 15.0,
+  "global_step": 2700,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.469135802469136e-07,
+      "loss": 1.9679,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.938271604938272e-07,
+      "loss": 2.4497,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.407407407407407e-07,
+      "loss": 1.3278,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.876543209876544e-07,
+      "loss": 1.1368,
+      "step": 4
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.234567901234568e-06,
+      "loss": 0.8084,
+      "step": 5
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 1.9155,
+      "step": 6
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7283950617283952e-06,
+      "loss": 0.871,
+      "step": 7
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9753086419753087e-06,
+      "loss": 1.9553,
+      "step": 8
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.8626,
+      "step": 9
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.469135802469136e-06,
+      "loss": 0.8049,
+      "step": 10
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7160493827160496e-06,
+      "loss": 1.0342,
+      "step": 11
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 0.8569,
+      "step": 12
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2098765432098767e-06,
+      "loss": 0.8774,
+      "step": 13
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4567901234567904e-06,
+      "loss": 1.05,
+      "step": 14
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.8864,
+      "step": 15
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9506172839506175e-06,
+      "loss": 2.3338,
+      "step": 16
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.197530864197531e-06,
+      "loss": 2.0949,
+      "step": 17
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.3458,
+      "step": 18
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.691358024691358e-06,
+      "loss": 0.8166,
+      "step": 19
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 1.2848,
+      "step": 20
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.8669,
+      "step": 21
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.432098765432099e-06,
+      "loss": 1.1391,
+      "step": 22
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.6790123456790125e-06,
+      "loss": 1.5942,
+      "step": 23
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 1.3178,
+      "step": 24
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.17283950617284e-06,
+      "loss": 1.8513,
+      "step": 25
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.419753086419753e-06,
+      "loss": 1.2217,
+      "step": 26
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.824,
+      "step": 27
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.913580246913581e-06,
+      "loss": 0.8527,
+      "step": 28
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.160493827160494e-06,
+      "loss": 0.908,
+      "step": 29
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 1.5864,
+      "step": 30
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.654320987654322e-06,
+      "loss": 1.8379,
+      "step": 31
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.901234567901235e-06,
+      "loss": 1.202,
+      "step": 32
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.148148148148148e-06,
+      "loss": 0.8248,
+      "step": 33
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.395061728395062e-06,
+      "loss": 1.3219,
+      "step": 34
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.641975308641975e-06,
+      "loss": 0.8031,
+      "step": 35
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.9123,
+      "step": 36
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135802469135803e-06,
+      "loss": 0.8443,
+      "step": 37
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.382716049382717e-06,
+      "loss": 0.8434,
+      "step": 38
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 2.2845,
+      "step": 39
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 1.0307,
+      "step": 40
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0123456790123458e-05,
+      "loss": 1.3577,
+      "step": 41
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 1.1748,
+      "step": 42
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0617283950617285e-05,
+      "loss": 0.7827,
+      "step": 43
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0864197530864198e-05,
+      "loss": 0.9275,
+      "step": 44
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.1686,
+      "step": 45
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1358024691358025e-05,
+      "loss": 0.7647,
+      "step": 46
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.160493827160494e-05,
+      "loss": 1.5052,
+      "step": 47
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 1.4286,
+      "step": 48
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2098765432098767e-05,
+      "loss": 1.1518,
+      "step": 49
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.234567901234568e-05,
+      "loss": 0.9073,
+      "step": 50
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2592592592592593e-05,
+      "loss": 0.8353,
+      "step": 51
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2839506172839507e-05,
+      "loss": 0.8511,
+      "step": 52
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.3086419753086422e-05,
+      "loss": 0.7945,
+      "step": 53
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.029,
+      "step": 54
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3580246913580248e-05,
+      "loss": 0.8063,
+      "step": 55
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3827160493827162e-05,
+      "loss": 0.8001,
+      "step": 56
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.9599,
+      "step": 57
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4320987654320988e-05,
+      "loss": 1.0965,
+      "step": 58
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4567901234567903e-05,
+      "loss": 0.8272,
+      "step": 59
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.8877,
+      "step": 60
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.506172839506173e-05,
+      "loss": 0.8304,
+      "step": 61
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308641975308643e-05,
+      "loss": 0.848,
+      "step": 62
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.3311,
+      "step": 63
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.580246913580247e-05,
+      "loss": 0.9286,
+      "step": 64
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6049382716049385e-05,
+      "loss": 0.8746,
+      "step": 65
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.7872,
+      "step": 66
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654320987654321e-05,
+      "loss": 0.82,
+      "step": 67
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6790123456790123e-05,
+      "loss": 0.8844,
+      "step": 68
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.7877,
+      "step": 69
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.728395061728395e-05,
+      "loss": 1.1559,
+      "step": 70
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7530864197530865e-05,
+      "loss": 0.7594,
+      "step": 71
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.9279,
+      "step": 72
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.802469135802469e-05,
+      "loss": 0.9164,
+      "step": 73
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8271604938271607e-05,
+      "loss": 0.9791,
+      "step": 74
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.851851851851852e-05,
+      "loss": 0.7742,
+      "step": 75
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8765432098765433e-05,
+      "loss": 1.0149,
+      "step": 76
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.901234567901235e-05,
+      "loss": 0.8637,
+      "step": 77
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.9429,
+      "step": 78
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9506172839506175e-05,
+      "loss": 0.8175,
+      "step": 79
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 0.8534,
+      "step": 80
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.9939,
+      "step": 81
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9999992805531744e-05,
+      "loss": 0.8103,
+      "step": 82
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9999971222137323e-05,
+      "loss": 0.8172,
+      "step": 83
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9999935249847798e-05,
+      "loss": 0.7838,
+      "step": 84
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.999988488871492e-05,
+      "loss": 0.7979,
+      "step": 85
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9999820138811166e-05,
+      "loss": 0.8573,
+      "step": 86
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9999741000229695e-05,
+      "loss": 0.8774,
+      "step": 87
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999647473084384e-05,
+      "loss": 0.8172,
+      "step": 88
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999539557509805e-05,
+      "loss": 0.8377,
+      "step": 89
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999417253661235e-05,
+      "loss": 0.7389,
+      "step": 90
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999280561714665e-05,
+      "loss": 0.7869,
+      "step": 91
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999129481866773e-05,
+      "loss": 0.8957,
+      "step": 92
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998964014334946e-05,
+      "loss": 1.0358,
+      "step": 93
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998784159357284e-05,
+      "loss": 0.8732,
+      "step": 94
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999858991719257e-05,
+      "loss": 0.7354,
+      "step": 95
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998381288120296e-05,
+      "loss": 0.7101,
+      "step": 96
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9998158272440663e-05,
+      "loss": 0.7945,
+      "step": 97
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9997920870474566e-05,
+      "loss": 0.7991,
+      "step": 98
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9997669082563597e-05,
+      "loss": 0.6828,
+      "step": 99
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999740290907006e-05,
+      "loss": 0.7245,
+      "step": 100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999712235037694e-05,
+      "loss": 0.738,
+      "step": 101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.999682740688794e-05,
+      "loss": 0.8885,
+      "step": 102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.999651807902745e-05,
+      "loss": 0.7335,
+      "step": 103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9996194367240557e-05,
+      "loss": 0.7278,
+      "step": 104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.999585627199305e-05,
+      "loss": 0.8434,
+      "step": 105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.999550379377141e-05,
+      "loss": 0.832,
+      "step": 106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.999513693308282e-05,
+      "loss": 0.7307,
+      "step": 107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9994755690455154e-05,
+      "loss": 0.7012,
+      "step": 108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999436006643697e-05,
+      "loss": 0.7911,
+      "step": 109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9993950061597536e-05,
+      "loss": 0.9348,
+      "step": 110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9993525676526807e-05,
+      "loss": 0.7792,
+      "step": 111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9993086911835425e-05,
+      "loss": 0.8214,
+      "step": 112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9992633768154728e-05,
+      "loss": 0.757,
+      "step": 113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9992166246136738e-05,
+      "loss": 0.7674,
+      "step": 114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999168434645417e-05,
+      "loss": 0.6459,
+      "step": 115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9991188069800434e-05,
+      "loss": 0.8519,
+      "step": 116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999067741688961e-05,
+      "loss": 0.752,
+      "step": 117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9990152388456475e-05,
+      "loss": 0.6926,
+      "step": 118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9989612985256492e-05,
+      "loss": 0.6907,
+      "step": 119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9989059208065803e-05,
+      "loss": 0.6778,
+      "step": 120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9988491057681235e-05,
+      "loss": 0.8137,
+      "step": 121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9987908534920294e-05,
+      "loss": 0.7783,
+      "step": 122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9987311640621173e-05,
+      "loss": 0.6881,
+      "step": 123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9986700375642733e-05,
+      "loss": 0.612,
+      "step": 124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9986074740864527e-05,
+      "loss": 0.7895,
+      "step": 125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.998543473718677e-05,
+      "loss": 0.6869,
+      "step": 126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9984780365530364e-05,
+      "loss": 0.5964,
+      "step": 127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9984111626836877e-05,
+      "loss": 0.7821,
+      "step": 128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9983428522068556e-05,
+      "loss": 0.8673,
+      "step": 129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.998273105220831e-05,
+      "loss": 0.8014,
+      "step": 130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.998201921825973e-05,
+      "loss": 0.7269,
+      "step": 131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.998129302124707e-05,
+      "loss": 0.5893,
+      "step": 132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9980552462215246e-05,
+      "loss": 0.6959,
+      "step": 133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9979797542229845e-05,
+      "loss": 0.7952,
+      "step": 134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.997902826237712e-05,
+      "loss": 0.7875,
+      "step": 135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9978244623763975e-05,
+      "loss": 0.8581,
+      "step": 136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9977446627517988e-05,
+      "loss": 0.6886,
+      "step": 137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9976634274787396e-05,
+      "loss": 0.8221,
+      "step": 138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9975807566741077e-05,
+      "loss": 0.845,
+      "step": 139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9974966504568583e-05,
+      "loss": 0.6314,
+      "step": 140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9974111089480112e-05,
+      "loss": 0.7959,
+      "step": 141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9973241322706515e-05,
+      "loss": 0.7309,
+      "step": 142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9972357205499295e-05,
+      "loss": 0.7387,
+      "step": 143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9971458739130598e-05,
+      "loss": 0.7944,
+      "step": 144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9970545924893227e-05,
+      "loss": 0.6262,
+      "step": 145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.996961876410062e-05,
+      "loss": 0.6638,
+      "step": 146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9968677258086868e-05,
+      "loss": 0.8054,
+      "step": 147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9967721408206695e-05,
+      "loss": 0.7664,
+      "step": 148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9966751215835466e-05,
+      "loss": 0.6883,
+      "step": 149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9965766682369188e-05,
+      "loss": 0.8392,
+      "step": 150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9964767809224495e-05,
+      "loss": 0.753,
+      "step": 151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9963754597838667e-05,
+      "loss": 0.8075,
+      "step": 152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99627270496696e-05,
+      "loss": 0.7079,
+      "step": 153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.996168516619583e-05,
+      "loss": 0.7963,
+      "step": 154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.996062894891652e-05,
+      "loss": 0.752,
+      "step": 155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9959558399351443e-05,
+      "loss": 0.8108,
+      "step": 156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9958473519041016e-05,
+      "loss": 0.6582,
+      "step": 157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9957374309546265e-05,
+      "loss": 0.8095,
+      "step": 158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9956260772448835e-05,
+      "loss": 0.7828,
+      "step": 159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9955132909350986e-05,
+      "loss": 0.6874,
+      "step": 160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.995399072187559e-05,
+      "loss": 0.7393,
+      "step": 161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.995283421166614e-05,
+      "loss": 0.8034,
+      "step": 162
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.995166338038673e-05,
+      "loss": 0.7841,
+      "step": 163
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9950478229722055e-05,
+      "loss": 0.7875,
+      "step": 164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.994927876137743e-05,
+      "loss": 0.6729,
+      "step": 165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9948064977078757e-05,
+      "loss": 0.7523,
+      "step": 166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9946836878572543e-05,
+      "loss": 0.6889,
+      "step": 167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9945594467625895e-05,
+      "loss": 0.5626,
+      "step": 168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9944337746026504e-05,
+      "loss": 0.7366,
+      "step": 169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.994306671558266e-05,
+      "loss": 0.691,
+      "step": 170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9941781378123244e-05,
+      "loss": 0.7358,
+      "step": 171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.994048173549772e-05,
+      "loss": 0.7253,
+      "step": 172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.993916778957613e-05,
+      "loss": 0.6864,
+      "step": 173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9937839542249108e-05,
+      "loss": 0.7875,
+      "step": 174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.993649699542786e-05,
+      "loss": 0.6224,
+      "step": 175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9935140151044166e-05,
+      "loss": 0.8025,
+      "step": 176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9933769011050382e-05,
+      "loss": 0.8491,
+      "step": 177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.7955,
+      "step": 178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9930983852144808e-05,
+      "loss": 0.7417,
+      "step": 179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9929569837240567e-05,
+      "loss": 0.7157,
+      "step": 180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9928141534741324e-05,
+      "loss": 0.4563,
+      "step": 181
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9926698946702255e-05,
+      "loss": 0.5155,
+      "step": 182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.992524207519909e-05,
+      "loss": 0.6969,
+      "step": 183
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9923770922328113e-05,
+      "loss": 0.6372,
+      "step": 184
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9922285490206155e-05,
+      "loss": 0.8519,
+      "step": 185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9920785780970604e-05,
+      "loss": 0.799,
+      "step": 186
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.991927179677937e-05,
+      "loss": 0.6067,
+      "step": 187
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.991774353981092e-05,
+      "loss": 0.686,
+      "step": 188
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9916201012264255e-05,
+      "loss": 0.6267,
+      "step": 189
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9914644216358904e-05,
+      "loss": 0.6014,
+      "step": 190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9913073154334933e-05,
+      "loss": 0.8714,
+      "step": 191
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9911487828452932e-05,
+      "loss": 0.7414,
+      "step": 192
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.990988824099402e-05,
+      "loss": 0.5565,
+      "step": 193
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.990827439425983e-05,
+      "loss": 0.6652,
+      "step": 194
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9906646290572514e-05,
+      "loss": 0.7987,
+      "step": 195
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.990500393227475e-05,
+      "loss": 0.6828,
+      "step": 196
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.99033473217297e-05,
+      "loss": 0.6423,
+      "step": 197
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.990167646132107e-05,
+      "loss": 0.7358,
+      "step": 198
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9899991353453033e-05,
+      "loss": 0.7162,
+      "step": 199
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.989829200055029e-05,
+      "loss": 0.5983,
+      "step": 200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9896578405058028e-05,
+      "loss": 0.7484,
+      "step": 201
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9894850569441922e-05,
+      "loss": 0.7608,
+      "step": 202
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9893108496188156e-05,
+      "loss": 0.8342,
+      "step": 203
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9891352187803375e-05,
+      "loss": 0.6314,
+      "step": 204
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.988958164681473e-05,
+      "loss": 0.7354,
+      "step": 205
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9887796875769836e-05,
+      "loss": 0.6426,
+      "step": 206
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9885997877236788e-05,
+      "loss": 0.511,
+      "step": 207
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.988418465380416e-05,
+      "loss": 0.7781,
+      "step": 208
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.988235720808098e-05,
+      "loss": 0.6919,
+      "step": 209
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.988051554269675e-05,
+      "loss": 0.9014,
+      "step": 210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9878659660301435e-05,
+      "loss": 0.7703,
+      "step": 211
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9876789563565446e-05,
+      "loss": 0.796,
+      "step": 212
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.987490525517966e-05,
+      "loss": 0.6371,
+      "step": 213
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.987300673785539e-05,
+      "loss": 0.6269,
+      "step": 214
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9871094014324405e-05,
+      "loss": 0.7756,
+      "step": 215
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9869167087338908e-05,
+      "loss": 0.6334,
+      "step": 216
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9867225959671545e-05,
+      "loss": 0.6351,
+      "step": 217
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.986527063411539e-05,
+      "loss": 0.6955,
+      "step": 218
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.986330111348395e-05,
+      "loss": 0.5813,
+      "step": 219
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.986131740061115e-05,
+      "loss": 0.6065,
+      "step": 220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9859319498351353e-05,
+      "loss": 0.7264,
+      "step": 221
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.985730740957932e-05,
+      "loss": 0.6817,
+      "step": 222
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.985528113719023e-05,
+      "loss": 0.764,
+      "step": 223
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9853240684099683e-05,
+      "loss": 0.7353,
+      "step": 224
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9851186053243667e-05,
+      "loss": 0.6539,
+      "step": 225
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.984911724757858e-05,
+      "loss": 0.774,
+      "step": 226
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.984703427008121e-05,
+      "loss": 0.597,
+      "step": 227
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9844937123748744e-05,
+      "loss": 0.8259,
+      "step": 228
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9842825811598748e-05,
+      "loss": 0.5955,
+      "step": 229
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9840700336669184e-05,
+      "loss": 0.7648,
+      "step": 230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9838560702018373e-05,
+      "loss": 0.7472,
+      "step": 231
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9836406910725032e-05,
+      "loss": 0.6346,
+      "step": 232
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9834238965888232e-05,
+      "loss": 0.6152,
+      "step": 233
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.983205687062742e-05,
+      "loss": 0.7273,
+      "step": 234
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9829860628082393e-05,
+      "loss": 0.7258,
+      "step": 235
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.982765024141331e-05,
+      "loss": 0.7267,
+      "step": 236
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.982542571380069e-05,
+      "loss": 0.718,
+      "step": 237
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9823187048445383e-05,
+      "loss": 0.785,
+      "step": 238
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9820934248568596e-05,
+      "loss": 0.7575,
+      "step": 239
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9818667317411865e-05,
+      "loss": 0.782,
+      "step": 240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.981638625823707e-05,
+      "loss": 0.758,
+      "step": 241
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9814091074326402e-05,
+      "loss": 0.7433,
+      "step": 242
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9811781768982392e-05,
+      "loss": 0.6718,
+      "step": 243
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9809458345527886e-05,
+      "loss": 0.6168,
+      "step": 244
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9807120807306042e-05,
+      "loss": 0.6419,
+      "step": 245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.980476915768033e-05,
+      "loss": 0.6009,
+      "step": 246
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9802403400034518e-05,
+      "loss": 0.6388,
+      "step": 247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9800023537772687e-05,
+      "loss": 0.6788,
+      "step": 248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9797629574319202e-05,
+      "loss": 0.5894,
+      "step": 249
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9795221513118723e-05,
+      "loss": 0.675,
+      "step": 250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.979279935763619e-05,
+      "loss": 0.6566,
+      "step": 251
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9790363111356838e-05,
+      "loss": 0.757,
+      "step": 252
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9787912777786154e-05,
+      "loss": 0.6402,
+      "step": 253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9785448360449914e-05,
+      "loss": 0.7101,
+      "step": 254
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.978296986289415e-05,
+      "loss": 0.7034,
+      "step": 255
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.978047728868516e-05,
+      "loss": 0.7413,
+      "step": 256
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9777970641409486e-05,
+      "loss": 0.5531,
+      "step": 257
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9775449924673937e-05,
+      "loss": 0.6523,
+      "step": 258
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9772915142105545e-05,
+      "loss": 0.64,
+      "step": 259
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.97703662973516e-05,
+      "loss": 0.7653,
+      "step": 260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9767803394079618e-05,
+      "loss": 0.7102,
+      "step": 261
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9765226435977344e-05,
+      "loss": 0.6858,
+      "step": 262
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.976263542675274e-05,
+      "loss": 0.7313,
+      "step": 263
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9760030370134004e-05,
+      "loss": 0.5887,
+      "step": 264
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.975741126986953e-05,
+      "loss": 0.7584,
+      "step": 265
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.975477812972792e-05,
+      "loss": 0.7952,
+      "step": 266
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.975213095349799e-05,
+      "loss": 0.7328,
+      "step": 267
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9749469744988743e-05,
+      "loss": 0.6569,
+      "step": 268
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.974679450802937e-05,
+      "loss": 0.6957,
+      "step": 269
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9744105246469264e-05,
+      "loss": 0.7651,
+      "step": 270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9741401964177973e-05,
+      "loss": 0.6228,
+      "step": 271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.973868466504524e-05,
+      "loss": 0.625,
+      "step": 272
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.973595335298097e-05,
+      "loss": 0.6785,
+      "step": 273
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9733208031915227e-05,
+      "loss": 0.6839,
+      "step": 274
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.7831,
+      "step": 275
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9727675378600382e-05,
+      "loss": 0.7094,
+      "step": 276
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9724888054312178e-05,
+      "loss": 0.7213,
+      "step": 277
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9722086736944292e-05,
+      "loss": 0.6996,
+      "step": 278
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.971927143052752e-05,
+      "loss": 0.782,
+      "step": 279
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.971644213911279e-05,
+      "loss": 0.7009,
+      "step": 280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.971359886677115e-05,
+      "loss": 0.5135,
+      "step": 281
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9710741617593768e-05,
+      "loss": 0.5765,
+      "step": 282
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.970787039569192e-05,
+      "loss": 0.6195,
+      "step": 283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.970498520519699e-05,
+      "loss": 0.6576,
+      "step": 284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.970208605026046e-05,
+      "loss": 0.7191,
+      "step": 285
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9699172935053907e-05,
+      "loss": 0.5775,
+      "step": 286
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.969624586376899e-05,
+      "loss": 0.706,
+      "step": 287
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9693304840617456e-05,
+      "loss": 0.5933,
+      "step": 288
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9690349869831128e-05,
+      "loss": 0.7383,
+      "step": 289
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968738095566189e-05,
+      "loss": 0.6273,
+      "step": 290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9684398102381694e-05,
+      "loss": 0.6488,
+      "step": 291
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.968140131428255e-05,
+      "loss": 0.6997,
+      "step": 292
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.967839059567652e-05,
+      "loss": 0.7512,
+      "step": 293
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9675365950895703e-05,
+      "loss": 0.6816,
+      "step": 294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.967232738429224e-05,
+      "loss": 0.6823,
+      "step": 295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9669274900238313e-05,
+      "loss": 0.6453,
+      "step": 296
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9666208503126115e-05,
+      "loss": 0.6348,
+      "step": 297
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9663128197367867e-05,
+      "loss": 0.7769,
+      "step": 298
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9660033987395802e-05,
+      "loss": 0.6968,
+      "step": 299
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.965692587766216e-05,
+      "loss": 0.7726,
+      "step": 300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.965380387263918e-05,
+      "loss": 0.7124,
+      "step": 301
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9650667976819094e-05,
+      "loss": 0.6811,
+      "step": 302
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9647518194714125e-05,
+      "loss": 0.6303,
+      "step": 303
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.964435453085647e-05,
+      "loss": 0.6874,
+      "step": 304
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9641176989798306e-05,
+      "loss": 0.5446,
+      "step": 305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.963798557611178e-05,
+      "loss": 0.6245,
+      "step": 306
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9634780294388997e-05,
+      "loss": 0.6632,
+      "step": 307
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9631561149242015e-05,
+      "loss": 0.6569,
+      "step": 308
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9628328145302837e-05,
+      "loss": 0.6729,
+      "step": 309
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.962508128722342e-05,
+      "loss": 0.6232,
+      "step": 310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.962182057967564e-05,
+      "loss": 0.6624,
+      "step": 311
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9618546027351315e-05,
+      "loss": 0.5551,
+      "step": 312
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9615257634962173e-05,
+      "loss": 0.7685,
+      "step": 313
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9611955407239863e-05,
+      "loss": 0.7284,
+      "step": 314
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9608639348935938e-05,
+      "loss": 0.6756,
+      "step": 315
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9605309464821857e-05,
+      "loss": 0.7626,
+      "step": 316
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.960196575968896e-05,
+      "loss": 0.6216,
+      "step": 317
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9598608238348493e-05,
+      "loss": 0.6903,
+      "step": 318
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.959523690563157e-05,
+      "loss": 0.6166,
+      "step": 319
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9591851766389177e-05,
+      "loss": 0.6845,
+      "step": 320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.958845282549217e-05,
+      "loss": 0.6537,
+      "step": 321
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9585040087831268e-05,
+      "loss": 0.7186,
+      "step": 322
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.958161355831703e-05,
+      "loss": 0.7496,
+      "step": 323
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.957817324187987e-05,
+      "loss": 0.5383,
+      "step": 324
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9574719143470044e-05,
+      "loss": 0.7474,
+      "step": 325
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9571251268057626e-05,
+      "loss": 0.6973,
+      "step": 326
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.956776962063252e-05,
+      "loss": 0.7407,
+      "step": 327
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.956427420620445e-05,
+      "loss": 0.7213,
+      "step": 328
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9560765029802944e-05,
+      "loss": 0.6306,
+      "step": 329
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955724209647733e-05,
+      "loss": 0.5895,
+      "step": 330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.955370541129674e-05,
+      "loss": 0.7811,
+      "step": 331
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.955015497935008e-05,
+      "loss": 0.7631,
+      "step": 332
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9546590805746054e-05,
+      "loss": 0.6877,
+      "step": 333
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.954301289561312e-05,
+      "loss": 0.6193,
+      "step": 334
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.953942125409952e-05,
+      "loss": 0.6883,
+      "step": 335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.953581588637323e-05,
+      "loss": 0.6149,
+      "step": 336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9532196797622007e-05,
+      "loss": 0.7496,
+      "step": 337
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9528563993053326e-05,
+      "loss": 0.5435,
+      "step": 338
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9524917477894406e-05,
+      "loss": 0.7413,
+      "step": 339
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9521257257392196e-05,
+      "loss": 0.5975,
+      "step": 340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9517583336813362e-05,
+      "loss": 0.5399,
+      "step": 341
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9513895721444286e-05,
+      "loss": 0.6437,
+      "step": 342
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.951019441659106e-05,
+      "loss": 0.4277,
+      "step": 343
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9506479427579456e-05,
+      "loss": 0.7429,
+      "step": 344
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9502750759754963e-05,
+      "loss": 0.7217,
+      "step": 345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9499008418482726e-05,
+      "loss": 0.7253,
+      "step": 346
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.949525240914758e-05,
+      "loss": 0.5664,
+      "step": 347
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9491482737154023e-05,
+      "loss": 0.4858,
+      "step": 348
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9487699407926213e-05,
+      "loss": 0.624,
+      "step": 349
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9483902426907953e-05,
+      "loss": 0.6605,
+      "step": 350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9480091799562706e-05,
+      "loss": 0.6749,
+      "step": 351
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.947626753137355e-05,
+      "loss": 0.623,
+      "step": 352
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9472429627843202e-05,
+      "loss": 0.7414,
+      "step": 353
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9468578094493998e-05,
+      "loss": 0.6733,
+      "step": 354
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9464712936867886e-05,
+      "loss": 0.54,
+      "step": 355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9460834160526416e-05,
+      "loss": 0.5649,
+      "step": 356
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9456941771050733e-05,
+      "loss": 0.6348,
+      "step": 357
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9453035774041574e-05,
+      "loss": 0.6961,
+      "step": 358
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9449116175119254e-05,
+      "loss": 0.749,
+      "step": 359
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9445182979923657e-05,
+      "loss": 0.6814,
+      "step": 360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9441236194114227e-05,
+      "loss": 0.6311,
+      "step": 361
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.943727582336998e-05,
+      "loss": 0.599,
+      "step": 362
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9433301873389463e-05,
+      "loss": 0.5541,
+      "step": 363
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9429314349890766e-05,
+      "loss": 0.7053,
+      "step": 364
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9425313258611513e-05,
+      "loss": 0.4942,
+      "step": 365
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9421298605308847e-05,
+      "loss": 0.5524,
+      "step": 366
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.941727039575943e-05,
+      "loss": 0.6193,
+      "step": 367
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9413228635759426e-05,
+      "loss": 0.6077,
+      "step": 368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.94091733311245e-05,
+      "loss": 0.5574,
+      "step": 369
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.94051044876898e-05,
+      "loss": 0.6731,
+      "step": 370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9401022111309958e-05,
+      "loss": 0.6931,
+      "step": 371
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.6377,
+      "step": 372
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.939281678323075e-05,
+      "loss": 0.6818,
+      "step": 373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.938869384333797e-05,
+      "loss": 0.7343,
+      "step": 374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.938455739411323e-05,
+      "loss": 0.7753,
+      "step": 375
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9380407441508427e-05,
+      "loss": 0.5579,
+      "step": 376
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.937624399149491e-05,
+      "loss": 0.6887,
+      "step": 377
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.937206705006344e-05,
+      "loss": 0.6516,
+      "step": 378
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9367876623224185e-05,
+      "loss": 0.5289,
+      "step": 379
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9363672717006735e-05,
+      "loss": 0.7521,
+      "step": 380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9359455337460054e-05,
+      "loss": 0.5779,
+      "step": 381
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9355224490652508e-05,
+      "loss": 0.5477,
+      "step": 382
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.935098018267183e-05,
+      "loss": 0.6716,
+      "step": 383
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9346722419625136e-05,
+      "loss": 0.5913,
+      "step": 384
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.934245120763889e-05,
+      "loss": 0.6293,
+      "step": 385
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9338166552858907e-05,
+      "loss": 0.5077,
+      "step": 386
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.933386846145036e-05,
+      "loss": 0.5457,
+      "step": 387
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9329556939597736e-05,
+      "loss": 0.6638,
+      "step": 388
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9325231993504863e-05,
+      "loss": 0.7154,
+      "step": 389
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9320893629394876e-05,
+      "loss": 0.5963,
+      "step": 390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9316541853510214e-05,
+      "loss": 0.6264,
+      "step": 391
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.931217667211263e-05,
+      "loss": 0.4708,
+      "step": 392
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9307798091483146e-05,
+      "loss": 0.6125,
+      "step": 393
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9303406117922082e-05,
+      "loss": 0.6195,
+      "step": 394
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9299000757749016e-05,
+      "loss": 0.6055,
+      "step": 395
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9294582017302797e-05,
+      "loss": 0.5915,
+      "step": 396
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9290149902941517e-05,
+      "loss": 0.5461,
+      "step": 397
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.928570442104252e-05,
+      "loss": 0.5961,
+      "step": 398
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9281245578002382e-05,
+      "loss": 0.7594,
+      "step": 399
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9276773380236903e-05,
+      "loss": 0.7359,
+      "step": 400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9272287834181102e-05,
+      "loss": 0.6078,
+      "step": 401
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9267788946289202e-05,
+      "loss": 0.6342,
+      "step": 402
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9263276723034622e-05,
+      "loss": 0.5272,
+      "step": 403
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9258751170909973e-05,
+      "loss": 0.6045,
+      "step": 404
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9254212296427043e-05,
+      "loss": 0.4815,
+      "step": 405
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9249660106116793e-05,
+      "loss": 0.7319,
+      "step": 406
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9245094606529333e-05,
+      "loss": 0.6941,
+      "step": 407
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.924051580423394e-05,
+      "loss": 0.5432,
+      "step": 408
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9235923705819015e-05,
+      "loss": 0.56,
+      "step": 409
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9231318317892108e-05,
+      "loss": 0.5304,
+      "step": 410
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9226699647079874e-05,
+      "loss": 0.5762,
+      "step": 411
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.92220677000281e-05,
+      "loss": 0.544,
+      "step": 412
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9217422483401652e-05,
+      "loss": 0.5791,
+      "step": 413
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.921276400388451e-05,
+      "loss": 0.6548,
+      "step": 414
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9208092268179733e-05,
+      "loss": 0.4772,
+      "step": 415
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9203407283009444e-05,
+      "loss": 0.7208,
+      "step": 416
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.919870905511485e-05,
+      "loss": 0.5652,
+      "step": 417
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.919399759125619e-05,
+      "loss": 0.6618,
+      "step": 418
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9189272898212764e-05,
+      "loss": 0.6328,
+      "step": 419
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.91845349827829e-05,
+      "loss": 0.7278,
+      "step": 420
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9179783851783968e-05,
+      "loss": 0.6767,
+      "step": 421
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.917501951205232e-05,
+      "loss": 0.7546,
+      "step": 422
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9170241970443344e-05,
+      "loss": 0.716,
+      "step": 423
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9165451233831415e-05,
+      "loss": 0.6541,
+      "step": 424
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.916064730910989e-05,
+      "loss": 0.7502,
+      "step": 425
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.915583020319111e-05,
+      "loss": 0.6769,
+      "step": 426
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.915099992300638e-05,
+      "loss": 0.7257,
+      "step": 427
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9146156475505956e-05,
+      "loss": 0.5032,
+      "step": 428
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9141299867659036e-05,
+      "loss": 0.7467,
+      "step": 429
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9136430106453777e-05,
+      "loss": 0.5,
+      "step": 430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9131547198897234e-05,
+      "loss": 0.6057,
+      "step": 431
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9126651152015404e-05,
+      "loss": 0.6186,
+      "step": 432
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.912174197285317e-05,
+      "loss": 0.6198,
+      "step": 433
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9116819668474318e-05,
+      "loss": 0.6664,
+      "step": 434
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9111884245961524e-05,
+      "loss": 0.7274,
+      "step": 435
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9106935712416336e-05,
+      "loss": 0.6201,
+      "step": 436
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9101974074959167e-05,
+      "loss": 0.5996,
+      "step": 437
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9096999340729284e-05,
+      "loss": 0.6584,
+      "step": 438
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.90920115168848e-05,
+      "loss": 0.7124,
+      "step": 439
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.908701061060267e-05,
+      "loss": 0.5565,
+      "step": 440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9081996629078655e-05,
+      "loss": 0.6215,
+      "step": 441
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.907696957952735e-05,
+      "loss": 0.5823,
+      "step": 442
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9071929469182143e-05,
+      "loss": 0.5801,
+      "step": 443
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9066876305295218e-05,
+      "loss": 0.6503,
+      "step": 444
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9061810095137533e-05,
+      "loss": 0.6918,
+      "step": 445
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9056730845998834e-05,
+      "loss": 0.7696,
+      "step": 446
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.905163856518762e-05,
+      "loss": 0.6035,
+      "step": 447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9046533260031134e-05,
+      "loss": 0.6875,
+      "step": 448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.904141493787537e-05,
+      "loss": 0.5754,
+      "step": 449
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9036283606085057e-05,
+      "loss": 0.6086,
+      "step": 450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9031139272043626e-05,
+      "loss": 0.7085,
+      "step": 451
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.902598194315323e-05,
+      "loss": 0.7045,
+      "step": 452
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.902081162683472e-05,
+      "loss": 0.5474,
+      "step": 453
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9015628330527626e-05,
+      "loss": 0.5298,
+      "step": 454
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9010432061690166e-05,
+      "loss": 0.5182,
+      "step": 455
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9005222827799213e-05,
+      "loss": 0.709,
+      "step": 456
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.90000006363503e-05,
+      "loss": 0.6669,
+      "step": 457
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.899476549485761e-05,
+      "loss": 0.6839,
+      "step": 458
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8989517410853956e-05,
+      "loss": 0.6187,
+      "step": 459
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8984256391890764e-05,
+      "loss": 0.5695,
+      "step": 460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.897898244553809e-05,
+      "loss": 0.463,
+      "step": 461
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8973695579384577e-05,
+      "loss": 0.6901,
+      "step": 462
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.896839580103746e-05,
+      "loss": 0.6564,
+      "step": 463
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8963083118122565e-05,
+      "loss": 0.6801,
+      "step": 464
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8957757538284273e-05,
+      "loss": 0.456,
+      "step": 465
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.895241906918553e-05,
+      "loss": 0.5766,
+      "step": 466
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8947067718507817e-05,
+      "loss": 0.6412,
+      "step": 467
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8941703493951163e-05,
+      "loss": 0.503,
+      "step": 468
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.6497,
+      "step": 469
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8930936454093754e-05,
+      "loss": 0.618,
+      "step": 470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8925533654285617e-05,
+      "loss": 0.6081,
+      "step": 471
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8920118011583766e-05,
+      "loss": 0.6317,
+      "step": 472
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8914689533780743e-05,
+      "loss": 0.6506,
+      "step": 473
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.890924822868754e-05,
+      "loss": 0.7271,
+      "step": 474
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8903794104133618e-05,
+      "loss": 0.5079,
+      "step": 475
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.889832716796689e-05,
+      "loss": 0.5357,
+      "step": 476
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8892847428053692e-05,
+      "loss": 0.588,
+      "step": 477
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8887354892278783e-05,
+      "loss": 0.5926,
+      "step": 478
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8881849568545343e-05,
+      "loss": 0.4338,
+      "step": 479
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8876331464774943e-05,
+      "loss": 0.6809,
+      "step": 480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.887080058890755e-05,
+      "loss": 0.6452,
+      "step": 481
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8865256948901507e-05,
+      "loss": 0.7398,
+      "step": 482
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8859700552733516e-05,
+      "loss": 0.5133,
+      "step": 483
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.885413140839865e-05,
+      "loss": 0.7096,
+      "step": 484
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8848549523910313e-05,
+      "loss": 0.5872,
+      "step": 485
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8842954907300236e-05,
+      "loss": 0.65,
+      "step": 486
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8837347566618484e-05,
+      "loss": 0.5375,
+      "step": 487
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8831727509933423e-05,
+      "loss": 0.6044,
+      "step": 488
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8826094745331717e-05,
+      "loss": 0.7167,
+      "step": 489
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8820449280918312e-05,
+      "loss": 0.6589,
+      "step": 490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8814791124816434e-05,
+      "loss": 0.7576,
+      "step": 491
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8809120285167568e-05,
+      "loss": 0.6712,
+      "step": 492
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8803436770131448e-05,
+      "loss": 0.504,
+      "step": 493
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8797740587886043e-05,
+      "loss": 0.6601,
+      "step": 494
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8792031746627563e-05,
+      "loss": 0.5679,
+      "step": 495
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.878631025457042e-05,
+      "loss": 0.7769,
+      "step": 496
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.878057611994723e-05,
+      "loss": 0.6538,
+      "step": 497
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8774829351008805e-05,
+      "loss": 0.7629,
+      "step": 498
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.876906995602413e-05,
+      "loss": 0.6456,
+      "step": 499
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8763297943280368e-05,
+      "loss": 0.5386,
+      "step": 500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.875751332108283e-05,
+      "loss": 0.6308,
+      "step": 501
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8751716097754967e-05,
+      "loss": 0.6576,
+      "step": 502
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8745906281638376e-05,
+      "loss": 0.5731,
+      "step": 503
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.874008388109276e-05,
+      "loss": 0.665,
+      "step": 504
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8734248904495932e-05,
+      "loss": 0.5102,
+      "step": 505
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8728401360243805e-05,
+      "loss": 0.5933,
+      "step": 506
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8722541256750372e-05,
+      "loss": 0.8162,
+      "step": 507
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8716668602447697e-05,
+      "loss": 0.6081,
+      "step": 508
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.871078340578591e-05,
+      "loss": 0.5437,
+      "step": 509
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.870488567523318e-05,
+      "loss": 0.5528,
+      "step": 510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.869897541927572e-05,
+      "loss": 0.6188,
+      "step": 511
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8693052646417745e-05,
+      "loss": 0.5379,
+      "step": 512
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8687117365181514e-05,
+      "loss": 0.4993,
+      "step": 513
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.868116958410725e-05,
+      "loss": 0.6448,
+      "step": 514
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8675209311753185e-05,
+      "loss": 0.7056,
+      "step": 515
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8669236556695517e-05,
+      "loss": 0.6745,
+      "step": 516
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8663251327528404e-05,
+      "loss": 0.6326,
+      "step": 517
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8657253632863953e-05,
+      "loss": 0.6935,
+      "step": 518
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.865124348133221e-05,
+      "loss": 0.6249,
+      "step": 519
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8645220881581145e-05,
+      "loss": 0.6662,
+      "step": 520
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8639185842276635e-05,
+      "loss": 0.6898,
+      "step": 521
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.863313837210247e-05,
+      "loss": 0.6514,
+      "step": 522
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.86270784797603e-05,
+      "loss": 0.6004,
+      "step": 523
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8621006173969678e-05,
+      "loss": 0.6264,
+      "step": 524
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8614921463468003e-05,
+      "loss": 0.4414,
+      "step": 525
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8608824357010526e-05,
+      "loss": 0.637,
+      "step": 526
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.860271486337033e-05,
+      "loss": 0.7036,
+      "step": 527
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8596592991338334e-05,
+      "loss": 0.8247,
+      "step": 528
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8590458749723264e-05,
+      "loss": 0.7317,
+      "step": 529
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.858431214735163e-05,
+      "loss": 0.629,
+      "step": 530
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8578153193067746e-05,
+      "loss": 0.7063,
+      "step": 531
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.857198189573369e-05,
+      "loss": 0.6376,
+      "step": 532
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.85657982642293e-05,
+      "loss": 0.644,
+      "step": 533
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.855960230745217e-05,
+      "loss": 0.68,
+      "step": 534
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8553394034317622e-05,
+      "loss": 0.6674,
+      "step": 535
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8547173453758698e-05,
+      "loss": 0.7108,
+      "step": 536
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8540940574726153e-05,
+      "loss": 0.5509,
+      "step": 537
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8534695406188436e-05,
+      "loss": 0.6514,
+      "step": 538
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.852843795713168e-05,
+      "loss": 0.5169,
+      "step": 539
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8522168236559693e-05,
+      "loss": 0.5052,
+      "step": 540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8515886253493934e-05,
+      "loss": 0.5791,
+      "step": 541
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8509592016973505e-05,
+      "loss": 0.6494,
+      "step": 542
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8503285536055146e-05,
+      "loss": 0.6,
+      "step": 543
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8496966819813214e-05,
+      "loss": 0.5233,
+      "step": 544
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8490635877339665e-05,
+      "loss": 0.6375,
+      "step": 545
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.848429271774406e-05,
+      "loss": 0.5804,
+      "step": 546
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.847793735015352e-05,
+      "loss": 0.6032,
+      "step": 547
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.847156978371275e-05,
+      "loss": 0.519,
+      "step": 548
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8465190027584007e-05,
+      "loss": 0.5973,
+      "step": 549
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8458798090947064e-05,
+      "loss": 0.6347,
+      "step": 550
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8452393982999256e-05,
+      "loss": 0.5399,
+      "step": 551
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8445977712955404e-05,
+      "loss": 0.5733,
+      "step": 552
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8439549290047835e-05,
+      "loss": 0.5434,
+      "step": 553
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8433108723526375e-05,
+      "loss": 0.5977,
+      "step": 554
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.842665602265831e-05,
+      "loss": 0.6636,
+      "step": 555
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8420191196728388e-05,
+      "loss": 0.6674,
+      "step": 556
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8413714255038813e-05,
+      "loss": 0.5276,
+      "step": 557
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.840722520690921e-05,
+      "loss": 0.6318,
+      "step": 558
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8400724061676627e-05,
+      "loss": 0.6402,
+      "step": 559
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8394210828695526e-05,
+      "loss": 0.6135,
+      "step": 560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8387685517337752e-05,
+      "loss": 0.6811,
+      "step": 561
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8381148136992536e-05,
+      "loss": 0.66,
+      "step": 562
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.837459869706648e-05,
+      "loss": 0.6377,
+      "step": 563
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8368037206983514e-05,
+      "loss": 0.7346,
+      "step": 564
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.836146367618494e-05,
+      "loss": 0.6088,
+      "step": 565
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.5701,
+      "step": 566
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8348280530292712e-05,
+      "loss": 0.7209,
+      "step": 567
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8341670934168205e-05,
+      "loss": 0.4013,
+      "step": 568
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8335049335266348e-05,
+      "loss": 0.599,
+      "step": 569
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8328415743114913e-05,
+      "loss": 0.6606,
+      "step": 570
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.832177016725894e-05,
+      "loss": 0.6543,
+      "step": 571
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8315112617260708e-05,
+      "loss": 0.6078,
+      "step": 572
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8308443102699713e-05,
+      "loss": 0.5323,
+      "step": 573
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8301761633172684e-05,
+      "loss": 0.5464,
+      "step": 574
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.829506821829355e-05,
+      "loss": 0.5299,
+      "step": 575
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8288362867693414e-05,
+      "loss": 0.6927,
+      "step": 576
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8281645591020567e-05,
+      "loss": 0.5739,
+      "step": 577
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8274916397940457e-05,
+      "loss": 0.6154,
+      "step": 578
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8268175298135672e-05,
+      "loss": 0.6181,
+      "step": 579
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.826142230130594e-05,
+      "loss": 0.7922,
+      "step": 580
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.825465741716811e-05,
+      "loss": 0.636,
+      "step": 581
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8247880655456125e-05,
+      "loss": 0.6143,
+      "step": 582
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8241092025921026e-05,
+      "loss": 0.6571,
+      "step": 583
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.823429153833093e-05,
+      "loss": 0.6779,
+      "step": 584
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8227479202471016e-05,
+      "loss": 0.6928,
+      "step": 585
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8220655028143507e-05,
+      "loss": 0.5443,
+      "step": 586
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8213819025167667e-05,
+      "loss": 0.4681,
+      "step": 587
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8206971203379774e-05,
+      "loss": 0.4047,
+      "step": 588
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8200111572633123e-05,
+      "loss": 0.6561,
+      "step": 589
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8193240142797988e-05,
+      "loss": 0.5573,
+      "step": 590
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.818635692376163e-05,
+      "loss": 0.6488,
+      "step": 591
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8179461925428256e-05,
+      "loss": 0.6437,
+      "step": 592
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8172555157719052e-05,
+      "loss": 0.5807,
+      "step": 593
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.816563663057211e-05,
+      "loss": 0.2293,
+      "step": 594
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8158706353942464e-05,
+      "loss": 0.6491,
+      "step": 595
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8151764337802034e-05,
+      "loss": 0.7113,
+      "step": 596
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8144810592139656e-05,
+      "loss": 0.7301,
+      "step": 597
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8137845126961024e-05,
+      "loss": 0.66,
+      "step": 598
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.81308679522887e-05,
+      "loss": 0.6379,
+      "step": 599
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.81238790781621e-05,
+      "loss": 0.6526,
+      "step": 600
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8116878514637465e-05,
+      "loss": 0.5434,
+      "step": 601
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.810986627178787e-05,
+      "loss": 0.6071,
+      "step": 602
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8102842359703177e-05,
+      "loss": 0.7042,
+      "step": 603
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.8095806788490055e-05,
+      "loss": 0.6215,
+      "step": 604
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.808875956827194e-05,
+      "loss": 0.5539,
+      "step": 605
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8081700709189038e-05,
+      "loss": 0.7517,
+      "step": 606
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.807463022139829e-05,
+      "loss": 0.632,
+      "step": 607
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8067548115073377e-05,
+      "loss": 0.5787,
+      "step": 608
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8060454400404697e-05,
+      "loss": 0.6248,
+      "step": 609
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.805334908759935e-05,
+      "loss": 0.6031,
+      "step": 610
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.8046232186881134e-05,
+      "loss": 0.7072,
+      "step": 611
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8039103708490503e-05,
+      "loss": 0.5797,
+      "step": 612
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.803196366268458e-05,
+      "loss": 0.6342,
+      "step": 613
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.8024812059737134e-05,
+      "loss": 0.6281,
+      "step": 614
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.801764890993856e-05,
+      "loss": 0.4961,
+      "step": 615
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.801047422359587e-05,
+      "loss": 0.5005,
+      "step": 616
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.8003288011032677e-05,
+      "loss": 0.6398,
+      "step": 617
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.799609028258917e-05,
+      "loss": 0.7163,
+      "step": 618
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7988881048622118e-05,
+      "loss": 0.6778,
+      "step": 619
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7981660319504844e-05,
+      "loss": 0.5592,
+      "step": 620
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.797442810562721e-05,
+      "loss": 0.6343,
+      "step": 621
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.79671844173956e-05,
+      "loss": 0.5466,
+      "step": 622
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7959929265232907e-05,
+      "loss": 0.6554,
+      "step": 623
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.795266265957853e-05,
+      "loss": 0.7094,
+      "step": 624
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7945384610888343e-05,
+      "loss": 0.5963,
+      "step": 625
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7938095129634672e-05,
+      "loss": 0.5226,
+      "step": 626
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.793079422630632e-05,
+      "loss": 0.3132,
+      "step": 627
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7923481911408505e-05,
+      "loss": 0.7115,
+      "step": 628
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7916158195462866e-05,
+      "loss": 0.6871,
+      "step": 629
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.790882308900746e-05,
+      "loss": 0.6074,
+      "step": 630
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7901476602596714e-05,
+      "loss": 0.6887,
+      "step": 631
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7894118746801453e-05,
+      "loss": 0.6633,
+      "step": 632
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.788674953220884e-05,
+      "loss": 0.7239,
+      "step": 633
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7879368969422394e-05,
+      "loss": 0.6546,
+      "step": 634
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.787197706906196e-05,
+      "loss": 0.6722,
+      "step": 635
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7864573841763695e-05,
+      "loss": 0.424,
+      "step": 636
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7857159298180057e-05,
+      "loss": 0.678,
+      "step": 637
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7849733448979788e-05,
+      "loss": 0.6043,
+      "step": 638
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7842296304847892e-05,
+      "loss": 0.5266,
+      "step": 639
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.783484787648563e-05,
+      "loss": 0.5972,
+      "step": 640
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7827388174610498e-05,
+      "loss": 0.5301,
+      "step": 641
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7819917209956213e-05,
+      "loss": 0.4396,
+      "step": 642
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7812434993272703e-05,
+      "loss": 0.6473,
+      "step": 643
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.780494153532607e-05,
+      "loss": 0.4304,
+      "step": 644
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.779743684689862e-05,
+      "loss": 0.5586,
+      "step": 645
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.778992093878879e-05,
+      "loss": 0.517,
+      "step": 646
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.778239382181117e-05,
+      "loss": 0.6408,
+      "step": 647
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7774855506796497e-05,
+      "loss": 0.3902,
+      "step": 648
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7767306004591588e-05,
+      "loss": 0.6471,
+      "step": 649
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.775974532605938e-05,
+      "loss": 0.4758,
+      "step": 650
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.775217348207888e-05,
+      "loss": 0.4679,
+      "step": 651
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7744590483545176e-05,
+      "loss": 0.6389,
+      "step": 652
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.773699634136939e-05,
+      "loss": 0.6764,
+      "step": 653
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.772939106647869e-05,
+      "loss": 0.7475,
+      "step": 654
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7721774669816252e-05,
+      "loss": 0.5052,
+      "step": 655
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7714147162341263e-05,
+      "loss": 0.5234,
+      "step": 656
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7706508555028895e-05,
+      "loss": 0.6174,
+      "step": 657
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7698858858870293e-05,
+      "loss": 0.5672,
+      "step": 658
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.769119808487255e-05,
+      "loss": 0.5183,
+      "step": 659
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.768352624405872e-05,
+      "loss": 0.5296,
+      "step": 660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7675843347467747e-05,
+      "loss": 0.494,
+      "step": 661
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7668149406154513e-05,
+      "loss": 0.4988,
+      "step": 662
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.3852,
+      "step": 663
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.765272843366019e-05,
+      "loss": 0.5114,
+      "step": 664
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.764500142466824e-05,
+      "loss": 0.6326,
+      "step": 665
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7637263415332272e-05,
+      "loss": 0.5655,
+      "step": 666
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.762951441678646e-05,
+      "loss": 0.5561,
+      "step": 667
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7621754440180792e-05,
+      "loss": 0.6376,
+      "step": 668
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7613983496681043e-05,
+      "loss": 0.5164,
+      "step": 669
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.760620159746878e-05,
+      "loss": 0.4959,
+      "step": 670
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7598408753741327e-05,
+      "loss": 0.6437,
+      "step": 671
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7590604976711755e-05,
+      "loss": 0.4671,
+      "step": 672
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7582790277608875e-05,
+      "loss": 0.7075,
+      "step": 673
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.75749646676772e-05,
+      "loss": 0.6085,
+      "step": 674
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7567128158176955e-05,
+      "loss": 0.575,
+      "step": 675
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.755928076038404e-05,
+      "loss": 0.7392,
+      "step": 676
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7551422485590032e-05,
+      "loss": 0.6181,
+      "step": 677
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7543553345102152e-05,
+      "loss": 0.6711,
+      "step": 678
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7535673350243246e-05,
+      "loss": 0.5902,
+      "step": 679
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7527782512351804e-05,
+      "loss": 0.4983,
+      "step": 680
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7519880842781892e-05,
+      "loss": 0.5476,
+      "step": 681
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7511968352903176e-05,
+      "loss": 0.6232,
+      "step": 682
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7504045054100886e-05,
+      "loss": 0.5677,
+      "step": 683
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.749611095777581e-05,
+      "loss": 0.4497,
+      "step": 684
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.748816607534426e-05,
+      "loss": 0.5117,
+      "step": 685
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7480210418238086e-05,
+      "loss": 0.5644,
+      "step": 686
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.747224399790463e-05,
+      "loss": 0.3669,
+      "step": 687
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.746426682580672e-05,
+      "loss": 0.6217,
+      "step": 688
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7456278913422662e-05,
+      "loss": 0.5745,
+      "step": 689
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.744828027224621e-05,
+      "loss": 0.3948,
+      "step": 690
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7440270913786564e-05,
+      "loss": 0.5945,
+      "step": 691
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.743225084956833e-05,
+      "loss": 0.5987,
+      "step": 692
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7424220091131536e-05,
+      "loss": 0.6302,
+      "step": 693
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7416178650031584e-05,
+      "loss": 0.6167,
+      "step": 694
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7408126537839255e-05,
+      "loss": 0.5787,
+      "step": 695
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.740006376614068e-05,
+      "loss": 0.5985,
+      "step": 696
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7391990346537328e-05,
+      "loss": 0.5818,
+      "step": 697
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7383906290645998e-05,
+      "loss": 0.6297,
+      "step": 698
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.737581161009878e-05,
+      "loss": 0.5023,
+      "step": 699
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7367706316543063e-05,
+      "loss": 0.6744,
+      "step": 700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7359590421641503e-05,
+      "loss": 0.6332,
+      "step": 701
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7351463937072008e-05,
+      "loss": 0.485,
+      "step": 702
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.734332687452772e-05,
+      "loss": 0.6149,
+      "step": 703
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7335179245717013e-05,
+      "loss": 0.6471,
+      "step": 704
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7327021062363457e-05,
+      "loss": 0.5231,
+      "step": 705
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7318852336205808e-05,
+      "loss": 0.456,
+      "step": 706
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7310673078998e-05,
+      "loss": 0.7387,
+      "step": 707
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.730248330250911e-05,
+      "loss": 0.7149,
+      "step": 708
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7294283018523352e-05,
+      "loss": 0.6478,
+      "step": 709
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.728607223884007e-05,
+      "loss": 0.6104,
+      "step": 710
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7277850975273694e-05,
+      "loss": 0.6526,
+      "step": 711
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7269619239653756e-05,
+      "loss": 0.4988,
+      "step": 712
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.726137704382485e-05,
+      "loss": 0.6295,
+      "step": 713
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7253124399646606e-05,
+      "loss": 0.6281,
+      "step": 714
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7244861318993714e-05,
+      "loss": 0.6806,
+      "step": 715
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7236587813755863e-05,
+      "loss": 0.4878,
+      "step": 716
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.722830389583775e-05,
+      "loss": 0.6868,
+      "step": 717
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.722000957715905e-05,
+      "loss": 0.6038,
+      "step": 718
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.72117048696544e-05,
+      "loss": 0.6736,
+      "step": 719
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7203389785273402e-05,
+      "loss": 0.721,
+      "step": 720
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.719506433598057e-05,
+      "loss": 0.62,
+      "step": 721
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7186728533755344e-05,
+      "loss": 0.6702,
+      "step": 722
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7178382390592058e-05,
+      "loss": 0.5433,
+      "step": 723
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7170025918499917e-05,
+      "loss": 0.5629,
+      "step": 724
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7161659129503003e-05,
+      "loss": 0.6649,
+      "step": 725
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.715328203564023e-05,
+      "loss": 0.4641,
+      "step": 726
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7144894648965354e-05,
+      "loss": 0.466,
+      "step": 727
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7136496981546926e-05,
+      "loss": 0.5968,
+      "step": 728
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7128089045468294e-05,
+      "loss": 0.5398,
+      "step": 729
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7119670852827587e-05,
+      "loss": 0.5441,
+      "step": 730
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7111242415737694e-05,
+      "loss": 0.6339,
+      "step": 731
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.710280374632623e-05,
+      "loss": 0.5604,
+      "step": 732
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7094354856735548e-05,
+      "loss": 0.5724,
+      "step": 733
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.70858957591227e-05,
+      "loss": 0.5173,
+      "step": 734
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7077426465659434e-05,
+      "loss": 0.5786,
+      "step": 735
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.706894698853215e-05,
+      "loss": 0.4485,
+      "step": 736
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7060457339941928e-05,
+      "loss": 0.5571,
+      "step": 737
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.705195753210446e-05,
+      "loss": 0.7309,
+      "step": 738
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.704344757725007e-05,
+      "loss": 0.4672,
+      "step": 739
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.703492748762367e-05,
+      "loss": 0.6841,
+      "step": 740
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7026397275484773e-05,
+      "loss": 0.5493,
+      "step": 741
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7017856953107444e-05,
+      "loss": 0.5814,
+      "step": 742
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7009306532780298e-05,
+      "loss": 0.5604,
+      "step": 743
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.700074602680648e-05,
+      "loss": 0.5712,
+      "step": 744
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.699217544750365e-05,
+      "loss": 0.7139,
+      "step": 745
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.6983594807203955e-05,
+      "loss": 0.6758,
+      "step": 746
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.697500411825403e-05,
+      "loss": 0.6387,
+      "step": 747
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.696640339301496e-05,
+      "loss": 0.4082,
+      "step": 748
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.695779264386227e-05,
+      "loss": 0.414,
+      "step": 749
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.694917188318592e-05,
+      "loss": 0.52,
+      "step": 750
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.694054112339026e-05,
+      "loss": 0.5247,
+      "step": 751
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6931900376894045e-05,
+      "loss": 0.62,
+      "step": 752
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.692324965613038e-05,
+      "loss": 0.6151,
+      "step": 753
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.691458897354674e-05,
+      "loss": 0.4113,
+      "step": 754
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.690591834160492e-05,
+      "loss": 0.5214,
+      "step": 755
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6897237772781046e-05,
+      "loss": 0.5921,
+      "step": 756
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6888547279565523e-05,
+      "loss": 0.601,
+      "step": 757
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6879846874463052e-05,
+      "loss": 0.5031,
+      "step": 758
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.6871136569992587e-05,
+      "loss": 0.5104,
+      "step": 759
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.5324,
+      "step": 760
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6853686313094724e-05,
+      "loss": 0.5156,
+      "step": 761
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6844946385776386e-05,
+      "loss": 0.4359,
+      "step": 762
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.683619660930815e-05,
+      "loss": 0.4713,
+      "step": 763
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.682743699628001e-05,
+      "loss": 0.5799,
+      "step": 764
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.681866755929612e-05,
+      "loss": 0.6393,
+      "step": 765
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.680988831097477e-05,
+      "loss": 0.6812,
+      "step": 766
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.680109926394836e-05,
+      "loss": 0.5153,
+      "step": 767
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6792300430863395e-05,
+      "loss": 0.6225,
+      "step": 768
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.678349182438046e-05,
+      "loss": 0.4634,
+      "step": 769
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.6774673457174205e-05,
+      "loss": 0.7231,
+      "step": 770
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.676584534193332e-05,
+      "loss": 0.4919,
+      "step": 771
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.6757007491360528e-05,
+      "loss": 0.5418,
+      "step": 772
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.6748159918172552e-05,
+      "loss": 0.4982,
+      "step": 773
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.673930263510011e-05,
+      "loss": 0.4644,
+      "step": 774
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.673043565488789e-05,
+      "loss": 0.6691,
+      "step": 775
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.6721558990294538e-05,
+      "loss": 0.5708,
+      "step": 776
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.671267265409262e-05,
+      "loss": 0.598,
+      "step": 777
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.6703776659068644e-05,
+      "loss": 0.5796,
+      "step": 778
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.6694871018022983e-05,
+      "loss": 0.5678,
+      "step": 779
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.668595574376992e-05,
+      "loss": 0.603,
+      "step": 780
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6677030849137585e-05,
+      "loss": 0.6237,
+      "step": 781
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.666809634696795e-05,
+      "loss": 0.5488,
+      "step": 782
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.665915225011681e-05,
+      "loss": 0.7365,
+      "step": 783
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.665019857145378e-05,
+      "loss": 0.6589,
+      "step": 784
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6641235323862236e-05,
+      "loss": 0.4806,
+      "step": 785
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.663226252023935e-05,
+      "loss": 0.5409,
+      "step": 786
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6623280173496026e-05,
+      "loss": 0.6228,
+      "step": 787
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6614288296556913e-05,
+      "loss": 0.5673,
+      "step": 788
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6605286902360357e-05,
+      "loss": 0.6683,
+      "step": 789
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.659627600385841e-05,
+      "loss": 0.5843,
+      "step": 790
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6587255614016802e-05,
+      "loss": 0.6697,
+      "step": 791
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6578225745814907e-05,
+      "loss": 0.6133,
+      "step": 792
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.656918641224575e-05,
+      "loss": 0.6483,
+      "step": 793
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6560137626315964e-05,
+      "loss": 0.6758,
+      "step": 794
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6551079401045798e-05,
+      "loss": 0.4785,
+      "step": 795
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6542011749469072e-05,
+      "loss": 0.6114,
+      "step": 796
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.653293468463317e-05,
+      "loss": 0.6042,
+      "step": 797
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6523848219599025e-05,
+      "loss": 0.7,
+      "step": 798
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6514752367441095e-05,
+      "loss": 0.436,
+      "step": 799
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6505647141247337e-05,
+      "loss": 0.5905,
+      "step": 800
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6496532554119214e-05,
+      "loss": 0.5829,
+      "step": 801
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6487408619171638e-05,
+      "loss": 0.6176,
+      "step": 802
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6478275349532987e-05,
+      "loss": 0.6063,
+      "step": 803
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6469132758345063e-05,
+      "loss": 0.6487,
+      "step": 804
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.645998085876308e-05,
+      "loss": 0.6382,
+      "step": 805
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6450819663955653e-05,
+      "loss": 0.5439,
+      "step": 806
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6441649187104763e-05,
+      "loss": 0.6733,
+      "step": 807
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6432469441405753e-05,
+      "loss": 0.5441,
+      "step": 808
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6423280440067302e-05,
+      "loss": 0.4475,
+      "step": 809
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6414082196311402e-05,
+      "loss": 0.6267,
+      "step": 810
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.640487472337335e-05,
+      "loss": 0.4288,
+      "step": 811
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6395658034501717e-05,
+      "loss": 0.4311,
+      "step": 812
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6386432142958342e-05,
+      "loss": 0.5704,
+      "step": 813
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.63771970620183e-05,
+      "loss": 0.6923,
+      "step": 814
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6367952804969896e-05,
+      "loss": 0.7154,
+      "step": 815
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6358699385114626e-05,
+      "loss": 0.6914,
+      "step": 816
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6349436815767174e-05,
+      "loss": 0.6475,
+      "step": 817
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6340165110255403e-05,
+      "loss": 0.6457,
+      "step": 818
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.63308842819203e-05,
+      "loss": 0.4517,
+      "step": 819
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6321594344115998e-05,
+      "loss": 0.4131,
+      "step": 820
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.631229531020973e-05,
+      "loss": 0.4655,
+      "step": 821
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.630298719358181e-05,
+      "loss": 0.3567,
+      "step": 822
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.629367000762563e-05,
+      "loss": 0.6025,
+      "step": 823
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.628434376574763e-05,
+      "loss": 0.614,
+      "step": 824
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6275008481367287e-05,
+      "loss": 0.6822,
+      "step": 825
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6265664167917076e-05,
+      "loss": 0.6269,
+      "step": 826
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6256310838842467e-05,
+      "loss": 0.6399,
+      "step": 827
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6246948507601915e-05,
+      "loss": 0.5377,
+      "step": 828
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.623757718766681e-05,
+      "loss": 0.5525,
+      "step": 829
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.622819689252149e-05,
+      "loss": 0.6133,
+      "step": 830
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6218807635663204e-05,
+      "loss": 0.4339,
+      "step": 831
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6209409430602086e-05,
+      "loss": 0.6451,
+      "step": 832
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.620000229086116e-05,
+      "loss": 0.5931,
+      "step": 833
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6190586229976305e-05,
+      "loss": 0.5931,
+      "step": 834
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6181161261496215e-05,
+      "loss": 0.5872,
+      "step": 835
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6171727398982432e-05,
+      "loss": 0.4768,
+      "step": 836
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6162284656009276e-05,
+      "loss": 0.6003,
+      "step": 837
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6152833046163844e-05,
+      "loss": 0.5079,
+      "step": 838
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6143372583046007e-05,
+      "loss": 0.4566,
+      "step": 839
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6133903280268365e-05,
+      "loss": 0.5622,
+      "step": 840
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.612442515145623e-05,
+      "loss": 0.588,
+      "step": 841
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6114938210247622e-05,
+      "loss": 0.638,
+      "step": 842
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.610544247029325e-05,
+      "loss": 0.4686,
+      "step": 843
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6095937945256464e-05,
+      "loss": 0.5357,
+      "step": 844
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6086424648813273e-05,
+      "loss": 0.612,
+      "step": 845
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.607690259465229e-05,
+      "loss": 0.687,
+      "step": 846
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6067371796474745e-05,
+      "loss": 0.5572,
+      "step": 847
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.605783226799444e-05,
+      "loss": 0.5301,
+      "step": 848
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.604828402293774e-05,
+      "loss": 0.527,
+      "step": 849
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.603872707504356e-05,
+      "loss": 0.723,
+      "step": 850
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6029161438063328e-05,
+      "loss": 0.5996,
+      "step": 851
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6019587125760977e-05,
+      "loss": 0.5697,
+      "step": 852
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6010004151912928e-05,
+      "loss": 0.5976,
+      "step": 853
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.600041253030806e-05,
+      "loss": 0.4917,
+      "step": 854
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5990812274747695e-05,
+      "loss": 0.5797,
+      "step": 855
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5981203399045577e-05,
+      "loss": 0.6318,
+      "step": 856
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.5114,
+      "step": 857
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5961959842533083e-05,
+      "loss": 0.5438,
+      "step": 858
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5952325189412134e-05,
+      "loss": 0.579,
+      "step": 859
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.594268197152826e-05,
+      "loss": 0.5809,
+      "step": 860
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.593303020275702e-05,
+      "loss": 0.5692,
+      "step": 861
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5923369896986294e-05,
+      "loss": 0.5753,
+      "step": 862
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5913701068116224e-05,
+      "loss": 0.6734,
+      "step": 863
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5904023730059227e-05,
+      "loss": 0.5321,
+      "step": 864
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.589433789673997e-05,
+      "loss": 0.6851,
+      "step": 865
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5884643582095334e-05,
+      "loss": 0.6031,
+      "step": 866
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5874940800074402e-05,
+      "loss": 0.5726,
+      "step": 867
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5865229564638452e-05,
+      "loss": 0.5277,
+      "step": 868
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5855509889760917e-05,
+      "loss": 0.5252,
+      "step": 869
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5845781789427376e-05,
+      "loss": 0.6031,
+      "step": 870
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.583604527763553e-05,
+      "loss": 0.7062,
+      "step": 871
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5826300368395186e-05,
+      "loss": 0.5091,
+      "step": 872
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5816547075728227e-05,
+      "loss": 0.4265,
+      "step": 873
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.580678541366861e-05,
+      "loss": 0.5656,
+      "step": 874
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.579701539626232e-05,
+      "loss": 0.6723,
+      "step": 875
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5787237037567383e-05,
+      "loss": 0.5724,
+      "step": 876
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.577745035165381e-05,
+      "loss": 0.5581,
+      "step": 877
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.576765535260361e-05,
+      "loss": 0.5313,
+      "step": 878
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5757852054510733e-05,
+      "loss": 0.385,
+      "step": 879
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5748040471481088e-05,
+      "loss": 0.5971,
+      "step": 880
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5738220617632504e-05,
+      "loss": 0.665,
+      "step": 881
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.57283925070947e-05,
+      "loss": 0.4562,
+      "step": 882
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5718556154009284e-05,
+      "loss": 0.5802,
+      "step": 883
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.570871157252972e-05,
+      "loss": 0.5458,
+      "step": 884
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.569885877682132e-05,
+      "loss": 0.5876,
+      "step": 885
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5688997781061204e-05,
+      "loss": 0.584,
+      "step": 886
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.56791285994383e-05,
+      "loss": 0.6241,
+      "step": 887
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5669251246153302e-05,
+      "loss": 0.5952,
+      "step": 888
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5659365735418683e-05,
+      "loss": 0.5787,
+      "step": 889
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.564947208145863e-05,
+      "loss": 0.5855,
+      "step": 890
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5639570298509067e-05,
+      "loss": 0.581,
+      "step": 891
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5629660400817604e-05,
+      "loss": 0.6515,
+      "step": 892
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5619742402643532e-05,
+      "loss": 0.6329,
+      "step": 893
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5609816318257788e-05,
+      "loss": 0.6281,
+      "step": 894
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5599882161942966e-05,
+      "loss": 0.6075,
+      "step": 895
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.558993994799325e-05,
+      "loss": 0.5471,
+      "step": 896
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5579989690714423e-05,
+      "loss": 0.5716,
+      "step": 897
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.557003140442386e-05,
+      "loss": 0.526,
+      "step": 898
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5560065103450473e-05,
+      "loss": 0.4678,
+      "step": 899
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.55500908021347e-05,
+      "loss": 0.5563,
+      "step": 900
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5540108514828508e-05,
+      "loss": 0.6207,
+      "step": 901
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.553011825589535e-05,
+      "loss": 0.6736,
+      "step": 902
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5520120039710137e-05,
+      "loss": 0.6991,
+      "step": 903
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.551011388065925e-05,
+      "loss": 0.4985,
+      "step": 904
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5500099793140474e-05,
+      "loss": 0.6464,
+      "step": 905
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5490077791563023e-05,
+      "loss": 0.5225,
+      "step": 906
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5480047890347494e-05,
+      "loss": 0.6338,
+      "step": 907
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5470010103925843e-05,
+      "loss": 0.5785,
+      "step": 908
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.545996444674138e-05,
+      "loss": 0.6029,
+      "step": 909
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.544991093324874e-05,
+      "loss": 0.6394,
+      "step": 910
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.5439849577913858e-05,
+      "loss": 0.6181,
+      "step": 911
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.542978039521395e-05,
+      "loss": 0.5824,
+      "step": 912
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5419703399637502e-05,
+      "loss": 0.4519,
+      "step": 913
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.540961860568424e-05,
+      "loss": 0.6613,
+      "step": 914
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.539952602786511e-05,
+      "loss": 0.3589,
+      "step": 915
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5389425680702257e-05,
+      "loss": 0.5614,
+      "step": 916
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5379317578729003e-05,
+      "loss": 0.4236,
+      "step": 917
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.536920173648984e-05,
+      "loss": 0.4812,
+      "step": 918
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5359078168540383e-05,
+      "loss": 0.5523,
+      "step": 919
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.534894688944737e-05,
+      "loss": 0.4613,
+      "step": 920
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.533880791378864e-05,
+      "loss": 0.6408,
+      "step": 921
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.532866125615309e-05,
+      "loss": 0.5459,
+      "step": 922
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5318506931140692e-05,
+      "loss": 0.5892,
+      "step": 923
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5308344953362434e-05,
+      "loss": 0.6634,
+      "step": 924
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.5298175337440322e-05,
+      "loss": 0.622,
+      "step": 925
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.528799809800735e-05,
+      "loss": 0.5146,
+      "step": 926
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.5277813249707488e-05,
+      "loss": 0.5903,
+      "step": 927
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.526762080719565e-05,
+      "loss": 0.3817,
+      "step": 928
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.5257420785137668e-05,
+      "loss": 0.5223,
+      "step": 929
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.52472131982103e-05,
+      "loss": 0.6061,
+      "step": 930
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5236998061101172e-05,
+      "loss": 0.5549,
+      "step": 931
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.522677538850878e-05,
+      "loss": 0.5666,
+      "step": 932
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.5216545195142463e-05,
+      "loss": 0.5621,
+      "step": 933
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.520630749572238e-05,
+      "loss": 0.545,
+      "step": 934
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5196062304979498e-05,
+      "loss": 0.5369,
+      "step": 935
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.5185809637655548e-05,
+      "loss": 0.5619,
+      "step": 936
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5175549508503036e-05,
+      "loss": 0.4902,
+      "step": 937
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5165281932285192e-05,
+      "loss": 0.5966,
+      "step": 938
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5155006923775966e-05,
+      "loss": 0.4875,
+      "step": 939
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5144724497760004e-05,
+      "loss": 0.3757,
+      "step": 940
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5134434669032624e-05,
+      "loss": 0.5668,
+      "step": 941
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5124137452399797e-05,
+      "loss": 0.5654,
+      "step": 942
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5113832862678115e-05,
+      "loss": 0.5729,
+      "step": 943
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5103520914694796e-05,
+      "loss": 0.6707,
+      "step": 944
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5093201623287631e-05,
+      "loss": 0.6337,
+      "step": 945
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5082875003304986e-05,
+      "loss": 0.5992,
+      "step": 946
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5072541069605767e-05,
+      "loss": 0.5241,
+      "step": 947
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5062199837059406e-05,
+      "loss": 0.5769,
+      "step": 948
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5051851320545834e-05,
+      "loss": 0.5376,
+      "step": 949
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5041495534955468e-05,
+      "loss": 0.6415,
+      "step": 950
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5031132495189186e-05,
+      "loss": 0.6411,
+      "step": 951
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5020762216158295e-05,
+      "loss": 0.5976,
+      "step": 952
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5010384712784524e-05,
+      "loss": 0.6006,
+      "step": 953
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5465,
+      "step": 954
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.498960809274722e-05,
+      "loss": 0.63,
+      "step": 955
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.497920900597903e-05,
+      "loss": 0.6233,
+      "step": 956
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4968802754658614e-05,
+      "loss": 0.4547,
+      "step": 957
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4958389353759452e-05,
+      "loss": 0.4063,
+      "step": 958
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4947968818265332e-05,
+      "loss": 0.6557,
+      "step": 959
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4937541163170293e-05,
+      "loss": 0.4813,
+      "step": 960
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4927106403478618e-05,
+      "loss": 0.4527,
+      "step": 961
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4916664554204818e-05,
+      "loss": 0.6254,
+      "step": 962
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4906215630373606e-05,
+      "loss": 0.5586,
+      "step": 963
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4895759647019871e-05,
+      "loss": 0.6641,
+      "step": 964
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4885296619188659e-05,
+      "loss": 0.5126,
+      "step": 965
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4874826561935157e-05,
+      "loss": 0.5767,
+      "step": 966
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.486434949032466e-05,
+      "loss": 0.5717,
+      "step": 967
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4853865419432565e-05,
+      "loss": 0.7174,
+      "step": 968
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4843374364344332e-05,
+      "loss": 0.5542,
+      "step": 969
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4832876340155475e-05,
+      "loss": 0.5526,
+      "step": 970
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4822371361971532e-05,
+      "loss": 0.4434,
+      "step": 971
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4811859444908053e-05,
+      "loss": 0.4937,
+      "step": 972
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4801340604090567e-05,
+      "loss": 0.6939,
+      "step": 973
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4790814854654562e-05,
+      "loss": 0.468,
+      "step": 974
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4780282211745481e-05,
+      "loss": 0.6191,
+      "step": 975
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4769742690518672e-05,
+      "loss": 0.4913,
+      "step": 976
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4759196306139388e-05,
+      "loss": 0.654,
+      "step": 977
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4748643073782754e-05,
+      "loss": 0.6237,
+      "step": 978
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4738083008633743e-05,
+      "loss": 0.5592,
+      "step": 979
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4727516125887176e-05,
+      "loss": 0.6467,
+      "step": 980
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.4716942440747663e-05,
+      "loss": 0.545,
+      "step": 981
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4706361968429618e-05,
+      "loss": 0.5874,
+      "step": 982
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.469577472415722e-05,
+      "loss": 0.4316,
+      "step": 983
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4685180723164377e-05,
+      "loss": 0.4452,
+      "step": 984
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4674579980694737e-05,
+      "loss": 0.6062,
+      "step": 985
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.466397251200164e-05,
+      "loss": 0.3315,
+      "step": 986
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4653358332348103e-05,
+      "loss": 0.6467,
+      "step": 987
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4642737457006804e-05,
+      "loss": 0.6832,
+      "step": 988
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.463210990126005e-05,
+      "loss": 0.6253,
+      "step": 989
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4621475680399771e-05,
+      "loss": 0.5803,
+      "step": 990
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4610834809727472e-05,
+      "loss": 0.4916,
+      "step": 991
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4600187304554236e-05,
+      "loss": 0.6283,
+      "step": 992
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4589533180200695e-05,
+      "loss": 0.5811,
+      "step": 993
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4578872451996996e-05,
+      "loss": 0.5976,
+      "step": 994
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4568205135282796e-05,
+      "loss": 0.6799,
+      "step": 995
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4557531245407226e-05,
+      "loss": 0.5547,
+      "step": 996
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.454685079772888e-05,
+      "loss": 0.5082,
+      "step": 997
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4536163807615787e-05,
+      "loss": 0.6736,
+      "step": 998
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4525470290445392e-05,
+      "loss": 0.5667,
+      "step": 999
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4514770261604523e-05,
+      "loss": 0.524,
+      "step": 1000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4504063736489384e-05,
+      "loss": 0.5504,
+      "step": 1001
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4493350730505533e-05,
+      "loss": 0.6446,
+      "step": 1002
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4482631259067841e-05,
+      "loss": 0.5392,
+      "step": 1003
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4471905337600487e-05,
+      "loss": 0.618,
+      "step": 1004
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.446117298153693e-05,
+      "loss": 0.5335,
+      "step": 1005
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4450434206319898e-05,
+      "loss": 0.5046,
+      "step": 1006
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4439689027401336e-05,
+      "loss": 0.5239,
+      "step": 1007
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4428937460242417e-05,
+      "loss": 0.5543,
+      "step": 1008
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4418179520313506e-05,
+      "loss": 0.5396,
+      "step": 1009
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4407415223094133e-05,
+      "loss": 0.5419,
+      "step": 1010
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4396644584072972e-05,
+      "loss": 0.599,
+      "step": 1011
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4385867618747836e-05,
+      "loss": 0.5402,
+      "step": 1012
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4375084342625622e-05,
+      "loss": 0.5564,
+      "step": 1013
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4364294771222325e-05,
+      "loss": 0.5804,
+      "step": 1014
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4353498920062986e-05,
+      "loss": 0.6296,
+      "step": 1015
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4342696804681691e-05,
+      "loss": 0.5484,
+      "step": 1016
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4331888440621533e-05,
+      "loss": 0.644,
+      "step": 1017
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4321073843434596e-05,
+      "loss": 0.3443,
+      "step": 1018
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4310253028681942e-05,
+      "loss": 0.4906,
+      "step": 1019
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4299426011933568e-05,
+      "loss": 0.5833,
+      "step": 1020
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4288592808768397e-05,
+      "loss": 0.3796,
+      "step": 1021
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4277753434774264e-05,
+      "loss": 0.4984,
+      "step": 1022
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4266907905547868e-05,
+      "loss": 0.6374,
+      "step": 1023
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4256056236694775e-05,
+      "loss": 0.6261,
+      "step": 1024
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4245198443829384e-05,
+      "loss": 0.6,
+      "step": 1025
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4234334542574906e-05,
+      "loss": 0.6404,
+      "step": 1026
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4223464548563331e-05,
+      "loss": 0.6186,
+      "step": 1027
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4212588477435432e-05,
+      "loss": 0.4363,
+      "step": 1028
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4201706344840714e-05,
+      "loss": 0.5896,
+      "step": 1029
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4190818166437413e-05,
+      "loss": 0.5511,
+      "step": 1030
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4179923957892454e-05,
+      "loss": 0.373,
+      "step": 1031
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4169023734881452e-05,
+      "loss": 0.6509,
+      "step": 1032
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4158117513088665e-05,
+      "loss": 0.584,
+      "step": 1033
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4147205308206984e-05,
+      "loss": 0.6207,
+      "step": 1034
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4136287135937915e-05,
+      "loss": 0.6137,
+      "step": 1035
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4125363011991544e-05,
+      "loss": 0.472,
+      "step": 1036
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4114432952086526e-05,
+      "loss": 0.5613,
+      "step": 1037
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4103496971950055e-05,
+      "loss": 0.6316,
+      "step": 1038
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4092555087317838e-05,
+      "loss": 0.5986,
+      "step": 1039
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4081607313934092e-05,
+      "loss": 0.6294,
+      "step": 1040
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.407065366755149e-05,
+      "loss": 0.4889,
+      "step": 1041
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4059694163931172e-05,
+      "loss": 0.5906,
+      "step": 1042
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4048728818842691e-05,
+      "loss": 0.4613,
+      "step": 1043
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4037757648064019e-05,
+      "loss": 0.6113,
+      "step": 1044
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4026780667381499e-05,
+      "loss": 0.4912,
+      "step": 1045
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4015797892589839e-05,
+      "loss": 0.3306,
+      "step": 1046
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.4004809339492087e-05,
+      "loss": 0.6258,
+      "step": 1047
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.3993815023899601e-05,
+      "loss": 0.4053,
+      "step": 1048
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3982814961632032e-05,
+      "loss": 0.4688,
+      "step": 1049
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3971809168517298e-05,
+      "loss": 0.6402,
+      "step": 1050
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.3522,
+      "step": 1051
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3949780453099232e-05,
+      "loss": 0.5782,
+      "step": 1052
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3938757562492873e-05,
+      "loss": 0.5616,
+      "step": 1053
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3927729004433265e-05,
+      "loss": 0.4616,
+      "step": 1054
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3916694794789324e-05,
+      "loss": 0.6536,
+      "step": 1055
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.390565494943811e-05,
+      "loss": 0.6052,
+      "step": 1056
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3894609484264788e-05,
+      "loss": 0.497,
+      "step": 1057
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3883558415162601e-05,
+      "loss": 0.4954,
+      "step": 1058
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3872501758032864e-05,
+      "loss": 0.5681,
+      "step": 1059
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3861439528784933e-05,
+      "loss": 0.4587,
+      "step": 1060
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3850371743336176e-05,
+      "loss": 0.4638,
+      "step": 1061
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3839298417611964e-05,
+      "loss": 0.5201,
+      "step": 1062
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.382821956754563e-05,
+      "loss": 0.3686,
+      "step": 1063
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3817135209078468e-05,
+      "loss": 0.4927,
+      "step": 1064
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3806045358159684e-05,
+      "loss": 0.5031,
+      "step": 1065
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3794950030746397e-05,
+      "loss": 0.5013,
+      "step": 1066
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3783849242803604e-05,
+      "loss": 0.5856,
+      "step": 1067
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3772743010304155e-05,
+      "loss": 0.6463,
+      "step": 1068
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3761631349228738e-05,
+      "loss": 0.4364,
+      "step": 1069
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3750514275565859e-05,
+      "loss": 0.6676,
+      "step": 1070
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3739391805311795e-05,
+      "loss": 0.6009,
+      "step": 1071
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3728263954470602e-05,
+      "loss": 0.4388,
+      "step": 1072
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3717130739054076e-05,
+      "loss": 0.4673,
+      "step": 1073
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3705992175081727e-05,
+      "loss": 0.4962,
+      "step": 1074
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3694848278580764e-05,
+      "loss": 0.5359,
+      "step": 1075
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3683699065586069e-05,
+      "loss": 0.7059,
+      "step": 1076
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3672544552140177e-05,
+      "loss": 0.6291,
+      "step": 1077
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3661384754293242e-05,
+      "loss": 0.4821,
+      "step": 1078
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3650219688103026e-05,
+      "loss": 0.6477,
+      "step": 1079
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3639049369634878e-05,
+      "loss": 0.5408,
+      "step": 1080
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3627873814961691e-05,
+      "loss": 0.6388,
+      "step": 1081
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3616693040163904e-05,
+      "loss": 0.4882,
+      "step": 1082
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3605507061329466e-05,
+      "loss": 0.6368,
+      "step": 1083
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3594315894553802e-05,
+      "loss": 0.5642,
+      "step": 1084
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3583119555939816e-05,
+      "loss": 0.4508,
+      "step": 1085
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3571918061597851e-05,
+      "loss": 0.6223,
+      "step": 1086
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3560711427645663e-05,
+      "loss": 0.6491,
+      "step": 1087
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3549499670208406e-05,
+      "loss": 0.3899,
+      "step": 1088
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.353828280541861e-05,
+      "loss": 0.5473,
+      "step": 1089
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3527060849416144e-05,
+      "loss": 0.6865,
+      "step": 1090
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3515833818348216e-05,
+      "loss": 0.5683,
+      "step": 1091
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3504601728369327e-05,
+      "loss": 0.5772,
+      "step": 1092
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3493364595641261e-05,
+      "loss": 0.5816,
+      "step": 1093
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3482122436333052e-05,
+      "loss": 0.5944,
+      "step": 1094
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3470875266620978e-05,
+      "loss": 0.4408,
+      "step": 1095
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3459623102688517e-05,
+      "loss": 0.6293,
+      "step": 1096
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3448365960726335e-05,
+      "loss": 0.6056,
+      "step": 1097
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3437103856932266e-05,
+      "loss": 0.6129,
+      "step": 1098
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3425836807511277e-05,
+      "loss": 0.566,
+      "step": 1099
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3414564828675456e-05,
+      "loss": 0.4819,
+      "step": 1100
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3403287936643978e-05,
+      "loss": 0.6966,
+      "step": 1101
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3392006147643096e-05,
+      "loss": 0.549,
+      "step": 1102
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.3380719477906103e-05,
+      "loss": 0.5425,
+      "step": 1103
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.3369427943673313e-05,
+      "loss": 0.4527,
+      "step": 1104
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.3358131561192047e-05,
+      "loss": 0.5864,
+      "step": 1105
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.3346830346716595e-05,
+      "loss": 0.5049,
+      "step": 1106
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.3335524316508208e-05,
+      "loss": 0.6295,
+      "step": 1107
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3324213486835054e-05,
+      "loss": 0.6455,
+      "step": 1108
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3312897873972221e-05,
+      "loss": 0.4883,
+      "step": 1109
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.3301577494201665e-05,
+      "loss": 0.3801,
+      "step": 1110
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.3290252363812215e-05,
+      "loss": 0.3345,
+      "step": 1111
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.3278922499099525e-05,
+      "loss": 0.4302,
+      "step": 1112
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.326758791636607e-05,
+      "loss": 0.6332,
+      "step": 1113
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3256248631921102e-05,
+      "loss": 0.5015,
+      "step": 1114
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3244904662080654e-05,
+      "loss": 0.4885,
+      "step": 1115
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.3233556023167487e-05,
+      "loss": 0.5791,
+      "step": 1116
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.3222202731511085e-05,
+      "loss": 0.5796,
+      "step": 1117
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.321084480344763e-05,
+      "loss": 0.6251,
+      "step": 1118
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.3199482255319971e-05,
+      "loss": 0.5593,
+      "step": 1119
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.3188115103477605e-05,
+      "loss": 0.3469,
+      "step": 1120
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3176743364276661e-05,
+      "loss": 0.6471,
+      "step": 1121
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3165367054079851e-05,
+      "loss": 0.6481,
+      "step": 1122
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.315398618925649e-05,
+      "loss": 0.4445,
+      "step": 1123
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.3142600786182417e-05,
+      "loss": 0.4398,
+      "step": 1124
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.3131210861240027e-05,
+      "loss": 0.4497,
+      "step": 1125
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3119816430818206e-05,
+      "loss": 0.6565,
+      "step": 1126
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3108417511312331e-05,
+      "loss": 0.593,
+      "step": 1127
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.309701411912423e-05,
+      "loss": 0.6076,
+      "step": 1128
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3085606270662177e-05,
+      "loss": 0.6795,
+      "step": 1129
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3074193982340848e-05,
+      "loss": 0.5059,
+      "step": 1130
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3062777270581312e-05,
+      "loss": 0.6066,
+      "step": 1131
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.305135615181101e-05,
+      "loss": 0.518,
+      "step": 1132
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.303993064246371e-05,
+      "loss": 0.5848,
+      "step": 1133
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3028500758979507e-05,
+      "loss": 0.5826,
+      "step": 1134
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3017066517804793e-05,
+      "loss": 0.6788,
+      "step": 1135
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3005627935392216e-05,
+      "loss": 0.5021,
+      "step": 1136
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2994185028200685e-05,
+      "loss": 0.4304,
+      "step": 1137
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2982737812695326e-05,
+      "loss": 0.5105,
+      "step": 1138
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2971286305347463e-05,
+      "loss": 0.636,
+      "step": 1139
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2959830522634598e-05,
+      "loss": 0.6423,
+      "step": 1140
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2948370481040384e-05,
+      "loss": 0.4781,
+      "step": 1141
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2936906197054607e-05,
+      "loss": 0.5796,
+      "step": 1142
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2925437687173144e-05,
+      "loss": 0.5513,
+      "step": 1143
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2913964967897963e-05,
+      "loss": 0.5047,
+      "step": 1144
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2902488055737092e-05,
+      "loss": 0.5333,
+      "step": 1145
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2891006967204585e-05,
+      "loss": 0.4986,
+      "step": 1146
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2879521718820503e-05,
+      "loss": 0.376,
+      "step": 1147
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.5376,
+      "step": 1148
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2856538808607794e-05,
+      "loss": 0.5081,
+      "step": 1149
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2845041179849129e-05,
+      "loss": 0.5594,
+      "step": 1150
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2833539457378774e-05,
+      "loss": 0.603,
+      "step": 1151
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2822033657746478e-05,
+      "loss": 0.4481,
+      "step": 1152
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.2810523797507873e-05,
+      "loss": 0.4413,
+      "step": 1153
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.2799009893224412e-05,
+      "loss": 0.4706,
+      "step": 1154
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.278749196146339e-05,
+      "loss": 0.5075,
+      "step": 1155
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.2775970018797876e-05,
+      "loss": 0.5794,
+      "step": 1156
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.2764444081806728e-05,
+      "loss": 0.5095,
+      "step": 1157
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.2752914167074541e-05,
+      "loss": 0.6452,
+      "step": 1158
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2741380291191636e-05,
+      "loss": 0.5743,
+      "step": 1159
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2729842470754032e-05,
+      "loss": 0.6388,
+      "step": 1160
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.2718300722363431e-05,
+      "loss": 0.6026,
+      "step": 1161
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2706755062627174e-05,
+      "loss": 0.4887,
+      "step": 1162
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2695205508158242e-05,
+      "loss": 0.5335,
+      "step": 1163
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.2683652075575218e-05,
+      "loss": 0.4384,
+      "step": 1164
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.2672094781502254e-05,
+      "loss": 0.6537,
+      "step": 1165
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.2660533642569072e-05,
+      "loss": 0.6235,
+      "step": 1166
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.2648968675410929e-05,
+      "loss": 0.5705,
+      "step": 1167
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2637399896668572e-05,
+      "loss": 0.4101,
+      "step": 1168
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2625827322988246e-05,
+      "loss": 0.6652,
+      "step": 1169
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.2614250971021658e-05,
+      "loss": 0.4276,
+      "step": 1170
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.2602670857425943e-05,
+      "loss": 0.5768,
+      "step": 1171
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.2591086998863653e-05,
+      "loss": 0.5999,
+      "step": 1172
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.2579499412002729e-05,
+      "loss": 0.4782,
+      "step": 1173
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.256790811351648e-05,
+      "loss": 0.5482,
+      "step": 1174
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2556313120083547e-05,
+      "loss": 0.4902,
+      "step": 1175
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2544714448387894e-05,
+      "loss": 0.5143,
+      "step": 1176
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.2533112115118779e-05,
+      "loss": 0.4625,
+      "step": 1177
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.252150613697072e-05,
+      "loss": 0.4638,
+      "step": 1178
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.2509896530643488e-05,
+      "loss": 0.555,
+      "step": 1179
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2498283312842072e-05,
+      "loss": 0.4855,
+      "step": 1180
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2486666500276662e-05,
+      "loss": 0.6399,
+      "step": 1181
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2475046109662605e-05,
+      "loss": 0.5952,
+      "step": 1182
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2463422157720417e-05,
+      "loss": 0.4482,
+      "step": 1183
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2451794661175725e-05,
+      "loss": 0.5849,
+      "step": 1184
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2440163636759261e-05,
+      "loss": 0.6142,
+      "step": 1185
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.242852910120683e-05,
+      "loss": 0.5034,
+      "step": 1186
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.2416891071259295e-05,
+      "loss": 0.53,
+      "step": 1187
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.2405249563662539e-05,
+      "loss": 0.4903,
+      "step": 1188
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2393604595167455e-05,
+      "loss": 0.5959,
+      "step": 1189
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2381956182529917e-05,
+      "loss": 0.3078,
+      "step": 1190
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.2370304342510745e-05,
+      "loss": 0.5483,
+      "step": 1191
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.2358649091875707e-05,
+      "loss": 0.5277,
+      "step": 1192
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2346990447395462e-05,
+      "loss": 0.4175,
+      "step": 1193
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2335328425845565e-05,
+      "loss": 0.4407,
+      "step": 1194
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.232366304400642e-05,
+      "loss": 0.5847,
+      "step": 1195
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.2311994318663272e-05,
+      "loss": 0.5036,
+      "step": 1196
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.2300322266606176e-05,
+      "loss": 0.6688,
+      "step": 1197
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.2288646904629977e-05,
+      "loss": 0.5523,
+      "step": 1198
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.2276968249534275e-05,
+      "loss": 0.6669,
+      "step": 1199
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.2265286318123414e-05,
+      "loss": 0.4755,
+      "step": 1200
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.2253601127206451e-05,
+      "loss": 0.6048,
+      "step": 1201
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2241912693597133e-05,
+      "loss": 0.5688,
+      "step": 1202
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2230221034113874e-05,
+      "loss": 0.5279,
+      "step": 1203
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.2218526165579728e-05,
+      "loss": 0.5584,
+      "step": 1204
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.2206828104822363e-05,
+      "loss": 0.4203,
+      "step": 1205
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.2195126868674052e-05,
+      "loss": 0.6657,
+      "step": 1206
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.2183422473971622e-05,
+      "loss": 0.4489,
+      "step": 1207
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.2171714937556455e-05,
+      "loss": 0.5487,
+      "step": 1208
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2160004276274453e-05,
+      "loss": 0.54,
+      "step": 1209
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2148290506976012e-05,
+      "loss": 0.5327,
+      "step": 1210
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.2136573646516e-05,
+      "loss": 0.5687,
+      "step": 1211
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.2124853711753728e-05,
+      "loss": 0.643,
+      "step": 1212
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2113130719552943e-05,
+      "loss": 0.4619,
+      "step": 1213
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2101404686781778e-05,
+      "loss": 0.4984,
+      "step": 1214
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.2089675630312755e-05,
+      "loss": 0.5579,
+      "step": 1215
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.2077943567022728e-05,
+      "loss": 0.5684,
+      "step": 1216
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.2066208513792896e-05,
+      "loss": 0.5679,
+      "step": 1217
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2054470487508751e-05,
+      "loss": 0.5911,
+      "step": 1218
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2042729505060066e-05,
+      "loss": 0.62,
+      "step": 1219
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.2030985583340862e-05,
+      "loss": 0.5176,
+      "step": 1220
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.2019238739249396e-05,
+      "loss": 0.5863,
+      "step": 1221
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.2007488989688127e-05,
+      "loss": 0.5895,
+      "step": 1222
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.1995736351563697e-05,
+      "loss": 0.4758,
+      "step": 1223
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.1983980841786899e-05,
+      "loss": 0.5147,
+      "step": 1224
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.1972222477272663e-05,
+      "loss": 0.6247,
+      "step": 1225
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.1960461274940029e-05,
+      "loss": 0.5425,
+      "step": 1226
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.194869725171211e-05,
+      "loss": 0.5743,
+      "step": 1227
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.1936930424516082e-05,
+      "loss": 0.6772,
+      "step": 1228
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1925160810283169e-05,
+      "loss": 0.5935,
+      "step": 1229
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1913388425948584e-05,
+      "loss": 0.4599,
+      "step": 1230
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1901613288451546e-05,
+      "loss": 0.563,
+      "step": 1231
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1889835414735214e-05,
+      "loss": 0.3813,
+      "step": 1232
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1878054821746703e-05,
+      "loss": 0.4116,
+      "step": 1233
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1866271526437032e-05,
+      "loss": 0.4374,
+      "step": 1234
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1854485545761108e-05,
+      "loss": 0.5516,
+      "step": 1235
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.1842696896677708e-05,
+      "loss": 0.4742,
+      "step": 1236
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.1830905596149438e-05,
+      "loss": 0.5122,
+      "step": 1237
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1819111661142731e-05,
+      "loss": 0.6227,
+      "step": 1238
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1807315108627805e-05,
+      "loss": 0.469,
+      "step": 1239
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.179551595557864e-05,
+      "loss": 0.4747,
+      "step": 1240
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.1783714218972966e-05,
+      "loss": 0.5029,
+      "step": 1241
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.177190991579223e-05,
+      "loss": 0.5554,
+      "step": 1242
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.176010306302156e-05,
+      "loss": 0.6171,
+      "step": 1243
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.174829367764977e-05,
+      "loss": 0.4703,
+      "step": 1244
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.6055,
+      "step": 1245
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1724667377076235e-05,
+      "loss": 0.5743,
+      "step": 1246
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.1712850495870225e-05,
+      "loss": 0.6849,
+      "step": 1247
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.170103115005451e-05,
+      "loss": 0.3386,
+      "step": 1248
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.1689209356635878e-05,
+      "loss": 0.5057,
+      "step": 1249
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.167738513262462e-05,
+      "loss": 0.5793,
+      "step": 1250
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.1665558495034546e-05,
+      "loss": 0.5188,
+      "step": 1251
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1653729460882928e-05,
+      "loss": 0.6756,
+      "step": 1252
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1641898047190485e-05,
+      "loss": 0.5451,
+      "step": 1253
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.1630064270981366e-05,
+      "loss": 0.4272,
+      "step": 1254
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.1618228149283118e-05,
+      "loss": 0.5455,
+      "step": 1255
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1606389699126657e-05,
+      "loss": 0.5903,
+      "step": 1256
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1594548937546258e-05,
+      "loss": 0.5732,
+      "step": 1257
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1582705881579515e-05,
+      "loss": 0.5399,
+      "step": 1258
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1570860548267324e-05,
+      "loss": 0.5125,
+      "step": 1259
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.1559012954653865e-05,
+      "loss": 0.4899,
+      "step": 1260
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.1547163117786563e-05,
+      "loss": 0.5025,
+      "step": 1261
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.1535311054716075e-05,
+      "loss": 0.6528,
+      "step": 1262
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.1523456782496254e-05,
+      "loss": 0.5086,
+      "step": 1263
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.1511600318184142e-05,
+      "loss": 0.5212,
+      "step": 1264
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.1499741678839929e-05,
+      "loss": 0.6073,
+      "step": 1265
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.1487880881526933e-05,
+      "loss": 0.5981,
+      "step": 1266
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1476017943311584e-05,
+      "loss": 0.4528,
+      "step": 1267
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1464152881263388e-05,
+      "loss": 0.5873,
+      "step": 1268
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1452285712454905e-05,
+      "loss": 0.497,
+      "step": 1269
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.144041645396173e-05,
+      "loss": 0.4246,
+      "step": 1270
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.1428545122862465e-05,
+      "loss": 0.5137,
+      "step": 1271
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1416671736238689e-05,
+      "loss": 0.5361,
+      "step": 1272
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1404796311174947e-05,
+      "loss": 0.4546,
+      "step": 1273
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1392918864758712e-05,
+      "loss": 0.6455,
+      "step": 1274
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1381039414080366e-05,
+      "loss": 0.3765,
+      "step": 1275
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1369157976233172e-05,
+      "loss": 0.6149,
+      "step": 1276
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1357274568313259e-05,
+      "loss": 0.5702,
+      "step": 1277
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1345389207419588e-05,
+      "loss": 0.43,
+      "step": 1278
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1333501910653926e-05,
+      "loss": 0.3607,
+      "step": 1279
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1321612695120832e-05,
+      "loss": 0.3413,
+      "step": 1280
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1309721577927619e-05,
+      "loss": 0.513,
+      "step": 1281
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1297828576184344e-05,
+      "loss": 0.5757,
+      "step": 1282
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.128593370700377e-05,
+      "loss": 0.5014,
+      "step": 1283
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.1274036987501348e-05,
+      "loss": 0.5221,
+      "step": 1284
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.126213843479519e-05,
+      "loss": 0.6149,
+      "step": 1285
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1250238066006056e-05,
+      "loss": 0.5101,
+      "step": 1286
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.1238335898257305e-05,
+      "loss": 0.5471,
+      "step": 1287
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.122643194867489e-05,
+      "loss": 0.6413,
+      "step": 1288
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.1214526234387329e-05,
+      "loss": 0.4576,
+      "step": 1289
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.1202618772525679e-05,
+      "loss": 0.6052,
+      "step": 1290
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.1190709580223516e-05,
+      "loss": 0.6359,
+      "step": 1291
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1178798674616894e-05,
+      "loss": 0.5471,
+      "step": 1292
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1166886072844344e-05,
+      "loss": 0.4564,
+      "step": 1293
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1154971792046831e-05,
+      "loss": 0.5524,
+      "step": 1294
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1143055849367738e-05,
+      "loss": 0.5727,
+      "step": 1295
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.1131138261952845e-05,
+      "loss": 0.6633,
+      "step": 1296
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1119219046950285e-05,
+      "loss": 0.6513,
+      "step": 1297
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1107298221510543e-05,
+      "loss": 0.5352,
+      "step": 1298
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.1095375802786418e-05,
+      "loss": 0.6163,
+      "step": 1299
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.1083451807933008e-05,
+      "loss": 0.4276,
+      "step": 1300
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.1071526254107667e-05,
+      "loss": 0.4406,
+      "step": 1301
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.1059599158470002e-05,
+      "loss": 0.4949,
+      "step": 1302
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.1047670538181833e-05,
+      "loss": 0.5325,
+      "step": 1303
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.1035740410407183e-05,
+      "loss": 0.3706,
+      "step": 1304
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.1023808792312226e-05,
+      "loss": 0.497,
+      "step": 1305
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.10118757010653e-05,
+      "loss": 0.6443,
+      "step": 1306
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.0999941153836853e-05,
+      "loss": 0.5628,
+      "step": 1307
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.0988005167799427e-05,
+      "loss": 0.3404,
+      "step": 1308
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.0976067760127636e-05,
+      "loss": 0.5746,
+      "step": 1309
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.0964128947998143e-05,
+      "loss": 0.4858,
+      "step": 1310
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.0952188748589625e-05,
+      "loss": 0.5318,
+      "step": 1311
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.0940247179082766e-05,
+      "loss": 0.4923,
+      "step": 1312
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.092830425666021e-05,
+      "loss": 0.5554,
+      "step": 1313
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0916359998506549e-05,
+      "loss": 0.5224,
+      "step": 1314
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.0904414421808303e-05,
+      "loss": 0.6339,
+      "step": 1315
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.0892467543753889e-05,
+      "loss": 0.65,
+      "step": 1316
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.0880519381533591e-05,
+      "loss": 0.6645,
+      "step": 1317
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.0868569952339547e-05,
+      "loss": 0.4912,
+      "step": 1318
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.085661927336571e-05,
+      "loss": 0.5735,
+      "step": 1319
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.0844667361807841e-05,
+      "loss": 0.57,
+      "step": 1320
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0832714234863466e-05,
+      "loss": 0.5231,
+      "step": 1321
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0820759909731867e-05,
+      "loss": 0.5999,
+      "step": 1322
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0808804403614044e-05,
+      "loss": 0.4096,
+      "step": 1323
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0796847733712697e-05,
+      "loss": 0.4197,
+      "step": 1324
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0784889917232206e-05,
+      "loss": 0.427,
+      "step": 1325
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0772930971378597e-05,
+      "loss": 0.524,
+      "step": 1326
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0760970913359519e-05,
+      "loss": 0.5349,
+      "step": 1327
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.0749009760384226e-05,
+      "loss": 0.5874,
+      "step": 1328
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.0737047529663545e-05,
+      "loss": 0.5954,
+      "step": 1329
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0725084238409849e-05,
+      "loss": 0.4996,
+      "step": 1330
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0713119903837046e-05,
+      "loss": 0.5478,
+      "step": 1331
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.070115454316054e-05,
+      "loss": 0.3348,
+      "step": 1332
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.0689188173597219e-05,
+      "loss": 0.3276,
+      "step": 1333
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.0677220812365405e-05,
+      "loss": 0.6157,
+      "step": 1334
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.0665252476684864e-05,
+      "loss": 0.6276,
+      "step": 1335
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.065328318377676e-05,
+      "loss": 0.5437,
+      "step": 1336
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.0641312950863627e-05,
+      "loss": 0.5684,
+      "step": 1337
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.0629341795169361e-05,
+      "loss": 0.5108,
+      "step": 1338
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0617369733919183e-05,
+      "loss": 0.5955,
+      "step": 1339
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0605396784339611e-05,
+      "loss": 0.5497,
+      "step": 1340
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.0593422963658453e-05,
+      "loss": 0.6472,
+      "step": 1341
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.4499,
+      "step": 1342
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0569472777908814e-05,
+      "loss": 0.6405,
+      "step": 1343
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.0557496447302103e-05,
+      "loss": 0.5114,
+      "step": 1344
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.0545519314517292e-05,
+      "loss": 0.4967,
+      "step": 1345
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.0533541396788206e-05,
+      "loss": 0.6657,
+      "step": 1346
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.052156271134979e-05,
+      "loss": 0.5046,
+      "step": 1347
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0509583275438096e-05,
+      "loss": 0.5479,
+      "step": 1348
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0497603106290266e-05,
+      "loss": 0.5391,
+      "step": 1349
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.0485622221144485e-05,
+      "loss": 0.5213,
+      "step": 1350
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.047364063723997e-05,
+      "loss": 0.3992,
+      "step": 1351
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0461658371816954e-05,
+      "loss": 0.6068,
+      "step": 1352
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.0449675442116634e-05,
+      "loss": 0.5534,
+      "step": 1353
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.0437691865381174e-05,
+      "loss": 0.542,
+      "step": 1354
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0425707658853672e-05,
+      "loss": 0.5504,
+      "step": 1355
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0413722839778117e-05,
+      "loss": 0.4237,
+      "step": 1356
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0401737425399396e-05,
+      "loss": 0.5365,
+      "step": 1357
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0389751432963248e-05,
+      "loss": 0.4856,
+      "step": 1358
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.0377764879716234e-05,
+      "loss": 0.459,
+      "step": 1359
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0365777782905735e-05,
+      "loss": 0.4208,
+      "step": 1360
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0353790159779904e-05,
+      "loss": 0.5965,
+      "step": 1361
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.034180202758766e-05,
+      "loss": 0.6307,
+      "step": 1362
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.0329813403578646e-05,
+      "loss": 0.5906,
+      "step": 1363
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0317824305003219e-05,
+      "loss": 0.4476,
+      "step": 1364
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0305834749112422e-05,
+      "loss": 0.6221,
+      "step": 1365
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0293844753157942e-05,
+      "loss": 0.5789,
+      "step": 1366
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0281854334392112e-05,
+      "loss": 0.549,
+      "step": 1367
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0269863510067872e-05,
+      "loss": 0.5333,
+      "step": 1368
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0257872297438737e-05,
+      "loss": 0.6506,
+      "step": 1369
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0245880713758793e-05,
+      "loss": 0.4786,
+      "step": 1370
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.023388877628265e-05,
+      "loss": 0.5897,
+      "step": 1371
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.0221896502265434e-05,
+      "loss": 0.5106,
+      "step": 1372
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.0209903908962747e-05,
+      "loss": 0.5859,
+      "step": 1373
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.0197911013630659e-05,
+      "loss": 0.5941,
+      "step": 1374
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.0185917833525669e-05,
+      "loss": 0.5957,
+      "step": 1375
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.0173924385904692e-05,
+      "loss": 0.4511,
+      "step": 1376
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0161930688025018e-05,
+      "loss": 0.5852,
+      "step": 1377
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.0149936757144306e-05,
+      "loss": 0.5956,
+      "step": 1378
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.0137942610520542e-05,
+      "loss": 0.5689,
+      "step": 1379
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0125948265412033e-05,
+      "loss": 0.3888,
+      "step": 1380
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0113953739077366e-05,
+      "loss": 0.4344,
+      "step": 1381
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0101959048775386e-05,
+      "loss": 0.6475,
+      "step": 1382
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0089964211765179e-05,
+      "loss": 0.5883,
+      "step": 1383
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.0077969245306038e-05,
+      "loss": 0.4893,
+      "step": 1384
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.0065974166657448e-05,
+      "loss": 0.4631,
+      "step": 1385
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.0053978993079046e-05,
+      "loss": 0.5813,
+      "step": 1386
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.0041983741830611e-05,
+      "loss": 0.5834,
+      "step": 1387
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.0029988430172041e-05,
+      "loss": 0.5755,
+      "step": 1388
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0017993075363305e-05,
+      "loss": 0.4734,
+      "step": 1389
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0005997694664451e-05,
+      "loss": 0.6005,
+      "step": 1390
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.99400230533555e-06,
+      "loss": 0.507,
+      "step": 1391
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.982006924636696e-06,
+      "loss": 0.4554,
+      "step": 1392
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.970011569827964e-06,
+      "loss": 0.5825,
+      "step": 1393
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.95801625816939e-06,
+      "loss": 0.6278,
+      "step": 1394
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.946021006920959e-06,
+      "loss": 0.4036,
+      "step": 1395
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.934025833342557e-06,
+      "loss": 0.2871,
+      "step": 1396
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.922030754693962e-06,
+      "loss": 0.4608,
+      "step": 1397
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.910035788234823e-06,
+      "loss": 0.5348,
+      "step": 1398
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.898040951224619e-06,
+      "loss": 0.3576,
+      "step": 1399
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.886046260922635e-06,
+      "loss": 0.3996,
+      "step": 1400
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.87405173458797e-06,
+      "loss": 0.5955,
+      "step": 1401
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.862057389479462e-06,
+      "loss": 0.575,
+      "step": 1402
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.8500632428557e-06,
+      "loss": 0.5894,
+      "step": 1403
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.838069311974986e-06,
+      "loss": 0.6588,
+      "step": 1404
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.826075614095311e-06,
+      "loss": 0.5394,
+      "step": 1405
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.814082166474333e-06,
+      "loss": 0.5674,
+      "step": 1406
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.802088986369343e-06,
+      "loss": 0.5887,
+      "step": 1407
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.790096091037253e-06,
+      "loss": 0.4568,
+      "step": 1408
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.77810349773457e-06,
+      "loss": 0.3765,
+      "step": 1409
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.766111223717352e-06,
+      "loss": 0.6501,
+      "step": 1410
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.754119286241207e-06,
+      "loss": 0.555,
+      "step": 1411
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.742127702561265e-06,
+      "loss": 0.5503,
+      "step": 1412
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.730136489932133e-06,
+      "loss": 0.3947,
+      "step": 1413
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.71814566560789e-06,
+      "loss": 0.6936,
+      "step": 1414
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.706155246842061e-06,
+      "loss": 0.541,
+      "step": 1415
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.694165250887583e-06,
+      "loss": 0.5161,
+      "step": 1416
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.682175694996781e-06,
+      "loss": 0.6138,
+      "step": 1417
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.670186596421358e-06,
+      "loss": 0.5773,
+      "step": 1418
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.658197972412345e-06,
+      "loss": 0.5952,
+      "step": 1419
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.646209840220097e-06,
+      "loss": 0.5757,
+      "step": 1420
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.634222217094268e-06,
+      "loss": 0.5486,
+      "step": 1421
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.622235120283769e-06,
+      "loss": 0.4801,
+      "step": 1422
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.610248567036755e-06,
+      "loss": 0.5878,
+      "step": 1423
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.598262574600606e-06,
+      "loss": 0.3732,
+      "step": 1424
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.586277160221885e-06,
+      "loss": 0.4976,
+      "step": 1425
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.574292341146332e-06,
+      "loss": 0.4215,
+      "step": 1426
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.562308134618828e-06,
+      "loss": 0.6222,
+      "step": 1427
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.550324557883373e-06,
+      "loss": 0.4943,
+      "step": 1428
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.53834162818305e-06,
+      "loss": 0.5247,
+      "step": 1429
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.526359362760032e-06,
+      "loss": 0.4962,
+      "step": 1430
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.514377778855521e-06,
+      "loss": 0.6123,
+      "step": 1431
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.502396893709736e-06,
+      "loss": 0.5233,
+      "step": 1432
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.490416724561907e-06,
+      "loss": 0.436,
+      "step": 1433
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.478437288650212e-06,
+      "loss": 0.604,
+      "step": 1434
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.466458603211796e-06,
+      "loss": 0.6185,
+      "step": 1435
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.45448068548271e-06,
+      "loss": 0.591,
+      "step": 1436
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.442503552697898e-06,
+      "loss": 0.4802,
+      "step": 1437
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.43052722209119e-06,
+      "loss": 0.419,
+      "step": 1438
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.6561,
+      "step": 1439
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.406577036341548e-06,
+      "loss": 0.3602,
+      "step": 1440
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.39460321566039e-06,
+      "loss": 0.4716,
+      "step": 1441
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.382630266080822e-06,
+      "loss": 0.5363,
+      "step": 1442
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.37065820483064e-06,
+      "loss": 0.5734,
+      "step": 1443
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.358687049136376e-06,
+      "loss": 0.5034,
+      "step": 1444
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.346716816223246e-06,
+      "loss": 0.4553,
+      "step": 1445
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.334747523315138e-06,
+      "loss": 0.5114,
+      "step": 1446
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.322779187634599e-06,
+      "loss": 0.6871,
+      "step": 1447
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.310811826402786e-06,
+      "loss": 0.4394,
+      "step": 1448
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.298845456839459e-06,
+      "loss": 0.5207,
+      "step": 1449
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.286880096162955e-06,
+      "loss": 0.4512,
+      "step": 1450
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.274915761590156e-06,
+      "loss": 0.6167,
+      "step": 1451
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.262952470336458e-06,
+      "loss": 0.3861,
+      "step": 1452
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.250990239615778e-06,
+      "loss": 0.5861,
+      "step": 1453
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.239029086640485e-06,
+      "loss": 0.564,
+      "step": 1454
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.227069028621406e-06,
+      "loss": 0.4605,
+      "step": 1455
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.215110082767797e-06,
+      "loss": 0.454,
+      "step": 1456
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.203152266287306e-06,
+      "loss": 0.5793,
+      "step": 1457
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.19119559638596e-06,
+      "loss": 0.5612,
+      "step": 1458
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.179240090268136e-06,
+      "loss": 0.67,
+      "step": 1459
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.167285765136533e-06,
+      "loss": 0.4719,
+      "step": 1460
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.15533263819216e-06,
+      "loss": 0.5201,
+      "step": 1461
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.143380726634291e-06,
+      "loss": 0.4425,
+      "step": 1462
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.131430047660454e-06,
+      "loss": 0.5496,
+      "step": 1463
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.11948061846641e-06,
+      "loss": 0.6232,
+      "step": 1464
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.107532456246114e-06,
+      "loss": 0.4562,
+      "step": 1465
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.095585578191697e-06,
+      "loss": 0.4648,
+      "step": 1466
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.083640001493455e-06,
+      "loss": 0.6131,
+      "step": 1467
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.071695743339795e-06,
+      "loss": 0.4554,
+      "step": 1468
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.059752820917237e-06,
+      "loss": 0.4416,
+      "step": 1469
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.047811251410377e-06,
+      "loss": 0.5318,
+      "step": 1470
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.03587105200186e-06,
+      "loss": 0.4018,
+      "step": 1471
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.023932239872366e-06,
+      "loss": 0.4467,
+      "step": 1472
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.011994832200577e-06,
+      "loss": 0.5809,
+      "step": 1473
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.000058846163152e-06,
+      "loss": 0.3821,
+      "step": 1474
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 8.9881242989347e-06,
+      "loss": 0.5025,
+      "step": 1475
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.976191207687775e-06,
+      "loss": 0.5378,
+      "step": 1476
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.964259589592824e-06,
+      "loss": 0.5782,
+      "step": 1477
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.952329461818167e-06,
+      "loss": 0.5883,
+      "step": 1478
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.940400841530001e-06,
+      "loss": 0.49,
+      "step": 1479
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.928473745892338e-06,
+      "loss": 0.5471,
+      "step": 1480
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.916548192066996e-06,
+      "loss": 0.5152,
+      "step": 1481
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.904624197213585e-06,
+      "loss": 0.6122,
+      "step": 1482
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.892701778489464e-06,
+      "loss": 0.5475,
+      "step": 1483
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.880780953049719e-06,
+      "loss": 0.537,
+      "step": 1484
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.868861738047158e-06,
+      "loss": 0.554,
+      "step": 1485
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.85694415063226e-06,
+      "loss": 0.6159,
+      "step": 1486
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.84502820795317e-06,
+      "loss": 0.4711,
+      "step": 1487
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.83311392715566e-06,
+      "loss": 0.6515,
+      "step": 1488
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.821201325383106e-06,
+      "loss": 0.5405,
+      "step": 1489
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.809290419776488e-06,
+      "loss": 0.495,
+      "step": 1490
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.797381227474324e-06,
+      "loss": 0.5803,
+      "step": 1491
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.785473765612673e-06,
+      "loss": 0.4297,
+      "step": 1492
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.773568051325114e-06,
+      "loss": 0.4814,
+      "step": 1493
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.7616641017427e-06,
+      "loss": 0.6352,
+      "step": 1494
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.749761933993946e-06,
+      "loss": 0.5446,
+      "step": 1495
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.737861565204811e-06,
+      "loss": 0.5779,
+      "step": 1496
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.725963012498657e-06,
+      "loss": 0.5503,
+      "step": 1497
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.714066292996232e-06,
+      "loss": 0.4511,
+      "step": 1498
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.702171423815658e-06,
+      "loss": 0.6438,
+      "step": 1499
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.690278422072384e-06,
+      "loss": 0.5464,
+      "step": 1500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.678387304879173e-06,
+      "loss": 0.514,
+      "step": 1501
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.666498089346077e-06,
+      "loss": 0.5127,
+      "step": 1502
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.654610792580415e-06,
+      "loss": 0.5487,
+      "step": 1503
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.642725431686743e-06,
+      "loss": 0.459,
+      "step": 1504
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.630842023766832e-06,
+      "loss": 0.515,
+      "step": 1505
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.61896058591964e-06,
+      "loss": 0.4562,
+      "step": 1506
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.60708113524129e-06,
+      "loss": 0.467,
+      "step": 1507
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.595203688825055e-06,
+      "loss": 0.4876,
+      "step": 1508
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.583328263761316e-06,
+      "loss": 0.6035,
+      "step": 1509
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.571454877137538e-06,
+      "loss": 0.3102,
+      "step": 1510
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.559583546038274e-06,
+      "loss": 0.6289,
+      "step": 1511
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.5477142875451e-06,
+      "loss": 0.445,
+      "step": 1512
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.535847118736615e-06,
+      "loss": 0.6178,
+      "step": 1513
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.523982056688418e-06,
+      "loss": 0.5203,
+      "step": 1514
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.512119118473068e-06,
+      "loss": 0.583,
+      "step": 1515
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.500258321160073e-06,
+      "loss": 0.5763,
+      "step": 1516
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.488399681815861e-06,
+      "loss": 0.6764,
+      "step": 1517
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.476543217503748e-06,
+      "loss": 0.5227,
+      "step": 1518
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.464688945283928e-06,
+      "loss": 0.4999,
+      "step": 1519
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.452836882213439e-06,
+      "loss": 0.5653,
+      "step": 1520
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.440987045346135e-06,
+      "loss": 0.615,
+      "step": 1521
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.429139451732678e-06,
+      "loss": 0.469,
+      "step": 1522
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.41729411842049e-06,
+      "loss": 0.385,
+      "step": 1523
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.405451062453744e-06,
+      "loss": 0.4811,
+      "step": 1524
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.393610300873346e-06,
+      "loss": 0.3988,
+      "step": 1525
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.381771850716887e-06,
+      "loss": 0.4694,
+      "step": 1526
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.369935729018635e-06,
+      "loss": 0.4692,
+      "step": 1527
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.358101952809517e-06,
+      "loss": 0.4616,
+      "step": 1528
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.346270539117077e-06,
+      "loss": 0.6041,
+      "step": 1529
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.334441504965456e-06,
+      "loss": 0.3467,
+      "step": 1530
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.322614867375383e-06,
+      "loss": 0.5391,
+      "step": 1531
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.310790643364127e-06,
+      "loss": 0.5132,
+      "step": 1532
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.29896884994549e-06,
+      "loss": 0.4599,
+      "step": 1533
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.287149504129778e-06,
+      "loss": 0.6054,
+      "step": 1534
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.275332622923768e-06,
+      "loss": 0.428,
+      "step": 1535
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.4435,
+      "step": 1536
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.251706322350235e-06,
+      "loss": 0.5586,
+      "step": 1537
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.239896936978444e-06,
+      "loss": 0.3969,
+      "step": 1538
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.228090084207773e-06,
+      "loss": 0.4939,
+      "step": 1539
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.216285781027036e-06,
+      "loss": 0.5054,
+      "step": 1540
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.204484044421361e-06,
+      "loss": 0.5231,
+      "step": 1541
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.192684891372198e-06,
+      "loss": 0.4528,
+      "step": 1542
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.18088833885727e-06,
+      "loss": 0.4738,
+      "step": 1543
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.169094403850562e-06,
+      "loss": 0.5568,
+      "step": 1544
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.157303103322296e-06,
+      "loss": 0.466,
+      "step": 1545
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.145514454238896e-06,
+      "loss": 0.4052,
+      "step": 1546
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.133728473562971e-06,
+      "loss": 0.5642,
+      "step": 1547
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.1219451782533e-06,
+      "loss": 0.4147,
+      "step": 1548
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.11016458526479e-06,
+      "loss": 0.541,
+      "step": 1549
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.098386711548459e-06,
+      "loss": 0.5732,
+      "step": 1550
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.086611574051418e-06,
+      "loss": 0.567,
+      "step": 1551
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.074839189716836e-06,
+      "loss": 0.325,
+      "step": 1552
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.063069575483918e-06,
+      "loss": 0.6324,
+      "step": 1553
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.051302748287895e-06,
+      "loss": 0.4996,
+      "step": 1554
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.039538725059976e-06,
+      "loss": 0.5209,
+      "step": 1555
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.027777522727335e-06,
+      "loss": 0.6024,
+      "step": 1556
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.016019158213103e-06,
+      "loss": 0.4847,
+      "step": 1557
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 8.004263648436308e-06,
+      "loss": 0.573,
+      "step": 1558
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 7.992511010311872e-06,
+      "loss": 0.5264,
+      "step": 1559
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.980761260750607e-06,
+      "loss": 0.6366,
+      "step": 1560
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.969014416659142e-06,
+      "loss": 0.6005,
+      "step": 1561
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.957270494939935e-06,
+      "loss": 0.4736,
+      "step": 1562
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.945529512491252e-06,
+      "loss": 0.4571,
+      "step": 1563
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.933791486207107e-06,
+      "loss": 0.5617,
+      "step": 1564
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.922056432977276e-06,
+      "loss": 0.5954,
+      "step": 1565
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.91032436968725e-06,
+      "loss": 0.5415,
+      "step": 1566
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.898595313218224e-06,
+      "loss": 0.5568,
+      "step": 1567
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.88686928044706e-06,
+      "loss": 0.5114,
+      "step": 1568
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.875146288246275e-06,
+      "loss": 0.3469,
+      "step": 1569
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.863426353484002e-06,
+      "loss": 0.5643,
+      "step": 1570
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.85170949302399e-06,
+      "loss": 0.6052,
+      "step": 1571
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.839995723725549e-06,
+      "loss": 0.5554,
+      "step": 1572
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.828285062443545e-06,
+      "loss": 0.5703,
+      "step": 1573
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.816577526028381e-06,
+      "loss": 0.5711,
+      "step": 1574
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.804873131325955e-06,
+      "loss": 0.5949,
+      "step": 1575
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.793171895177638e-06,
+      "loss": 0.5306,
+      "step": 1576
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.781473834420277e-06,
+      "loss": 0.5761,
+      "step": 1577
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.76977896588613e-06,
+      "loss": 0.5988,
+      "step": 1578
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.758087306402868e-06,
+      "loss": 0.6972,
+      "step": 1579
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.74639887279355e-06,
+      "loss": 0.4991,
+      "step": 1580
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.734713681876589e-06,
+      "loss": 0.5741,
+      "step": 1581
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.723031750465726e-06,
+      "loss": 0.576,
+      "step": 1582
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.711353095370025e-06,
+      "loss": 0.5496,
+      "step": 1583
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.699677733393827e-06,
+      "loss": 0.5471,
+      "step": 1584
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.68800568133673e-06,
+      "loss": 0.5021,
+      "step": 1585
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.676336955993585e-06,
+      "loss": 0.4158,
+      "step": 1586
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.66467157415444e-06,
+      "loss": 0.6368,
+      "step": 1587
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.653009552604538e-06,
+      "loss": 0.5439,
+      "step": 1588
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.641350908124297e-06,
+      "loss": 0.4939,
+      "step": 1589
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.629695657489258e-06,
+      "loss": 0.4644,
+      "step": 1590
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.618043817470087e-06,
+      "loss": 0.5597,
+      "step": 1591
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.606395404832548e-06,
+      "loss": 0.5526,
+      "step": 1592
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.594750436337467e-06,
+      "loss": 0.5321,
+      "step": 1593
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.583108928740708e-06,
+      "loss": 0.4094,
+      "step": 1594
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.571470898793173e-06,
+      "loss": 0.4645,
+      "step": 1595
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.559836363240741e-06,
+      "loss": 0.6365,
+      "step": 1596
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.548205338824277e-06,
+      "loss": 0.4517,
+      "step": 1597
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.536577842279584e-06,
+      "loss": 0.4202,
+      "step": 1598
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.5249538903373954e-06,
+      "loss": 0.3987,
+      "step": 1599
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.5133334997233435e-06,
+      "loss": 0.6289,
+      "step": 1600
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.501716687157931e-06,
+      "loss": 0.5454,
+      "step": 1601
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.490103469356513e-06,
+      "loss": 0.5099,
+      "step": 1602
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.4784938630292835e-06,
+      "loss": 0.5137,
+      "step": 1603
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.466887884881226e-06,
+      "loss": 0.5897,
+      "step": 1604
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.455285551612106e-06,
+      "loss": 0.523,
+      "step": 1605
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.443686879916455e-06,
+      "loss": 0.4473,
+      "step": 1606
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.432091886483524e-06,
+      "loss": 0.3423,
+      "step": 1607
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.42050058799727e-06,
+      "loss": 0.5476,
+      "step": 1608
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.40891300113635e-06,
+      "loss": 0.538,
+      "step": 1609
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.3973291425740625e-06,
+      "loss": 0.6187,
+      "step": 1610
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.385749028978347e-06,
+      "loss": 0.5027,
+      "step": 1611
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.374172677011755e-06,
+      "loss": 0.4491,
+      "step": 1612
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.362600103331432e-06,
+      "loss": 0.4712,
+      "step": 1613
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.351031324589074e-06,
+      "loss": 0.5304,
+      "step": 1614
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.3394663574309285e-06,
+      "loss": 0.5691,
+      "step": 1615
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.327905218497752e-06,
+      "loss": 0.4001,
+      "step": 1616
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.316347924424787e-06,
+      "loss": 0.6189,
+      "step": 1617
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.304794491841761e-06,
+      "loss": 0.6983,
+      "step": 1618
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.293244937372831e-06,
+      "loss": 0.5829,
+      "step": 1619
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.2816992776365714e-06,
+      "loss": 0.5848,
+      "step": 1620
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 7.270157529245969e-06,
+      "loss": 0.6247,
+      "step": 1621
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 7.258619708808368e-06,
+      "loss": 0.4082,
+      "step": 1622
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 7.2470858329254605e-06,
+      "loss": 0.5576,
+      "step": 1623
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 7.2355559181932745e-06,
+      "loss": 0.5165,
+      "step": 1624
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 7.224029981202123e-06,
+      "loss": 0.5589,
+      "step": 1625
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 7.2125080385366146e-06,
+      "loss": 0.5288,
+      "step": 1626
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 7.200990106775591e-06,
+      "loss": 0.342,
+      "step": 1627
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 7.1894762024921315e-06,
+      "loss": 0.5777,
+      "step": 1628
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 7.1779663422535235e-06,
+      "loss": 0.4437,
+      "step": 1629
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 7.166460542621229e-06,
+      "loss": 0.6088,
+      "step": 1630
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 7.154958820150872e-06,
+      "loss": 0.4743,
+      "step": 1631
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 7.1434611913922066e-06,
+      "loss": 0.4222,
+      "step": 1632
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.5438,
+      "step": 1633
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 7.1204782811794974e-06,
+      "loss": 0.5206,
+      "step": 1634
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 7.108993032795418e-06,
+      "loss": 0.4801,
+      "step": 1635
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 7.097511944262912e-06,
+      "loss": 0.5628,
+      "step": 1636
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 7.086035032102037e-06,
+      "loss": 0.5061,
+      "step": 1637
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 7.0745623128268605e-06,
+      "loss": 0.4422,
+      "step": 1638
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 7.0630938029454e-06,
+      "loss": 0.4718,
+      "step": 1639
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 7.051629518959615e-06,
+      "loss": 0.5705,
+      "step": 1640
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 7.040169477365404e-06,
+      "loss": 0.3999,
+      "step": 1641
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 7.0287136946525415e-06,
+      "loss": 0.5099,
+      "step": 1642
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 7.0172621873046785e-06,
+      "loss": 0.5088,
+      "step": 1643
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 7.005814971799319e-06,
+      "loss": 0.6242,
+      "step": 1644
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 6.994372064607788e-06,
+      "loss": 0.5644,
+      "step": 1645
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 6.982933482195211e-06,
+      "loss": 0.4482,
+      "step": 1646
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 6.971499241020495e-06,
+      "loss": 0.5393,
+      "step": 1647
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 6.960069357536294e-06,
+      "loss": 0.4233,
+      "step": 1648
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 6.948643848188991e-06,
+      "loss": 0.4096,
+      "step": 1649
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 6.93722272941869e-06,
+      "loss": 0.5392,
+      "step": 1650
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 6.925806017659153e-06,
+      "loss": 0.5261,
+      "step": 1651
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 6.914393729337827e-06,
+      "loss": 0.5747,
+      "step": 1652
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 6.902985880875773e-06,
+      "loss": 0.5207,
+      "step": 1653
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 6.8915824886876715e-06,
+      "loss": 0.5505,
+      "step": 1654
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 6.880183569181795e-06,
+      "loss": 0.5886,
+      "step": 1655
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 6.868789138759977e-06,
+      "loss": 0.6188,
+      "step": 1656
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 6.8573992138175835e-06,
+      "loss": 0.5601,
+      "step": 1657
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 6.846013810743515e-06,
+      "loss": 0.5357,
+      "step": 1658
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 6.83463294592015e-06,
+      "loss": 0.3215,
+      "step": 1659
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 6.823256635723343e-06,
+      "loss": 0.3792,
+      "step": 1660
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 6.811884896522397e-06,
+      "loss": 0.53,
+      "step": 1661
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 6.800517744680032e-06,
+      "loss": 0.6126,
+      "step": 1662
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 6.789155196552373e-06,
+      "loss": 0.5984,
+      "step": 1663
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 6.777797268488918e-06,
+      "loss": 0.4952,
+      "step": 1664
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 6.766443976832518e-06,
+      "loss": 0.5683,
+      "step": 1665
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 6.755095337919349e-06,
+      "loss": 0.3693,
+      "step": 1666
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 6.743751368078899e-06,
+      "loss": 0.5016,
+      "step": 1667
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 6.7324120836339356e-06,
+      "loss": 0.4857,
+      "step": 1668
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 6.721077500900476e-06,
+      "loss": 0.4749,
+      "step": 1669
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 6.709747636187789e-06,
+      "loss": 0.5222,
+      "step": 1670
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 6.698422505798339e-06,
+      "loss": 0.6383,
+      "step": 1671
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 6.687102126027782e-06,
+      "loss": 0.4982,
+      "step": 1672
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 6.675786513164948e-06,
+      "loss": 0.4356,
+      "step": 1673
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 6.664475683491797e-06,
+      "loss": 0.6702,
+      "step": 1674
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.653169653283407e-06,
+      "loss": 0.567,
+      "step": 1675
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 6.641868438807955e-06,
+      "loss": 0.4607,
+      "step": 1676
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 6.630572056326687e-06,
+      "loss": 0.5915,
+      "step": 1677
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 6.6192805220939e-06,
+      "loss": 0.5136,
+      "step": 1678
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 6.607993852356907e-06,
+      "loss": 0.5423,
+      "step": 1679
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 6.596712063356021e-06,
+      "loss": 0.585,
+      "step": 1680
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.585435171324546e-06,
+      "loss": 0.5051,
+      "step": 1681
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 6.574163192488725e-06,
+      "loss": 0.6047,
+      "step": 1682
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 6.562896143067734e-06,
+      "loss": 0.6548,
+      "step": 1683
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 6.551634039273667e-06,
+      "loss": 0.3085,
+      "step": 1684
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 6.540376897311489e-06,
+      "loss": 0.5771,
+      "step": 1685
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.529124733379024e-06,
+      "loss": 0.5208,
+      "step": 1686
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 6.517877563666951e-06,
+      "loss": 0.581,
+      "step": 1687
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 6.506635404358745e-06,
+      "loss": 0.5752,
+      "step": 1688
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 6.495398271630675e-06,
+      "loss": 0.501,
+      "step": 1689
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 6.484166181651786e-06,
+      "loss": 0.4404,
+      "step": 1690
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 6.472939150583859e-06,
+      "loss": 0.4991,
+      "step": 1691
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 6.461717194581394e-06,
+      "loss": 0.5072,
+      "step": 1692
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 6.450500329791597e-06,
+      "loss": 0.4213,
+      "step": 1693
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 6.439288572354338e-06,
+      "loss": 0.4485,
+      "step": 1694
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 6.42808193840215e-06,
+      "loss": 0.5146,
+      "step": 1695
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 6.416880444060185e-06,
+      "loss": 0.5792,
+      "step": 1696
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 6.405684105446202e-06,
+      "loss": 0.5673,
+      "step": 1697
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 6.3944929386705375e-06,
+      "loss": 0.5581,
+      "step": 1698
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 6.383306959836098e-06,
+      "loss": 0.4733,
+      "step": 1699
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 6.372126185038313e-06,
+      "loss": 0.4936,
+      "step": 1700
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 6.360950630365126e-06,
+      "loss": 0.4984,
+      "step": 1701
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 6.3497803118969766e-06,
+      "loss": 0.3641,
+      "step": 1702
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 6.338615245706765e-06,
+      "loss": 0.5701,
+      "step": 1703
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 6.327455447859827e-06,
+      "loss": 0.4945,
+      "step": 1704
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 6.316300934413935e-06,
+      "loss": 0.5165,
+      "step": 1705
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 6.305151721419238e-06,
+      "loss": 0.5158,
+      "step": 1706
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 6.294007824918276e-06,
+      "loss": 0.425,
+      "step": 1707
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 6.282869260945928e-06,
+      "loss": 0.616,
+      "step": 1708
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 6.271736045529399e-06,
+      "loss": 0.6389,
+      "step": 1709
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 6.260608194688207e-06,
+      "loss": 0.5032,
+      "step": 1710
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 6.249485724434145e-06,
+      "loss": 0.5077,
+      "step": 1711
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 6.238368650771261e-06,
+      "loss": 0.5375,
+      "step": 1712
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 6.227256989695849e-06,
+      "loss": 0.4382,
+      "step": 1713
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 6.216150757196402e-06,
+      "loss": 0.6261,
+      "step": 1714
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 6.205049969253604e-06,
+      "loss": 0.4276,
+      "step": 1715
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 6.193954641840318e-06,
+      "loss": 0.5407,
+      "step": 1716
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 6.182864790921536e-06,
+      "loss": 0.3821,
+      "step": 1717
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 6.17178043245437e-06,
+      "loss": 0.5463,
+      "step": 1718
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 6.160701582388039e-06,
+      "loss": 0.5475,
+      "step": 1719
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 6.149628256663827e-06,
+      "loss": 0.6382,
+      "step": 1720
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 6.1385604712150696e-06,
+      "loss": 0.4535,
+      "step": 1721
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 6.127498241967139e-06,
+      "loss": 0.4167,
+      "step": 1722
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 6.116441584837404e-06,
+      "loss": 0.5736,
+      "step": 1723
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 6.105390515735217e-06,
+      "loss": 0.3346,
+      "step": 1724
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 6.094345050561892e-06,
+      "loss": 0.5295,
+      "step": 1725
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 6.0833052052106766e-06,
+      "loss": 0.5798,
+      "step": 1726
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 6.072270995566738e-06,
+      "loss": 0.6229,
+      "step": 1727
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 6.061242437507131e-06,
+      "loss": 0.6223,
+      "step": 1728
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 6.050219546900774e-06,
+      "loss": 0.7088,
+      "step": 1729
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.5259,
+      "step": 1730
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 6.028190831482703e-06,
+      "loss": 0.5464,
+      "step": 1731
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 6.017185038367967e-06,
+      "loss": 0.5106,
+      "step": 1732
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 6.0061849761003996e-06,
+      "loss": 0.5439,
+      "step": 1733
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.995190660507916e-06,
+      "loss": 0.3606,
+      "step": 1734
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.984202107410161e-06,
+      "loss": 0.5801,
+      "step": 1735
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.9732193326185035e-06,
+      "loss": 0.5031,
+      "step": 1736
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.962242351935985e-06,
+      "loss": 0.6059,
+      "step": 1737
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.951271181157311e-06,
+      "loss": 0.4608,
+      "step": 1738
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.94030583606883e-06,
+      "loss": 0.5924,
+      "step": 1739
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.929346332448511e-06,
+      "loss": 0.5494,
+      "step": 1740
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.91839268606591e-06,
+      "loss": 0.5617,
+      "step": 1741
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.907444912682163e-06,
+      "loss": 0.4529,
+      "step": 1742
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.89650302804995e-06,
+      "loss": 0.6045,
+      "step": 1743
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.885567047913474e-06,
+      "loss": 0.4037,
+      "step": 1744
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.874636988008457e-06,
+      "loss": 0.4747,
+      "step": 1745
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.863712864062089e-06,
+      "loss": 0.5754,
+      "step": 1746
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.852794691793016e-06,
+      "loss": 0.5827,
+      "step": 1747
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.841882486911338e-06,
+      "loss": 0.5599,
+      "step": 1748
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.8309762651185485e-06,
+      "loss": 0.5103,
+      "step": 1749
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.820076042107545e-06,
+      "loss": 0.4355,
+      "step": 1750
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.809181833562592e-06,
+      "loss": 0.6014,
+      "step": 1751
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.798293655159291e-06,
+      "loss": 0.3803,
+      "step": 1752
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.787411522564574e-06,
+      "loss": 0.6212,
+      "step": 1753
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.776535451436672e-06,
+      "loss": 0.5582,
+      "step": 1754
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.765665457425102e-06,
+      "loss": 0.5524,
+      "step": 1755
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.754801556170615e-06,
+      "loss": 0.5698,
+      "step": 1756
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.743943763305227e-06,
+      "loss": 0.5376,
+      "step": 1757
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.733092094452135e-06,
+      "loss": 0.494,
+      "step": 1758
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.722246565225739e-06,
+      "loss": 0.6206,
+      "step": 1759
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.711407191231603e-06,
+      "loss": 0.4213,
+      "step": 1760
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.700573988066434e-06,
+      "loss": 0.5577,
+      "step": 1761
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.689746971318058e-06,
+      "loss": 0.3651,
+      "step": 1762
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.678926156565405e-06,
+      "loss": 0.5928,
+      "step": 1763
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.6681115593784705e-06,
+      "loss": 0.6096,
+      "step": 1764
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.6573031953183115e-06,
+      "loss": 0.5423,
+      "step": 1765
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.646501079937019e-06,
+      "loss": 0.5451,
+      "step": 1766
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.635705228777677e-06,
+      "loss": 0.5624,
+      "step": 1767
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.624915657374383e-06,
+      "loss": 0.3957,
+      "step": 1768
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.61413238125217e-06,
+      "loss": 0.5379,
+      "step": 1769
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.6033554159270294e-06,
+      "loss": 0.5217,
+      "step": 1770
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.59258477690587e-06,
+      "loss": 0.5172,
+      "step": 1771
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.581820479686498e-06,
+      "loss": 0.4129,
+      "step": 1772
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.571062539757582e-06,
+      "loss": 0.6088,
+      "step": 1773
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.560310972598667e-06,
+      "loss": 0.5897,
+      "step": 1774
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.549565793680105e-06,
+      "loss": 0.5662,
+      "step": 1775
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.5388270184630684e-06,
+      "loss": 0.339,
+      "step": 1776
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.528094662399515e-06,
+      "loss": 0.3864,
+      "step": 1777
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.517368740932164e-06,
+      "loss": 0.6345,
+      "step": 1778
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.5066492694944706e-06,
+      "loss": 0.5705,
+      "step": 1779
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.495936263510617e-06,
+      "loss": 0.6429,
+      "step": 1780
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.485229738395481e-06,
+      "loss": 0.4863,
+      "step": 1781
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.4745297095546125e-06,
+      "loss": 0.5109,
+      "step": 1782
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.463836192384218e-06,
+      "loss": 0.474,
+      "step": 1783
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.453149202271124e-06,
+      "loss": 0.3991,
+      "step": 1784
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.442468754592778e-06,
+      "loss": 0.5277,
+      "step": 1785
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.4317948647172085e-06,
+      "loss": 0.5881,
+      "step": 1786
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.421127548003005e-06,
+      "loss": 0.4489,
+      "step": 1787
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.410466819799306e-06,
+      "loss": 0.3216,
+      "step": 1788
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.399812695445767e-06,
+      "loss": 0.4754,
+      "step": 1789
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.389165190272527e-06,
+      "loss": 0.4578,
+      "step": 1790
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.378524319600231e-06,
+      "loss": 0.5075,
+      "step": 1791
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.3678900987399495e-06,
+      "loss": 0.5082,
+      "step": 1792
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.357262542993198e-06,
+      "loss": 0.4173,
+      "step": 1793
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.346641667651897e-06,
+      "loss": 0.4446,
+      "step": 1794
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.336027487998365e-06,
+      "loss": 0.4921,
+      "step": 1795
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.325420019305267e-06,
+      "loss": 0.5474,
+      "step": 1796
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.3148192768356255e-06,
+      "loss": 0.5768,
+      "step": 1797
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.304225275842788e-06,
+      "loss": 0.4815,
+      "step": 1798
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.293638031570382e-06,
+      "loss": 0.5064,
+      "step": 1799
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.2830575592523415e-06,
+      "loss": 0.4859,
+      "step": 1800
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.27248387411283e-06,
+      "loss": 0.5492,
+      "step": 1801
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.261916991366259e-06,
+      "loss": 0.5359,
+      "step": 1802
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.2513569262172505e-06,
+      "loss": 0.5973,
+      "step": 1803
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.240803693860616e-06,
+      "loss": 0.4808,
+      "step": 1804
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.230257309481327e-06,
+      "loss": 0.4481,
+      "step": 1805
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.2197177882545215e-06,
+      "loss": 0.584,
+      "step": 1806
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.209185145345439e-06,
+      "loss": 0.5488,
+      "step": 1807
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.198659395909438e-06,
+      "loss": 0.5085,
+      "step": 1808
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.18814055509195e-06,
+      "loss": 0.4123,
+      "step": 1809
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.177628638028472e-06,
+      "loss": 0.5503,
+      "step": 1810
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.1671236598445295e-06,
+      "loss": 0.518,
+      "step": 1811
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.156625635655672e-06,
+      "loss": 0.3895,
+      "step": 1812
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.146134580567438e-06,
+      "loss": 0.4698,
+      "step": 1813
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.135650509675342e-06,
+      "loss": 0.4859,
+      "step": 1814
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.12517343806485e-06,
+      "loss": 0.5322,
+      "step": 1815
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.114703380811342e-06,
+      "loss": 0.5319,
+      "step": 1816
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.104240352980134e-06,
+      "loss": 0.5331,
+      "step": 1817
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.093784369626397e-06,
+      "loss": 0.521,
+      "step": 1818
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.083335445795183e-06,
+      "loss": 0.575,
+      "step": 1819
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.072893596521383e-06,
+      "loss": 0.5906,
+      "step": 1820
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.0624588368297115e-06,
+      "loss": 0.3922,
+      "step": 1821
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.052031181734667e-06,
+      "loss": 0.6111,
+      "step": 1822
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.041610646240549e-06,
+      "loss": 0.4906,
+      "step": 1823
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.03119724534139e-06,
+      "loss": 0.5277,
+      "step": 1824
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.020790994020972e-06,
+      "loss": 0.5461,
+      "step": 1825
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.010391907252781e-06,
+      "loss": 0.5873,
+      "step": 1826
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.5082,
+      "step": 1827
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.989615287215478e-06,
+      "loss": 0.5325,
+      "step": 1828
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.979237783841707e-06,
+      "loss": 0.4904,
+      "step": 1829
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9688675048108204e-06,
+      "loss": 0.5189,
+      "step": 1830
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.958504465044532e-06,
+      "loss": 0.3789,
+      "step": 1831
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.948148679454171e-06,
+      "loss": 0.5123,
+      "step": 1832
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.9378001629406e-06,
+      "loss": 0.5588,
+      "step": 1833
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.927458930394238e-06,
+      "loss": 0.5352,
+      "step": 1834
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.917124996695017e-06,
+      "loss": 0.5828,
+      "step": 1835
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.9067983767123736e-06,
+      "loss": 0.4883,
+      "step": 1836
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.8964790853052045e-06,
+      "loss": 0.6074,
+      "step": 1837
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.886167137321887e-06,
+      "loss": 0.4691,
+      "step": 1838
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.8758625476002075e-06,
+      "loss": 0.5105,
+      "step": 1839
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.8655653309673776e-06,
+      "loss": 0.5434,
+      "step": 1840
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.855275502239998e-06,
+      "loss": 0.63,
+      "step": 1841
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.844993076224036e-06,
+      "loss": 0.3962,
+      "step": 1842
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.834718067714813e-06,
+      "loss": 0.633,
+      "step": 1843
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.824450491496968e-06,
+      "loss": 0.4985,
+      "step": 1844
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.814190362344454e-06,
+      "loss": 0.4852,
+      "step": 1845
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.803937695020507e-06,
+      "loss": 0.5127,
+      "step": 1846
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.793692504277625e-06,
+      "loss": 0.4956,
+      "step": 1847
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.7834548048575395e-06,
+      "loss": 0.5094,
+      "step": 1848
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.773224611491225e-06,
+      "loss": 0.4584,
+      "step": 1849
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.763001938898832e-06,
+      "loss": 0.4966,
+      "step": 1850
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.752786801789703e-06,
+      "loss": 0.2895,
+      "step": 1851
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.7425792148623315e-06,
+      "loss": 0.5695,
+      "step": 1852
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.732379192804356e-06,
+      "loss": 0.4572,
+      "step": 1853
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.722186750292511e-06,
+      "loss": 0.5207,
+      "step": 1854
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.712001901992652e-06,
+      "loss": 0.5576,
+      "step": 1855
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.701824662559682e-06,
+      "loss": 0.4165,
+      "step": 1856
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.6916550466375686e-06,
+      "loss": 0.3757,
+      "step": 1857
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.68149306885931e-06,
+      "loss": 0.4993,
+      "step": 1858
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.671338743846914e-06,
+      "loss": 0.6223,
+      "step": 1859
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.661192086211366e-06,
+      "loss": 0.4497,
+      "step": 1860
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.651053110552631e-06,
+      "loss": 0.5236,
+      "step": 1861
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.640921831459623e-06,
+      "loss": 0.6135,
+      "step": 1862
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.630798263510162e-06,
+      "loss": 0.3363,
+      "step": 1863
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.620682421271e-06,
+      "loss": 0.5307,
+      "step": 1864
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.610574319297748e-06,
+      "loss": 0.5222,
+      "step": 1865
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.6004739721348945e-06,
+      "loss": 0.4793,
+      "step": 1866
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.590381394315763e-06,
+      "loss": 0.6021,
+      "step": 1867
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.580296600362499e-06,
+      "loss": 0.457,
+      "step": 1868
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.570219604786051e-06,
+      "loss": 0.4668,
+      "step": 1869
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.560150422086147e-06,
+      "loss": 0.6156,
+      "step": 1870
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.550089066751258e-06,
+      "loss": 0.442,
+      "step": 1871
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.54003555325862e-06,
+      "loss": 0.48,
+      "step": 1872
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.529989896074157e-06,
+      "loss": 0.5035,
+      "step": 1873
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.519952109652508e-06,
+      "loss": 0.4305,
+      "step": 1874
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.509922208436981e-06,
+      "loss": 0.4652,
+      "step": 1875
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.499900206859531e-06,
+      "loss": 0.3652,
+      "step": 1876
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.489886119340756e-06,
+      "loss": 0.4291,
+      "step": 1877
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.479879960289863e-06,
+      "loss": 0.4019,
+      "step": 1878
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.469881744104655e-06,
+      "loss": 0.4098,
+      "step": 1879
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.45989148517149e-06,
+      "loss": 0.4431,
+      "step": 1880
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.449909197865303e-06,
+      "loss": 0.527,
+      "step": 1881
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.439934896549533e-06,
+      "loss": 0.526,
+      "step": 1882
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.429968595576142e-06,
+      "loss": 0.629,
+      "step": 1883
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.420010309285577e-06,
+      "loss": 0.5178,
+      "step": 1884
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.410060052006758e-06,
+      "loss": 0.5747,
+      "step": 1885
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.400117838057037e-06,
+      "loss": 0.4751,
+      "step": 1886
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.390183681742213e-06,
+      "loss": 0.5936,
+      "step": 1887
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.380257597356472e-06,
+      "loss": 0.497,
+      "step": 1888
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.370339599182398e-06,
+      "loss": 0.5794,
+      "step": 1889
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.360429701490935e-06,
+      "loss": 0.4616,
+      "step": 1890
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.350527918541375e-06,
+      "loss": 0.4806,
+      "step": 1891
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.340634264581324e-06,
+      "loss": 0.5318,
+      "step": 1892
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.330748753846701e-06,
+      "loss": 0.4785,
+      "step": 1893
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.320871400561709e-06,
+      "loss": 0.4845,
+      "step": 1894
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.311002218938798e-06,
+      "loss": 0.5147,
+      "step": 1895
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.3011412231786845e-06,
+      "loss": 0.5566,
+      "step": 1896
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.291288427470278e-06,
+      "loss": 0.5167,
+      "step": 1897
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.281443845990719e-06,
+      "loss": 0.6136,
+      "step": 1898
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.271607492905303e-06,
+      "loss": 0.4092,
+      "step": 1899
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.261779382367499e-06,
+      "loss": 0.264,
+      "step": 1900
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.251959528518912e-06,
+      "loss": 0.594,
+      "step": 1901
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.242147945489272e-06,
+      "loss": 0.5525,
+      "step": 1902
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.232344647396393e-06,
+      "loss": 0.5498,
+      "step": 1903
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.222549648346191e-06,
+      "loss": 0.5671,
+      "step": 1904
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.212762962432619e-06,
+      "loss": 0.4547,
+      "step": 1905
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.20298460373768e-06,
+      "loss": 0.5505,
+      "step": 1906
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.193214586331397e-06,
+      "loss": 0.5993,
+      "step": 1907
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.183452924271776e-06,
+      "loss": 0.682,
+      "step": 1908
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.1736996316048175e-06,
+      "loss": 0.5257,
+      "step": 1909
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.163954722364471e-06,
+      "loss": 0.6067,
+      "step": 1910
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.154218210572627e-06,
+      "loss": 0.5996,
+      "step": 1911
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.144490110239081e-06,
+      "loss": 0.4725,
+      "step": 1912
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.1347704353615495e-06,
+      "loss": 0.4844,
+      "step": 1913
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.1250591999256e-06,
+      "loss": 0.533,
+      "step": 1914
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.11535641790467e-06,
+      "loss": 0.5501,
+      "step": 1915
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.105662103260032e-06,
+      "loss": 0.3915,
+      "step": 1916
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.095976269940777e-06,
+      "loss": 0.4941,
+      "step": 1917
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.086298931883779e-06,
+      "loss": 0.3854,
+      "step": 1918
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.0766301030137115e-06,
+      "loss": 0.5318,
+      "step": 1919
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.06696979724298e-06,
+      "loss": 0.6443,
+      "step": 1920
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.057318028471743e-06,
+      "loss": 0.4837,
+      "step": 1921
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.047674810587867e-06,
+      "loss": 0.618,
+      "step": 1922
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.0380401574669184e-06,
+      "loss": 0.454,
+      "step": 1923
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 0.5912,
+      "step": 1924
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.018796600954426e-06,
+      "loss": 0.4119,
+      "step": 1925
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.009187725252309e-06,
+      "loss": 0.3411,
+      "step": 1926
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.999587469691942e-06,
+      "loss": 0.5136,
+      "step": 1927
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.989995848087076e-06,
+      "loss": 0.6132,
+      "step": 1928
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.9804128742390225e-06,
+      "loss": 0.5663,
+      "step": 1929
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.970838561936675e-06,
+      "loss": 0.5092,
+      "step": 1930
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.961272924956442e-06,
+      "loss": 0.4968,
+      "step": 1931
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.951715977062261e-06,
+      "loss": 0.4702,
+      "step": 1932
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.942167732005562e-06,
+      "loss": 0.5701,
+      "step": 1933
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.93262820352526e-06,
+      "loss": 0.5161,
+      "step": 1934
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.923097405347709e-06,
+      "loss": 0.5579,
+      "step": 1935
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.91357535118673e-06,
+      "loss": 0.5606,
+      "step": 1936
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.904062054743536e-06,
+      "loss": 0.6098,
+      "step": 1937
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.89455752970675e-06,
+      "loss": 0.5762,
+      "step": 1938
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.885061789752379e-06,
+      "loss": 0.4924,
+      "step": 1939
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.875574848543774e-06,
+      "loss": 0.5968,
+      "step": 1940
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.866096719731639e-06,
+      "loss": 0.4493,
+      "step": 1941
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.856627416953993e-06,
+      "loss": 0.6,
+      "step": 1942
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.847166953836159e-06,
+      "loss": 0.4134,
+      "step": 1943
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.837715343990727e-06,
+      "loss": 0.5296,
+      "step": 1944
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.828272601017572e-06,
+      "loss": 0.4535,
+      "step": 1945
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.8188387385037876e-06,
+      "loss": 0.5743,
+      "step": 1946
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.809413770023701e-06,
+      "loss": 0.6378,
+      "step": 1947
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.7999977091388394e-06,
+      "loss": 0.4843,
+      "step": 1948
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.790590569397917e-06,
+      "loss": 0.4254,
+      "step": 1949
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.7811923643367974e-06,
+      "loss": 0.3893,
+      "step": 1950
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.771803107478512e-06,
+      "loss": 0.5658,
+      "step": 1951
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.7624228123331918e-06,
+      "loss": 0.4996,
+      "step": 1952
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.753051492398089e-06,
+      "loss": 0.4743,
+      "step": 1953
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.743689161157533e-06,
+      "loss": 0.3865,
+      "step": 1954
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.734335832082927e-06,
+      "loss": 0.4919,
+      "step": 1955
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.724991518632717e-06,
+      "loss": 0.5259,
+      "step": 1956
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.7156562342523706e-06,
+      "loss": 0.601,
+      "step": 1957
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.7063299923743733e-06,
+      "loss": 0.5351,
+      "step": 1958
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.697012806418194e-06,
+      "loss": 0.5128,
+      "step": 1959
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.687704689790277e-06,
+      "loss": 0.6018,
+      "step": 1960
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.6784056558840022e-06,
+      "loss": 0.576,
+      "step": 1961
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.669115718079702e-06,
+      "loss": 0.5555,
+      "step": 1962
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.659834889744601e-06,
+      "loss": 0.5483,
+      "step": 1963
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.650563184232827e-06,
+      "loss": 0.5338,
+      "step": 1964
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.6413006148853782e-06,
+      "loss": 0.2774,
+      "step": 1965
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.6320471950301085e-06,
+      "loss": 0.5548,
+      "step": 1966
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.6228029379816974e-06,
+      "loss": 0.4426,
+      "step": 1967
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.61356785704166e-06,
+      "loss": 0.4655,
+      "step": 1968
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.604341965498285e-06,
+      "loss": 0.516,
+      "step": 1969
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.595125276626653e-06,
+      "loss": 0.5952,
+      "step": 1970
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.585917803688603e-06,
+      "loss": 0.4936,
+      "step": 1971
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.576719559932703e-06,
+      "loss": 0.4798,
+      "step": 1972
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.5675305585942498e-06,
+      "loss": 0.4686,
+      "step": 1973
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.5583508128952384e-06,
+      "loss": 0.5221,
+      "step": 1974
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.549180336044352e-06,
+      "loss": 0.5524,
+      "step": 1975
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.54001914123692e-06,
+      "loss": 0.5454,
+      "step": 1976
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.5308672416549416e-06,
+      "loss": 0.5083,
+      "step": 1977
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.5217246504670132e-06,
+      "loss": 0.5052,
+      "step": 1978
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.5125913808283653e-06,
+      "loss": 0.5314,
+      "step": 1979
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.5034674458807893e-06,
+      "loss": 0.58,
+      "step": 1980
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.4943528587526644e-06,
+      "loss": 0.4934,
+      "step": 1981
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.4852476325589093e-06,
+      "loss": 0.4839,
+      "step": 1982
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.4761517804009793e-06,
+      "loss": 0.6444,
+      "step": 1983
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.4670653153668333e-06,
+      "loss": 0.5923,
+      "step": 1984
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.4579882505309314e-06,
+      "loss": 0.3948,
+      "step": 1985
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.448920598954203e-06,
+      "loss": 0.5441,
+      "step": 1986
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.4398623736840355e-06,
+      "loss": 0.6278,
+      "step": 1987
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.430813587754256e-06,
+      "loss": 0.6415,
+      "step": 1988
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.421774254185096e-06,
+      "loss": 0.4469,
+      "step": 1989
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.412744385983201e-06,
+      "loss": 0.6564,
+      "step": 1990
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.403723996141589e-06,
+      "loss": 0.4057,
+      "step": 1991
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.394713097639647e-06,
+      "loss": 0.4378,
+      "step": 1992
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.3857117034430874e-06,
+      "loss": 0.5718,
+      "step": 1993
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.3767198265039748e-06,
+      "loss": 0.5799,
+      "step": 1994
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.367737479760652e-06,
+      "loss": 0.4721,
+      "step": 1995
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.3587646761377645e-06,
+      "loss": 0.6112,
+      "step": 1996
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.3498014285462242e-06,
+      "loss": 0.5939,
+      "step": 1997
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.3408477498831917e-06,
+      "loss": 0.4795,
+      "step": 1998
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.3319036530320503e-06,
+      "loss": 0.5734,
+      "step": 1999
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.3229691508624162e-06,
+      "loss": 0.6154,
+      "step": 2000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.31404425623008e-06,
+      "loss": 0.4256,
+      "step": 2001
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.3051289819770173e-06,
+      "loss": 0.4481,
+      "step": 2002
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.2962233409313625e-06,
+      "loss": 0.5806,
+      "step": 2003
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.2873273459073805e-06,
+      "loss": 0.5114,
+      "step": 2004
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.2784410097054665e-06,
+      "loss": 0.5005,
+      "step": 2005
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.2695643451121105e-06,
+      "loss": 0.3769,
+      "step": 2006
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.2606973648998918e-06,
+      "loss": 0.4531,
+      "step": 2007
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.25184008182745e-06,
+      "loss": 0.4174,
+      "step": 2008
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.2429925086394765e-06,
+      "loss": 0.5088,
+      "step": 2009
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.2341546580666792e-06,
+      "loss": 0.4941,
+      "step": 2010
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.2253265428257985e-06,
+      "loss": 0.484,
+      "step": 2011
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.2165081756195425e-06,
+      "loss": 0.5142,
+      "step": 2012
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.2076995691366076e-06,
+      "loss": 0.4836,
+      "step": 2013
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.198900736051642e-06,
+      "loss": 0.5808,
+      "step": 2014
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.190111689025235e-06,
+      "loss": 0.4752,
+      "step": 2015
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1813324407038826e-06,
+      "loss": 0.4911,
+      "step": 2016
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.172563003719993e-06,
+      "loss": 0.5001,
+      "step": 2017
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.163803390691853e-06,
+      "loss": 0.5681,
+      "step": 2018
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1550536142236143e-06,
+      "loss": 0.5401,
+      "step": 2019
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.146313686905279e-06,
+      "loss": 0.4857,
+      "step": 2020
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 0.4576,
+      "step": 2021
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.128863430007414e-06,
+      "loss": 0.4109,
+      "step": 2022
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.120153125536951e-06,
+      "loss": 0.4864,
+      "step": 2023
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.1114527204344836e-06,
+      "loss": 0.2708,
+      "step": 2024
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.1027622272189572e-06,
+      "loss": 0.4846,
+      "step": 2025
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.0940816583950815e-06,
+      "loss": 0.5255,
+      "step": 2026
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.085411026453263e-06,
+      "loss": 0.5102,
+      "step": 2027
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.076750343869621e-06,
+      "loss": 0.4153,
+      "step": 2028
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.068099623105958e-06,
+      "loss": 0.5035,
+      "step": 2029
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.059458876609742e-06,
+      "loss": 0.5119,
+      "step": 2030
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.050828116814081e-06,
+      "loss": 0.507,
+      "step": 2031
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.042207356137733e-06,
+      "loss": 0.5349,
+      "step": 2032
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.033596606985044e-06,
+      "loss": 0.3468,
+      "step": 2033
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.024995881745972e-06,
+      "loss": 0.5001,
+      "step": 2034
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.016405192796049e-06,
+      "loss": 0.3434,
+      "step": 2035
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.007824552496351e-06,
+      "loss": 0.536,
+      "step": 2036
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.999253973193522e-06,
+      "loss": 0.5076,
+      "step": 2037
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.990693467219704e-06,
+      "loss": 0.654,
+      "step": 2038
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.982143046892557e-06,
+      "loss": 0.5562,
+      "step": 2039
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 2.9736027245152276e-06,
+      "loss": 0.6514,
+      "step": 2040
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.9650725123763334e-06,
+      "loss": 0.5659,
+      "step": 2041
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.9565524227499344e-06,
+      "loss": 0.6007,
+      "step": 2042
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 2.948042467895544e-06,
+      "loss": 0.5178,
+      "step": 2043
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.9395426600580757e-06,
+      "loss": 0.5075,
+      "step": 2044
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.9310530114678503e-06,
+      "loss": 0.4928,
+      "step": 2045
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.922573534340569e-06,
+      "loss": 0.4092,
+      "step": 2046
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 2.9141042408773024e-06,
+      "loss": 0.5266,
+      "step": 2047
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.905645143264455e-06,
+      "loss": 0.5522,
+      "step": 2048
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.897196253673773e-06,
+      "loss": 0.5536,
+      "step": 2049
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.8887575842623097e-06,
+      "loss": 0.6138,
+      "step": 2050
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 2.8803291471724124e-06,
+      "loss": 0.4858,
+      "step": 2051
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.8719109545317102e-06,
+      "loss": 0.309,
+      "step": 2052
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.8635030184530787e-06,
+      "loss": 0.4205,
+      "step": 2053
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 2.8551053510346483e-06,
+      "loss": 0.5657,
+      "step": 2054
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.8467179643597698e-06,
+      "loss": 0.4521,
+      "step": 2055
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.838340870497002e-06,
+      "loss": 0.4258,
+      "step": 2056
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.8299740815000843e-06,
+      "loss": 0.4769,
+      "step": 2057
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.8216176094079485e-06,
+      "loss": 0.5393,
+      "step": 2058
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.813271466244656e-06,
+      "loss": 0.3981,
+      "step": 2059
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 2.8049356640194316e-06,
+      "loss": 0.5233,
+      "step": 2060
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.7966102147265993e-06,
+      "loss": 0.6392,
+      "step": 2061
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.7882951303455996e-06,
+      "loss": 0.5395,
+      "step": 2062
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 2.7799904228409536e-06,
+      "loss": 0.6822,
+      "step": 2063
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.7716961041622537e-06,
+      "loss": 0.4285,
+      "step": 2064
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.7634121862441386e-06,
+      "loss": 0.5041,
+      "step": 2065
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.7551386810062877e-06,
+      "loss": 0.5429,
+      "step": 2066
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 2.746875600353398e-06,
+      "loss": 0.5375,
+      "step": 2067
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.738622956175153e-06,
+      "loss": 0.5976,
+      "step": 2068
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.7303807603462453e-06,
+      "loss": 0.4628,
+      "step": 2069
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 2.722149024726307e-06,
+      "loss": 0.5319,
+      "step": 2070
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.7139277611599336e-06,
+      "loss": 0.4841,
+      "step": 2071
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.7057169814766494e-06,
+      "loss": 0.522,
+      "step": 2072
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.697516697490896e-06,
+      "loss": 0.525,
+      "step": 2073
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.6893269210020002e-06,
+      "loss": 0.4858,
+      "step": 2074
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.6811476637941923e-06,
+      "loss": 0.4702,
+      "step": 2075
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 2.6729789376365455e-06,
+      "loss": 0.4272,
+      "step": 2076
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.6648207542829886e-06,
+      "loss": 0.507,
+      "step": 2077
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.656673125472282e-06,
+      "loss": 0.315,
+      "step": 2078
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 2.648536062927999e-06,
+      "loss": 0.5598,
+      "step": 2079
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.6404095783585003e-06,
+      "loss": 0.5304,
+      "step": 2080
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.632293683456938e-06,
+      "loss": 0.5674,
+      "step": 2081
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.624188389901221e-06,
+      "loss": 0.5694,
+      "step": 2082
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 2.616093709354004e-06,
+      "loss": 0.523,
+      "step": 2083
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.6080096534626753e-06,
+      "loss": 0.4307,
+      "step": 2084
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.599936233859326e-06,
+      "loss": 0.5054,
+      "step": 2085
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.59187346216075e-06,
+      "loss": 0.3567,
+      "step": 2086
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 2.5838213499684183e-06,
+      "loss": 0.6313,
+      "step": 2087
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.5757799088684654e-06,
+      "loss": 0.572,
+      "step": 2088
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.56774915043167e-06,
+      "loss": 0.5965,
+      "step": 2089
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.5597290862134405e-06,
+      "loss": 0.5127,
+      "step": 2090
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.5517197277537888e-06,
+      "loss": 0.5472,
+      "step": 2091
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 2.543721086577341e-06,
+      "loss": 0.564,
+      "step": 2092
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.5357331741932823e-06,
+      "loss": 0.3761,
+      "step": 2093
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.527756002095373e-06,
+      "loss": 0.464,
+      "step": 2094
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.5197895817619157e-06,
+      "loss": 0.5115,
+      "step": 2095
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 2.5118339246557433e-06,
+      "loss": 0.5006,
+      "step": 2096
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.5038890422241958e-06,
+      "loss": 0.3933,
+      "step": 2097
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.495954945899116e-06,
+      "loss": 0.4856,
+      "step": 2098
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 2.488031647096826e-06,
+      "loss": 0.3672,
+      "step": 2099
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.480119157218108e-06,
+      "loss": 0.5877,
+      "step": 2100
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.4722174876481986e-06,
+      "loss": 0.4941,
+      "step": 2101
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.4643266497567543e-06,
+      "loss": 0.465,
+      "step": 2102
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 2.4564466548978527e-06,
+      "loss": 0.5905,
+      "step": 2103
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.4485775144099687e-06,
+      "loss": 0.5601,
+      "step": 2104
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.440719239615963e-06,
+      "loss": 0.6052,
+      "step": 2105
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 2.432871841823047e-06,
+      "loss": 0.484,
+      "step": 2106
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.4250353323228037e-06,
+      "loss": 0.4917,
+      "step": 2107
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.4172097223911296e-06,
+      "loss": 0.5809,
+      "step": 2108
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.4093950232882457e-06,
+      "loss": 0.4408,
+      "step": 2109
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.401591246258673e-06,
+      "loss": 0.5988,
+      "step": 2110
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.393798402531222e-06,
+      "loss": 0.527,
+      "step": 2111
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 2.3860165033189587e-06,
+      "loss": 0.4408,
+      "step": 2112
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.3782455598192123e-06,
+      "loss": 0.5742,
+      "step": 2113
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.370485583213542e-06,
+      "loss": 0.5347,
+      "step": 2114
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 2.362736584667731e-06,
+      "loss": 0.5095,
+      "step": 2115
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.354998575331765e-06,
+      "loss": 0.4175,
+      "step": 2116
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.3472715663398105e-06,
+      "loss": 0.6153,
+      "step": 2117
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.4268,
+      "step": 2118
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 2.3318505938454883e-06,
+      "loss": 0.3355,
+      "step": 2119
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.3241566525322556e-06,
+      "loss": 0.4722,
+      "step": 2120
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.3164737559412852e-06,
+      "loss": 0.5674,
+      "step": 2121
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.3088019151274523e-06,
+      "loss": 0.5784,
+      "step": 2122
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 2.3011411411297103e-06,
+      "loss": 0.4453,
+      "step": 2123
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.293491444971109e-06,
+      "loss": 0.3506,
+      "step": 2124
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.285852837658741e-06,
+      "loss": 0.4102,
+      "step": 2125
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.278225330183751e-06,
+      "loss": 0.5733,
+      "step": 2126
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.2706089335213124e-06,
+      "loss": 0.6658,
+      "step": 2127
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 2.2630036586306128e-06,
+      "loss": 0.5735,
+      "step": 2128
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.255409516454826e-06,
+      "loss": 0.539,
+      "step": 2129
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.247826517921121e-06,
+      "loss": 0.4474,
+      "step": 2130
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.240254673940625e-06,
+      "loss": 0.5869,
+      "step": 2131
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 2.2326939954084157e-06,
+      "loss": 0.568,
+      "step": 2132
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.2251444932035094e-06,
+      "loss": 0.4233,
+      "step": 2133
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.217606178188829e-06,
+      "loss": 0.4623,
+      "step": 2134
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 2.210079061211213e-06,
+      "loss": 0.4688,
+      "step": 2135
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.2025631531013826e-06,
+      "loss": 0.574,
+      "step": 2136
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.1950584646739304e-06,
+      "loss": 0.4434,
+      "step": 2137
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.1875650067273e-06,
+      "loss": 0.6164,
+      "step": 2138
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 2.180082790043789e-06,
+      "loss": 0.5951,
+      "step": 2139
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.1726118253895036e-06,
+      "loss": 0.5557,
+      "step": 2140
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.1651521235143715e-06,
+      "loss": 0.4268,
+      "step": 2141
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.157703695152109e-06,
+      "loss": 0.52,
+      "step": 2142
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.1502665510202135e-06,
+      "loss": 0.5295,
+      "step": 2143
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 2.142840701819946e-06,
+      "loss": 0.5069,
+      "step": 2144
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.135426158236309e-06,
+      "loss": 0.6139,
+      "step": 2145
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.128022930938045e-06,
+      "loss": 0.4693,
+      "step": 2146
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.12063103057761e-06,
+      "loss": 0.4624,
+      "step": 2147
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 2.1132504677911658e-06,
+      "loss": 0.6366,
+      "step": 2148
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.1058812531985506e-06,
+      "loss": 0.5295,
+      "step": 2149
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.098523397403288e-06,
+      "loss": 0.3776,
+      "step": 2150
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 2.091176910992545e-06,
+      "loss": 0.419,
+      "step": 2151
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.0838418045371357e-06,
+      "loss": 0.6023,
+      "step": 2152
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.0765180885914983e-06,
+      "loss": 0.517,
+      "step": 2153
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0692057736936832e-06,
+      "loss": 0.5677,
+      "step": 2154
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 2.0619048703653268e-06,
+      "loss": 0.4229,
+      "step": 2155
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.0546153891116627e-06,
+      "loss": 0.5309,
+      "step": 2156
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.0473373404214724e-06,
+      "loss": 0.6318,
+      "step": 2157
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.0400707347670943e-06,
+      "loss": 0.3504,
+      "step": 2158
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 2.032815582604404e-06,
+      "loss": 0.5897,
+      "step": 2159
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.025571894372794e-06,
+      "loss": 0.515,
+      "step": 2160
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.018339680495157e-06,
+      "loss": 0.4307,
+      "step": 2161
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.011118951377883e-06,
+      "loss": 0.5561,
+      "step": 2162
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.0039097174108313e-06,
+      "loss": 0.475,
+      "step": 2163
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.9967119889673258e-06,
+      "loss": 0.5594,
+      "step": 2164
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.989525776404132e-06,
+      "loss": 0.6437,
+      "step": 2165
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.9823510900614418e-06,
+      "loss": 0.5357,
+      "step": 2166
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9751879402628682e-06,
+      "loss": 0.4903,
+      "step": 2167
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.9680363373154233e-06,
+      "loss": 0.5603,
+      "step": 2168
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.9608962915095e-06,
+      "loss": 0.5902,
+      "step": 2169
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.9537678131188674e-06,
+      "loss": 0.5124,
+      "step": 2170
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.9466509124006506e-06,
+      "loss": 0.3902,
+      "step": 2171
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.939545599595304e-06,
+      "loss": 0.5341,
+      "step": 2172
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.932451884926627e-06,
+      "loss": 0.5622,
+      "step": 2173
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9253697786017132e-06,
+      "loss": 0.3825,
+      "step": 2174
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.9182992908109643e-06,
+      "loss": 0.4651,
+      "step": 2175
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.9112404317280606e-06,
+      "loss": 0.6085,
+      "step": 2176
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.904193211509947e-06,
+      "loss": 0.4398,
+      "step": 2177
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.897157640296825e-06,
+      "loss": 0.5493,
+      "step": 2178
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.8901337282121335e-06,
+      "loss": 0.5544,
+      "step": 2179
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.8831214853625379e-06,
+      "loss": 0.373,
+      "step": 2180
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.8761209218379018e-06,
+      "loss": 0.5782,
+      "step": 2181
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.8691320477113028e-06,
+      "loss": 0.5488,
+      "step": 2182
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.8621548730389782e-06,
+      "loss": 0.4679,
+      "step": 2183
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.855189407860344e-06,
+      "loss": 0.4274,
+      "step": 2184
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.8482356621979646e-06,
+      "loss": 0.4766,
+      "step": 2185
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.8412936460575414e-06,
+      "loss": 0.5135,
+      "step": 2186
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.8343633694278895e-06,
+      "loss": 0.4751,
+      "step": 2187
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.827444842280952e-06,
+      "loss": 0.435,
+      "step": 2188
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.820538074571745e-06,
+      "loss": 0.5833,
+      "step": 2189
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.813643076238375e-06,
+      "loss": 0.4499,
+      "step": 2190
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.8067598572020129e-06,
+      "loss": 0.4491,
+      "step": 2191
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.7998884273668794e-06,
+      "loss": 0.4318,
+      "step": 2192
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.7930287966202264e-06,
+      "loss": 0.5057,
+      "step": 2193
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.7861809748323367e-06,
+      "loss": 0.5472,
+      "step": 2194
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.779344971856497e-06,
+      "loss": 0.4328,
+      "step": 2195
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.7725207975289883e-06,
+      "loss": 0.5712,
+      "step": 2196
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.7657084616690745e-06,
+      "loss": 0.4816,
+      "step": 2197
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.758907974078975e-06,
+      "loss": 0.5778,
+      "step": 2198
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.7521193445438788e-06,
+      "loss": 0.4946,
+      "step": 2199
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.7453425828318936e-06,
+      "loss": 0.3795,
+      "step": 2200
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.7385776986940607e-06,
+      "loss": 0.598,
+      "step": 2201
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.7318247018643309e-06,
+      "loss": 0.4696,
+      "step": 2202
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.7250836020595486e-06,
+      "loss": 0.5067,
+      "step": 2203
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.7183544089794334e-06,
+      "loss": 0.5969,
+      "step": 2204
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.7116371323065883e-06,
+      "loss": 0.6019,
+      "step": 2205
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.7049317817064525e-06,
+      "loss": 0.6029,
+      "step": 2206
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.698238366827315e-06,
+      "loss": 0.6695,
+      "step": 2207
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.6915568973002904e-06,
+      "loss": 0.5465,
+      "step": 2208
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.6848873827392975e-06,
+      "loss": 0.3722,
+      "step": 2209
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.6782298327410618e-06,
+      "loss": 0.4541,
+      "step": 2210
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.6715842568850894e-06,
+      "loss": 0.487,
+      "step": 2211
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.6649506647336588e-06,
+      "loss": 0.4959,
+      "step": 2212
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.6583290658317963e-06,
+      "loss": 0.6291,
+      "step": 2213
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.6517194697072903e-06,
+      "loss": 0.4852,
+      "step": 2214
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 0.5107,
+      "step": 2215
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.638536323815062e-06,
+      "loss": 0.5965,
+      "step": 2216
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.631962793016487e-06,
+      "loss": 0.3993,
+      "step": 2217
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.6254013029335281e-06,
+      "loss": 0.5965,
+      "step": 2218
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.6188518630074635e-06,
+      "loss": 0.5219,
+      "step": 2219
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.6123144826622506e-06,
+      "loss": 0.4542,
+      "step": 2220
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.6057891713044772e-06,
+      "loss": 0.3853,
+      "step": 2221
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.5992759383233748e-06,
+      "loss": 0.4993,
+      "step": 2222
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.5927747930907921e-06,
+      "loss": 0.3958,
+      "step": 2223
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.5862857449611869e-06,
+      "loss": 0.4549,
+      "step": 2224
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.5798088032716119e-06,
+      "loss": 0.3574,
+      "step": 2225
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.5733439773416914e-06,
+      "loss": 0.4877,
+      "step": 2226
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.5668912764736256e-06,
+      "loss": 0.3516,
+      "step": 2227
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.560450709952165e-06,
+      "loss": 0.5202,
+      "step": 2228
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.554022287044602e-06,
+      "loss": 0.4779,
+      "step": 2229
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.5476060170007457e-06,
+      "loss": 0.485,
+      "step": 2230
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.5412019090529362e-06,
+      "loss": 0.5879,
+      "step": 2231
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.534809972415998e-06,
+      "loss": 0.6138,
+      "step": 2232
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.528430216287249e-06,
+      "loss": 0.5651,
+      "step": 2233
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.5220626498464798e-06,
+      "loss": 0.3957,
+      "step": 2234
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.5157072822559438e-06,
+      "loss": 0.3371,
+      "step": 2235
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.5093641226603329e-06,
+      "loss": 0.5117,
+      "step": 2236
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.503033180186787e-06,
+      "loss": 0.5776,
+      "step": 2237
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.4967144639448539e-06,
+      "loss": 0.5289,
+      "step": 2238
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.490407983026496e-06,
+      "loss": 0.5688,
+      "step": 2239
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.4841137465060672e-06,
+      "loss": 0.4662,
+      "step": 2240
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.4778317634403082e-06,
+      "loss": 0.5435,
+      "step": 2241
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.4715620428683208e-06,
+      "loss": 0.5968,
+      "step": 2242
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.4653045938115662e-06,
+      "loss": 0.5313,
+      "step": 2243
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.4590594252738521e-06,
+      "loss": 0.5075,
+      "step": 2244
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.4528265462413038e-06,
+      "loss": 0.5422,
+      "step": 2245
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.4466059656823816e-06,
+      "loss": 0.4954,
+      "step": 2246
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.4403976925478314e-06,
+      "loss": 0.515,
+      "step": 2247
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.4342017357707017e-06,
+      "loss": 0.541,
+      "step": 2248
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.4280181042663143e-06,
+      "loss": 0.5263,
+      "step": 2249
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.4218468069322576e-06,
+      "loss": 0.5583,
+      "step": 2250
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.415687852648372e-06,
+      "loss": 0.4993,
+      "step": 2251
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.4095412502767402e-06,
+      "loss": 0.4432,
+      "step": 2252
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.4034070086616646e-06,
+      "loss": 0.3801,
+      "step": 2253
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.397285136629669e-06,
+      "loss": 0.5009,
+      "step": 2254
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.3911756429894762e-06,
+      "loss": 0.5833,
+      "step": 2255
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.3850785365319985e-06,
+      "loss": 0.5087,
+      "step": 2256
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.3789938260303237e-06,
+      "loss": 0.6055,
+      "step": 2257
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.3729215202397018e-06,
+      "loss": 0.4108,
+      "step": 2258
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.3668616278975343e-06,
+      "loss": 0.5359,
+      "step": 2259
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.3608141577233636e-06,
+      "loss": 0.5561,
+      "step": 2260
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.3547791184188574e-06,
+      "loss": 0.495,
+      "step": 2261
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.3487565186677897e-06,
+      "loss": 0.5059,
+      "step": 2262
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.3427463671360496e-06,
+      "loss": 0.4672,
+      "step": 2263
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.3367486724715983e-06,
+      "loss": 0.5851,
+      "step": 2264
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.3307634433044847e-06,
+      "loss": 0.57,
+      "step": 2265
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.3247906882468153e-06,
+      "loss": 0.5445,
+      "step": 2266
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.3188304158927533e-06,
+      "loss": 0.4743,
+      "step": 2267
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.3128826348184886e-06,
+      "loss": 0.4602,
+      "step": 2268
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.3069473535822552e-06,
+      "loss": 0.3704,
+      "step": 2269
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.3010245807242849e-06,
+      "loss": 0.512,
+      "step": 2270
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.2951143247668197e-06,
+      "loss": 0.5091,
+      "step": 2271
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.28921659421409e-06,
+      "loss": 0.5451,
+      "step": 2272
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.2833313975523054e-06,
+      "loss": 0.507,
+      "step": 2273
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.2774587432496323e-06,
+      "loss": 0.4644,
+      "step": 2274
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.2715986397561997e-06,
+      "loss": 0.39,
+      "step": 2275
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.265751095504073e-06,
+      "loss": 0.5783,
+      "step": 2276
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.2599161189072428e-06,
+      "loss": 0.5307,
+      "step": 2277
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.2540937183616264e-06,
+      "loss": 0.4862,
+      "step": 2278
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.248283902245031e-06,
+      "loss": 0.4726,
+      "step": 2279
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.242486678917173e-06,
+      "loss": 0.3852,
+      "step": 2280
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.2367020567196341e-06,
+      "loss": 0.4661,
+      "step": 2281
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.2309300439758709e-06,
+      "loss": 0.5698,
+      "step": 2282
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.2251706489911986e-06,
+      "loss": 0.5749,
+      "step": 2283
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.2194238800527747e-06,
+      "loss": 0.4815,
+      "step": 2284
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.2136897454295837e-06,
+      "loss": 0.4548,
+      "step": 2285
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.207968253372438e-06,
+      "loss": 0.4467,
+      "step": 2286
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.2022594121139575e-06,
+      "loss": 0.5145,
+      "step": 2287
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.1965632298685559e-06,
+      "loss": 0.5405,
+      "step": 2288
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.1908797148324358e-06,
+      "loss": 0.5464,
+      "step": 2289
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.1852088751835688e-06,
+      "loss": 0.4327,
+      "step": 2290
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.1795507190816901e-06,
+      "loss": 0.4701,
+      "step": 2291
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.173905254668285e-06,
+      "loss": 0.6336,
+      "step": 2292
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1682724900665798e-06,
+      "loss": 0.557,
+      "step": 2293
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.1626524333815147e-06,
+      "loss": 0.4594,
+      "step": 2294
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.1570450926997657e-06,
+      "loss": 0.3567,
+      "step": 2295
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1514504760896894e-06,
+      "loss": 0.6327,
+      "step": 2296
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.1458685916013502e-06,
+      "loss": 0.3795,
+      "step": 2297
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.140299447266483e-06,
+      "loss": 0.4025,
+      "step": 2298
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.134743051098497e-06,
+      "loss": 0.5461,
+      "step": 2299
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.129199411092451e-06,
+      "loss": 0.3877,
+      "step": 2300
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.1236685352250599e-06,
+      "loss": 0.5645,
+      "step": 2301
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.1181504314546598e-06,
+      "loss": 0.545,
+      "step": 2302
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.1126451077212187e-06,
+      "loss": 0.543,
+      "step": 2303
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.1071525719463094e-06,
+      "loss": 0.5585,
+      "step": 2304
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.1016728320331093e-06,
+      "loss": 0.5666,
+      "step": 2305
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.0962058958663824e-06,
+      "loss": 0.4704,
+      "step": 2306
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.0907517713124637e-06,
+      "loss": 0.4111,
+      "step": 2307
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.0853104662192615e-06,
+      "loss": 0.6236,
+      "step": 2308
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.0798819884162349e-06,
+      "loss": 0.463,
+      "step": 2309
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.0744663457143878e-06,
+      "loss": 0.5458,
+      "step": 2310
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.0690635459062482e-06,
+      "loss": 0.5809,
+      "step": 2311
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 0.529,
+      "step": 2312
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.058296506048836e-06,
+      "loss": 0.6403,
+      "step": 2313
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.052932281492185e-06,
+      "loss": 0.4222,
+      "step": 2314
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.0475809308144747e-06,
+      "loss": 0.6065,
+      "step": 2315
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.042242461715729e-06,
+      "loss": 0.5285,
+      "step": 2316
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.0369168818774367e-06,
+      "loss": 0.5071,
+      "step": 2317
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.0316041989625403e-06,
+      "loss": 0.4835,
+      "step": 2318
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.026304420615426e-06,
+      "loss": 0.4314,
+      "step": 2319
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.0210175544619116e-06,
+      "loss": 0.5124,
+      "step": 2320
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.0157436081092375e-06,
+      "loss": 0.6277,
+      "step": 2321
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.010482589146048e-06,
+      "loss": 0.5102,
+      "step": 2322
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.0052345051423907e-06,
+      "loss": 0.4797,
+      "step": 2323
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 9.999993636497008e-07,
+      "loss": 0.539,
+      "step": 2324
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 9.947771722007915e-07,
+      "loss": 0.4904,
+      "step": 2325
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 9.895679383098366e-07,
+      "loss": 0.4717,
+      "step": 2326
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 9.84371669472376e-07,
+      "loss": 0.5118,
+      "step": 2327
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 9.79188373165283e-07,
+      "loss": 0.4027,
+      "step": 2328
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 9.740180568467716e-07,
+      "loss": 0.6603,
+      "step": 2329
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 9.688607279563767e-07,
+      "loss": 0.4726,
+      "step": 2330
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 9.637163939149485e-07,
+      "loss": 0.5911,
+      "step": 2331
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 9.585850621246306e-07,
+      "loss": 0.5906,
+      "step": 2332
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 9.534667399688702e-07,
+      "loss": 0.5787,
+      "step": 2333
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 9.48361434812386e-07,
+      "loss": 0.4401,
+      "step": 2334
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 9.432691540011674e-07,
+      "loss": 0.4264,
+      "step": 2335
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 9.381899048624677e-07,
+      "loss": 0.6046,
+      "step": 2336
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 9.331236947047839e-07,
+      "loss": 0.5007,
+      "step": 2337
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 9.280705308178584e-07,
+      "loss": 0.4453,
+      "step": 2338
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 9.230304204726503e-07,
+      "loss": 0.5084,
+      "step": 2339
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 9.180033709213454e-07,
+      "loss": 0.5132,
+      "step": 2340
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 9.129893893973329e-07,
+      "loss": 0.5876,
+      "step": 2341
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 9.07988483115202e-07,
+      "loss": 0.5382,
+      "step": 2342
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 9.030006592707175e-07,
+      "loss": 0.4553,
+      "step": 2343
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 8.980259250408363e-07,
+      "loss": 0.5863,
+      "step": 2344
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 8.93064287583667e-07,
+      "loss": 0.4959,
+      "step": 2345
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 8.881157540384777e-07,
+      "loss": 0.478,
+      "step": 2346
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 8.831803315256838e-07,
+      "loss": 0.6232,
+      "step": 2347
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 8.782580271468343e-07,
+      "loss": 0.4428,
+      "step": 2348
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 8.733488479845997e-07,
+      "loss": 0.638,
+      "step": 2349
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 8.68452801102766e-07,
+      "loss": 0.5537,
+      "step": 2350
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 8.635698935462267e-07,
+      "loss": 0.5278,
+      "step": 2351
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 8.587001323409639e-07,
+      "loss": 0.4413,
+      "step": 2352
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 8.538435244940502e-07,
+      "loss": 0.5466,
+      "step": 2353
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 8.490000769936224e-07,
+      "loss": 0.5978,
+      "step": 2354
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 8.441697968088891e-07,
+      "loss": 0.6553,
+      "step": 2355
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 8.393526908901095e-07,
+      "loss": 0.4071,
+      "step": 2356
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 8.34548766168588e-07,
+      "loss": 0.4802,
+      "step": 2357
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 8.297580295566576e-07,
+      "loss": 0.5701,
+      "step": 2358
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 8.249804879476842e-07,
+      "loss": 0.4479,
+      "step": 2359
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 8.202161482160354e-07,
+      "loss": 0.5587,
+      "step": 2360
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 8.154650172170975e-07,
+      "loss": 0.3483,
+      "step": 2361
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 8.107271017872364e-07,
+      "loss": 0.5112,
+      "step": 2362
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 8.060024087438112e-07,
+      "loss": 0.5722,
+      "step": 2363
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 8.012909448851514e-07,
+      "loss": 0.5994,
+      "step": 2364
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 7.965927169905551e-07,
+      "loss": 0.471,
+      "step": 2365
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 7.919077318202694e-07,
+      "loss": 0.4146,
+      "step": 2366
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 7.872359961154907e-07,
+      "loss": 0.321,
+      "step": 2367
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 7.825775165983506e-07,
+      "loss": 0.5789,
+      "step": 2368
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 7.779322999719041e-07,
+      "loss": 0.5408,
+      "step": 2369
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 7.733003529201277e-07,
+      "loss": 0.2634,
+      "step": 2370
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 7.686816821078958e-07,
+      "loss": 0.4758,
+      "step": 2371
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 7.640762941809865e-07,
+      "loss": 0.5499,
+      "step": 2372
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 7.594841957660637e-07,
+      "loss": 0.6166,
+      "step": 2373
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 7.549053934706707e-07,
+      "loss": 0.5658,
+      "step": 2374
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 7.503398938832107e-07,
+      "loss": 0.4324,
+      "step": 2375
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 7.457877035729588e-07,
+      "loss": 0.4668,
+      "step": 2376
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 7.412488290900288e-07,
+      "loss": 0.5875,
+      "step": 2377
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 7.367232769653798e-07,
+      "loss": 0.3672,
+      "step": 2378
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 7.322110537108007e-07,
+      "loss": 0.5419,
+      "step": 2379
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 7.277121658189001e-07,
+      "loss": 0.2979,
+      "step": 2380
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 7.232266197630988e-07,
+      "loss": 0.3919,
+      "step": 2381
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 7.187544219976205e-07,
+      "loss": 0.5527,
+      "step": 2382
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 7.142955789574824e-07,
+      "loss": 0.4703,
+      "step": 2383
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 7.098500970584854e-07,
+      "loss": 0.5132,
+      "step": 2384
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 7.054179826972074e-07,
+      "loss": 0.5922,
+      "step": 2385
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 7.00999242250986e-07,
+      "loss": 0.3158,
+      "step": 2386
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 6.965938820779205e-07,
+      "loss": 0.5729,
+      "step": 2387
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 6.922019085168552e-07,
+      "loss": 0.452,
+      "step": 2388
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 6.878233278873736e-07,
+      "loss": 0.5759,
+      "step": 2389
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 6.834581464897872e-07,
+      "loss": 0.5503,
+      "step": 2390
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 6.791063706051292e-07,
+      "loss": 0.4543,
+      "step": 2391
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 6.74768006495139e-07,
+      "loss": 0.5704,
+      "step": 2392
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 6.704430604022649e-07,
+      "loss": 0.5543,
+      "step": 2393
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 6.661315385496426e-07,
+      "loss": 0.4813,
+      "step": 2394
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 6.618334471410926e-07,
+      "loss": 0.4423,
+      "step": 2395
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 6.575487923611124e-07,
+      "loss": 0.4415,
+      "step": 2396
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 6.532775803748659e-07,
+      "loss": 0.5791,
+      "step": 2397
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 6.490198173281704e-07,
+      "loss": 0.5754,
+      "step": 2398
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 6.44775509347495e-07,
+      "loss": 0.4205,
+      "step": 2399
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 6.405446625399481e-07,
+      "loss": 0.2888,
+      "step": 2400
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 6.363272829932676e-07,
+      "loss": 0.5589,
+      "step": 2401
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 6.321233767758162e-07,
+      "loss": 0.5746,
+      "step": 2402
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 6.279329499365649e-07,
+      "loss": 0.4532,
+      "step": 2403
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 6.237560085050931e-07,
+      "loss": 0.5729,
+      "step": 2404
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 6.195925584915752e-07,
+      "loss": 0.4574,
+      "step": 2405
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 6.154426058867758e-07,
+      "loss": 0.4747,
+      "step": 2406
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 6.113061566620304e-07,
+      "loss": 0.5298,
+      "step": 2407
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 6.071832167692549e-07,
+      "loss": 0.4981,
+      "step": 2408
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.4562,
+      "step": 2409
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 5.989778886900433e-07,
+      "loss": 0.5588,
+      "step": 2410
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 5.948955123102029e-07,
+      "loss": 0.5704,
+      "step": 2411
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 5.908266688755049e-07,
+      "loss": 0.4594,
+      "step": 2412
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 5.867713642405748e-07,
+      "loss": 0.4247,
+      "step": 2413
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 5.827296042405717e-07,
+      "loss": 0.3708,
+      "step": 2414
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 5.787013946911546e-07,
+      "loss": 0.6572,
+      "step": 2415
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 5.746867413884904e-07,
+      "loss": 0.5209,
+      "step": 2416
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 5.706856501092384e-07,
+      "loss": 0.4844,
+      "step": 2417
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 5.666981266105398e-07,
+      "loss": 0.4164,
+      "step": 2418
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 5.627241766300228e-07,
+      "loss": 0.6067,
+      "step": 2419
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 5.587638058857736e-07,
+      "loss": 0.5134,
+      "step": 2420
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 5.54817020076347e-07,
+      "loss": 0.5353,
+      "step": 2421
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 5.50883824880748e-07,
+      "loss": 0.4501,
+      "step": 2422
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 5.469642259584274e-07,
+      "loss": 0.4858,
+      "step": 2423
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 5.430582289492658e-07,
+      "loss": 0.4631,
+      "step": 2424
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 5.391658394735855e-07,
+      "loss": 0.4468,
+      "step": 2425
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 5.352870631321139e-07,
+      "loss": 0.4024,
+      "step": 2426
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 5.314219055060022e-07,
+      "loss": 0.4871,
+      "step": 2427
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 5.275703721567982e-07,
+      "loss": 0.5959,
+      "step": 2428
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 5.237324686264522e-07,
+      "loss": 0.3665,
+      "step": 2429
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 5.199082004372958e-07,
+      "loss": 0.5026,
+      "step": 2430
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 5.160975730920459e-07,
+      "loss": 0.4523,
+      "step": 2431
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 5.123005920737889e-07,
+      "loss": 0.5654,
+      "step": 2432
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 5.085172628459778e-07,
+      "loss": 0.605,
+      "step": 2433
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 5.047475908524225e-07,
+      "loss": 0.4869,
+      "step": 2434
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 5.009915815172772e-07,
+      "loss": 0.5897,
+      "step": 2435
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.972492402450402e-07,
+      "loss": 0.4342,
+      "step": 2436
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.935205724205438e-07,
+      "loss": 0.4073,
+      "step": 2437
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.898055834089455e-07,
+      "loss": 0.4044,
+      "step": 2438
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.861042785557147e-07,
+      "loss": 0.4787,
+      "step": 2439
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.824166631866412e-07,
+      "loss": 0.5536,
+      "step": 2440
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.787427426078072e-07,
+      "loss": 0.6533,
+      "step": 2441
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.7508252210559656e-07,
+      "loss": 0.5773,
+      "step": 2442
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.714360069466761e-07,
+      "loss": 0.4163,
+      "step": 2443
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.678032023779933e-07,
+      "loss": 0.5687,
+      "step": 2444
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.6418411362676664e-07,
+      "loss": 0.5992,
+      "step": 2445
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.605787459004829e-07,
+      "loss": 0.56,
+      "step": 2446
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.56987104386879e-07,
+      "loss": 0.4879,
+      "step": 2447
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.534091942539476e-07,
+      "loss": 0.5894,
+      "step": 2448
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.498450206499216e-07,
+      "loss": 0.5731,
+      "step": 2449
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.4629458870326327e-07,
+      "loss": 0.4546,
+      "step": 2450
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.427579035226726e-07,
+      "loss": 0.581,
+      "step": 2451
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3923497019705996e-07,
+      "loss": 0.5973,
+      "step": 2452
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.357257937955517e-07,
+      "loss": 0.4946,
+      "step": 2453
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.322303793674798e-07,
+      "loss": 0.5687,
+      "step": 2454
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.2874873194237553e-07,
+      "loss": 0.3916,
+      "step": 2455
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.25280856529956e-07,
+      "loss": 0.6855,
+      "step": 2456
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.218267581201296e-07,
+      "loss": 0.4692,
+      "step": 2457
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.18386441682973e-07,
+      "loss": 0.393,
+      "step": 2458
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.1495991216873643e-07,
+      "loss": 0.6388,
+      "step": 2459
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.115471745078314e-07,
+      "loss": 0.6071,
+      "step": 2460
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.0814823361082665e-07,
+      "loss": 0.355,
+      "step": 2461
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.047630943684333e-07,
+      "loss": 0.6847,
+      "step": 2462
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.013917616515084e-07,
+      "loss": 0.3567,
+      "step": 2463
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.9803424031104045e-07,
+      "loss": 0.6204,
+      "step": 2464
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.9469053517814715e-07,
+      "loss": 0.6249,
+      "step": 2465
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.913606510640644e-07,
+      "loss": 0.483,
+      "step": 2466
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.880445927601395e-07,
+      "loss": 0.5608,
+      "step": 2467
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 3.84742365037829e-07,
+      "loss": 0.4465,
+      "step": 2468
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.8145397264868655e-07,
+      "loss": 0.4713,
+      "step": 2469
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.781794203243605e-07,
+      "loss": 0.4238,
+      "step": 2470
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.749187127765819e-07,
+      "loss": 0.4158,
+      "step": 2471
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.716718546971643e-07,
+      "loss": 0.5746,
+      "step": 2472
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.684388507579872e-07,
+      "loss": 0.4514,
+      "step": 2473
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 3.652197056110052e-07,
+      "loss": 0.5599,
+      "step": 2474
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.620144238882206e-07,
+      "loss": 0.5285,
+      "step": 2475
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.588230102016954e-07,
+      "loss": 0.5568,
+      "step": 2476
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 3.556454691435329e-07,
+      "loss": 0.615,
+      "step": 2477
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.5248180528588026e-07,
+      "loss": 0.5057,
+      "step": 2478
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.493320231809083e-07,
+      "loss": 0.5689,
+      "step": 2479
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.4619612736082274e-07,
+      "loss": 0.4655,
+      "step": 2480
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.430741223378431e-07,
+      "loss": 0.4295,
+      "step": 2481
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.399660126041993e-07,
+      "loss": 0.5116,
+      "step": 2482
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 3.3687180263213603e-07,
+      "loss": 0.6163,
+      "step": 2483
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.3379149687388866e-07,
+      "loss": 0.4607,
+      "step": 2484
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.307250997616906e-07,
+      "loss": 0.4547,
+      "step": 2485
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.2767261570776144e-07,
+      "loss": 0.367,
+      "step": 2486
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.246340491043021e-07,
+      "loss": 0.3583,
+      "step": 2487
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 3.21609404323483e-07,
+      "loss": 0.3929,
+      "step": 2488
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.1859868571745057e-07,
+      "loss": 0.3357,
+      "step": 2489
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.156018976183073e-07,
+      "loss": 0.5908,
+      "step": 2490
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.1261904433811165e-07,
+      "loss": 0.4482,
+      "step": 2491
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.0965013016887367e-07,
+      "loss": 0.5707,
+      "step": 2492
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 3.0669515938254404e-07,
+      "loss": 0.5459,
+      "step": 2493
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.037541362310115e-07,
+      "loss": 0.5678,
+      "step": 2494
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.008270649460965e-07,
+      "loss": 0.3733,
+      "step": 2495
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.979139497395422e-07,
+      "loss": 0.5834,
+      "step": 2496
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.9501479480301245e-07,
+      "loss": 0.6227,
+      "step": 2497
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.9212960430808346e-07,
+      "loss": 0.6357,
+      "step": 2498
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.892583824062334e-07,
+      "loss": 0.4968,
+      "step": 2499
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.864011332288519e-07,
+      "loss": 0.5247,
+      "step": 2500
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.8355786088721137e-07,
+      "loss": 0.6379,
+      "step": 2501
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.807285694724804e-07,
+      "loss": 0.4924,
+      "step": 2502
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.779132630557091e-07,
+      "loss": 0.5534,
+      "step": 2503
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.751119456878226e-07,
+      "loss": 0.4173,
+      "step": 2504
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.723246213996178e-07,
+      "loss": 0.4704,
+      "step": 2505
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 0.4049,
+      "step": 2506
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6679196808477414e-07,
+      "loss": 0.6277,
+      "step": 2507
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.640466470190317e-07,
+      "loss": 0.4574,
+      "step": 2508
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6131533495476037e-07,
+      "loss": 0.3772,
+      "step": 2509
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.585980358220297e-07,
+      "loss": 0.4194,
+      "step": 2510
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.5589475353073987e-07,
+      "loss": 0.5524,
+      "step": 2511
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.532054919706295e-07,
+      "loss": 0.6483,
+      "step": 2512
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.505302550112609e-07,
+      "loss": 0.4841,
+      "step": 2513
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.478690465020117e-07,
+      "loss": 0.6541,
+      "step": 2514
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.452218702720821e-07,
+      "loss": 0.4883,
+      "step": 2515
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.4258873013047415e-07,
+      "loss": 0.4429,
+      "step": 2516
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.3996962986599815e-07,
+      "loss": 0.4367,
+      "step": 2517
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.3736457324725958e-07,
+      "loss": 0.5326,
+      "step": 2518
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.3477356402266005e-07,
+      "loss": 0.5067,
+      "step": 2519
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.3219660592038285e-07,
+      "loss": 0.5257,
+      "step": 2520
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.296337026484008e-07,
+      "loss": 0.5198,
+      "step": 2521
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.2708485789445621e-07,
+      "loss": 0.525,
+      "step": 2522
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.2455007532606654e-07,
+      "loss": 0.6039,
+      "step": 2523
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.2202935859051422e-07,
+      "loss": 0.4956,
+      "step": 2524
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.1952271131484237e-07,
+      "loss": 0.5374,
+      "step": 2525
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.170301371058503e-07,
+      "loss": 0.4606,
+      "step": 2526
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.1455163955008685e-07,
+      "loss": 0.464,
+      "step": 2527
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.1208722221384704e-07,
+      "loss": 0.4369,
+      "step": 2528
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.0963688864316324e-07,
+      "loss": 0.5019,
+      "step": 2529
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.0720064236380843e-07,
+      "loss": 0.4947,
+      "step": 2530
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.0477848688127854e-07,
+      "loss": 0.506,
+      "step": 2531
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 2.0237042568080012e-07,
+      "loss": 0.6083,
+      "step": 2532
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.9997646222731482e-07,
+      "loss": 0.484,
+      "step": 2533
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.9759659996548385e-07,
+      "loss": 0.4563,
+      "step": 2534
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.952308423196736e-07,
+      "loss": 0.5228,
+      "step": 2535
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.9287919269396104e-07,
+      "loss": 0.4786,
+      "step": 2536
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.9054165447211504e-07,
+      "loss": 0.4977,
+      "step": 2537
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.8821823101760949e-07,
+      "loss": 0.5691,
+      "step": 2538
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.859089256736013e-07,
+      "loss": 0.6603,
+      "step": 2539
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.8361374176293468e-07,
+      "loss": 0.6128,
+      "step": 2540
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.8133268258813564e-07,
+      "loss": 0.536,
+      "step": 2541
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.7906575143140647e-07,
+      "loss": 0.5784,
+      "step": 2542
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.7681295155461909e-07,
+      "loss": 0.4651,
+      "step": 2543
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.7457428619931383e-07,
+      "loss": 0.5478,
+      "step": 2544
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.7234975858669179e-07,
+      "loss": 0.5968,
+      "step": 2545
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.701393719176103e-07,
+      "loss": 0.5035,
+      "step": 2546
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.6794312937258417e-07,
+      "loss": 0.3309,
+      "step": 2547
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.657610341117688e-07,
+      "loss": 0.5891,
+      "step": 2548
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.6359308927496932e-07,
+      "loss": 0.3752,
+      "step": 2549
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.6143929798162705e-07,
+      "loss": 0.5601,
+      "step": 2550
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.592996633308197e-07,
+      "loss": 0.6259,
+      "step": 2551
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.5717418840125231e-07,
+      "loss": 0.5015,
+      "step": 2552
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.550628762512596e-07,
+      "loss": 0.4567,
+      "step": 2553
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.5296572991879254e-07,
+      "loss": 0.5219,
+      "step": 2554
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.5088275242142403e-07,
+      "loss": 0.5704,
+      "step": 2555
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.4881394675633543e-07,
+      "loss": 0.503,
+      "step": 2556
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.467593159003189e-07,
+      "loss": 0.5776,
+      "step": 2557
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.4471886280977066e-07,
+      "loss": 0.4918,
+      "step": 2558
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.426925904206833e-07,
+      "loss": 0.5165,
+      "step": 2559
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.40680501648649e-07,
+      "loss": 0.4449,
+      "step": 2560
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.3868259938884964e-07,
+      "loss": 0.6099,
+      "step": 2561
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.3669888651605346e-07,
+      "loss": 0.4188,
+      "step": 2562
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.347293658846116e-07,
+      "loss": 0.478,
+      "step": 2563
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.327740403284572e-07,
+      "loss": 0.5007,
+      "step": 2564
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.30832912661093e-07,
+      "loss": 0.3631,
+      "step": 2565
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.2890598567559697e-07,
+      "loss": 0.3745,
+      "step": 2566
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.2699326214461238e-07,
+      "loss": 0.4761,
+      "step": 2567
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.2509474482034434e-07,
+      "loss": 0.613,
+      "step": 2568
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.2321043643455543e-07,
+      "loss": 0.5686,
+      "step": 2569
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.2134033969856907e-07,
+      "loss": 0.498,
+      "step": 2570
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.1948445730325164e-07,
+      "loss": 0.5437,
+      "step": 2571
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.1764279191902372e-07,
+      "loss": 0.512,
+      "step": 2572
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.1581534619584334e-07,
+      "loss": 0.3968,
+      "step": 2573
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.1400212276321377e-07,
+      "loss": 0.5473,
+      "step": 2574
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.1220312423016689e-07,
+      "loss": 0.5491,
+      "step": 2575
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.1041835318527206e-07,
+      "loss": 0.5277,
+      "step": 2576
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.0864781219662613e-07,
+      "loss": 0.5502,
+      "step": 2577
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.0689150381184676e-07,
+      "loss": 0.48,
+      "step": 2578
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.0514943055807803e-07,
+      "loss": 0.5813,
+      "step": 2579
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.0342159494197479e-07,
+      "loss": 0.4961,
+      "step": 2580
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.0170799944971165e-07,
+      "loss": 0.5255,
+      "step": 2581
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.0000864654696851e-07,
+      "loss": 0.5086,
+      "step": 2582
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 9.832353867893385e-08,
+      "loss": 0.5912,
+      "step": 2583
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 9.665267827029812e-08,
+      "loss": 0.4839,
+      "step": 2584
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 9.499606772525371e-08,
+      "loss": 0.5329,
+      "step": 2585
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 9.335370942748389e-08,
+      "loss": 0.5267,
+      "step": 2586
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 9.172560574017165e-08,
+      "loss": 0.409,
+      "step": 2587
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 9.01117590059819e-08,
+      "loss": 0.6163,
+      "step": 2588
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 8.851217154706826e-08,
+      "loss": 0.4484,
+      "step": 2589
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 8.69268456650696e-08,
+      "loss": 0.5637,
+      "step": 2590
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 8.535578364109898e-08,
+      "loss": 0.4701,
+      "step": 2591
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 8.379898773574924e-08,
+      "loss": 0.6227,
+      "step": 2592
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 8.225646018908185e-08,
+      "loss": 0.4928,
+      "step": 2593
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 8.072820322063246e-08,
+      "loss": 0.4743,
+      "step": 2594
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 7.921421902939874e-08,
+      "loss": 0.3708,
+      "step": 2595
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 7.771450979384366e-08,
+      "loss": 0.4379,
+      "step": 2596
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 7.622907767188881e-08,
+      "loss": 0.4079,
+      "step": 2597
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 7.475792480091226e-08,
+      "loss": 0.2852,
+      "step": 2598
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 7.33010532977485e-08,
+      "loss": 0.3908,
+      "step": 2599
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 7.185846525867956e-08,
+      "loss": 0.5849,
+      "step": 2600
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 7.043016275943614e-08,
+      "loss": 0.4955,
+      "step": 2601
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.901614785519429e-08,
+      "loss": 0.5392,
+      "step": 2602
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 0.4569,
+      "step": 2603
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 6.623098894962043e-08,
+      "loss": 0.6297,
+      "step": 2604
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 6.485984895583608e-08,
+      "loss": 0.5202,
+      "step": 2605
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 6.350300457214187e-08,
+      "loss": 0.5045,
+      "step": 2606
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 6.216045775089275e-08,
+      "loss": 0.4212,
+      "step": 2607
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 6.083221042387121e-08,
+      "loss": 0.4568,
+      "step": 2608
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 5.951826450228293e-08,
+      "loss": 0.4447,
+      "step": 2609
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 5.821862187675775e-08,
+      "loss": 0.5101,
+      "step": 2610
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 5.693328441734203e-08,
+      "loss": 0.4994,
+      "step": 2611
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 5.5662253973498557e-08,
+      "loss": 0.5676,
+      "step": 2612
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 5.4405532374107726e-08,
+      "loss": 0.531,
+      "step": 2613
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 5.31631214274575e-08,
+      "loss": 0.4999,
+      "step": 2614
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 5.193502292124342e-08,
+      "loss": 0.5033,
+      "step": 2615
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 5.0721238622569724e-08,
+      "loss": 0.5193,
+      "step": 2616
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.95217702779438e-08,
+      "loss": 0.456,
+      "step": 2617
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.833661961327174e-08,
+      "loss": 0.5454,
+      "step": 2618
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.716578833386054e-08,
+      "loss": 0.5116,
+      "step": 2619
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.600927812441036e-08,
+      "loss": 0.4251,
+      "step": 2620
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.486709064901673e-08,
+      "loss": 0.4272,
+      "step": 2621
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.373922755116722e-08,
+      "loss": 0.5777,
+      "step": 2622
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.262569045373588e-08,
+      "loss": 0.6041,
+      "step": 2623
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.152648095898437e-08,
+      "loss": 0.4982,
+      "step": 2624
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.044160064855751e-08,
+      "loss": 0.4704,
+      "step": 2625
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.9371051083484377e-08,
+      "loss": 0.5112,
+      "step": 2626
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 3.831483380417056e-08,
+      "loss": 0.5752,
+      "step": 2627
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 3.727295033040035e-08,
+      "loss": 0.5926,
+      "step": 2628
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.6245402161334545e-08,
+      "loss": 0.6019,
+      "step": 2629
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 3.523219077550488e-08,
+      "loss": 0.6306,
+      "step": 2630
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.4233317630814054e-08,
+      "loss": 0.3907,
+      "step": 2631
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 3.32487841645357e-08,
+      "loss": 0.5227,
+      "step": 2632
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.227859179330661e-08,
+      "loss": 0.527,
+      "step": 2633
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 3.132274191313345e-08,
+      "loss": 0.4539,
+      "step": 2634
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 3.038123589938047e-08,
+      "loss": 0.5033,
+      "step": 2635
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.9454075106776227e-08,
+      "loss": 0.5301,
+      "step": 2636
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 2.8541260869403565e-08,
+      "loss": 0.417,
+      "step": 2637
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.7642794500708504e-08,
+      "loss": 0.5622,
+      "step": 2638
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.6758677293486912e-08,
+      "loss": 0.5976,
+      "step": 2639
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.5888910519888953e-08,
+      "loss": 0.509,
+      "step": 2640
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.5033495431417976e-08,
+      "loss": 0.671,
+      "step": 2641
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.4192433258923842e-08,
+      "loss": 0.5233,
+      "step": 2642
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 2.336572521260738e-08,
+      "loss": 0.4956,
+      "step": 2643
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.2553372482010395e-08,
+      "loss": 0.3934,
+      "step": 2644
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.1755376236025637e-08,
+      "loss": 0.2172,
+      "step": 2645
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.0971737622883515e-08,
+      "loss": 0.4023,
+      "step": 2646
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 2.020245777015539e-08,
+      "loss": 0.3717,
+      "step": 2647
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.9447537784754722e-08,
+      "loss": 0.6244,
+      "step": 2648
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.870697875293148e-08,
+      "loss": 0.5551,
+      "step": 2649
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.798078174026885e-08,
+      "loss": 0.3859,
+      "step": 2650
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.726894779168986e-08,
+      "loss": 0.4841,
+      "step": 2651
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.6571477931446312e-08,
+      "loss": 0.4896,
+      "step": 2652
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.5888373163124304e-08,
+      "loss": 0.5419,
+      "step": 2653
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.521963446963759e-08,
+      "loss": 0.582,
+      "step": 2654
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.4565262813230896e-08,
+      "loss": 0.6236,
+      "step": 2655
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.3925259135474379e-08,
+      "loss": 0.5838,
+      "step": 2656
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.3299624357266949e-08,
+      "loss": 0.5648,
+      "step": 2657
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.2688359378829617e-08,
+      "loss": 0.5323,
+      "step": 2658
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.20914650797066e-08,
+      "loss": 0.5365,
+      "step": 2659
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.1508942318767536e-08,
+      "loss": 0.473,
+      "step": 2660
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.0940791934198613e-08,
+      "loss": 0.569,
+      "step": 2661
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.0387014743510338e-08,
+      "loss": 0.4457,
+      "step": 2662
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 9.847611543526426e-09,
+      "loss": 0.4818,
+      "step": 2663
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 9.322583110392692e-09,
+      "loss": 0.4055,
+      "step": 2664
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 8.811930199568164e-09,
+      "loss": 0.4983,
+      "step": 2665
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 8.315653545828417e-09,
+      "loss": 0.5427,
+      "step": 2666
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 7.833753863263349e-09,
+      "loss": 0.6144,
+      "step": 2667
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 7.366231845273852e-09,
+      "loss": 0.5384,
+      "step": 2668
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 6.913088164575144e-09,
+      "loss": 0.4303,
+      "step": 2669
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 6.474323473194543e-09,
+      "loss": 0.4073,
+      "step": 2670
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 6.049938402464817e-09,
+      "loss": 0.4523,
+      "step": 2671
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.639933563033051e-09,
+      "loss": 0.4,
+      "step": 2672
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 5.2443095448506674e-09,
+      "loss": 0.5475,
+      "step": 2673
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.8630669171811916e-09,
+      "loss": 0.514,
+      "step": 2674
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.496206228590261e-09,
+      "loss": 0.5936,
+      "step": 2675
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.143728006951176e-09,
+      "loss": 0.5397,
+      "step": 2676
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 3.8056327594448995e-09,
+      "loss": 0.6043,
+      "step": 2677
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.4819209725511783e-09,
+      "loss": 0.4452,
+      "step": 2678
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 3.172593112059641e-09,
+      "loss": 0.6547,
+      "step": 2679
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.8776496230598083e-09,
+      "loss": 0.3852,
+      "step": 2680
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.5970909299422032e-09,
+      "loss": 0.526,
+      "step": 2681
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 2.330917436402791e-09,
+      "loss": 0.4771,
+      "step": 2682
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.0791295254363186e-09,
+      "loss": 0.4857,
+      "step": 2683
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.8417275593385354e-09,
+      "loss": 0.4945,
+      "step": 2684
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.6187118797061918e-09,
+      "loss": 0.3156,
+      "step": 2685
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.4100828074348205e-09,
+      "loss": 0.4785,
+      "step": 2686
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.2158406427187352e-09,
+      "loss": 0.5132,
+      "step": 2687
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.035985665053252e-09,
+      "loss": 0.5567,
+      "step": 2688
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 8.705181332291368e-10,
+      "loss": 0.5699,
+      "step": 2689
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 7.194382853370486e-10,
+      "loss": 0.6044,
+      "step": 2690
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 5.827463387653165e-10,
+      "loss": 0.3864,
+      "step": 2691
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.6044249019883136e-10,
+      "loss": 0.4543,
+      "step": 2692
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 3.5252691562015497e-10,
+      "loss": 0.4158,
+      "step": 2693
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.5899977030729997e-10,
+      "loss": 0.5832,
+      "step": 2694
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.7986118883706048e-10,
+      "loss": 0.4981,
+      "step": 2695
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.1511128508057135e-10,
+      "loss": 0.4657,
+      "step": 2696
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 6.475015220552827e-11,
+      "loss": 0.4541,
+      "step": 2697
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 2.877786267840854e-11,
+      "loss": 0.5429,
+      "step": 2698
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 7.1944682578095615e-12,
+      "loss": 0.4102,
+      "step": 2699
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.0,
+      "loss": 0.498,
+      "step": 2700
+    },
+    {
+      "epoch": 15.0,
+      "step": 2700,
+      "total_flos": 5.095972157029089e+17,
+      "train_loss": 0.5729767621998434,
+      "train_runtime": 1180.905,
+      "train_samples_per_second": 18.291,
+      "train_steps_per_second": 2.286
+    }
+  ],
+  "max_steps": 2700,
+  "num_train_epochs": 15,
+  "total_flos": 5.095972157029089e+17,
+  "trial_name": null,
+  "trial_params": null
+}