diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15288 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999438265363443,
+  "eval_steps": 500,
+  "global_step": 2543,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.194805194805196e-07,
+      "loss": 4.333,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0389610389610392e-06,
+      "loss": 4.4929,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5584415584415584e-06,
+      "loss": 4.4299,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0779220779220784e-06,
+      "loss": 3.974,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.597402597402597e-06,
+      "loss": 3.9993,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.116883116883117e-06,
+      "loss": 3.0864,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.9034,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.155844155844157e-06,
+      "loss": 1.2156,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.675324675324676e-06,
+      "loss": 0.7811,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.194805194805194e-06,
+      "loss": 0.8151,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.5756,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.233766233766234e-06,
+      "loss": 0.8955,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.753246753246754e-06,
+      "loss": 0.8179,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.6165,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.792207792207793e-06,
+      "loss": 0.644,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.311688311688313e-06,
+      "loss": 0.6263,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.831168831168832e-06,
+      "loss": 0.604,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.350649350649352e-06,
+      "loss": 0.6234,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.87012987012987e-06,
+      "loss": 0.4765,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0389610389610389e-05,
+      "loss": 0.6287,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.5614,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.5288,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1948051948051949e-05,
+      "loss": 0.5402,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2467532467532468e-05,
+      "loss": 0.7049,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2987012987012988e-05,
+      "loss": 0.5522,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3506493506493508e-05,
+      "loss": 0.5248,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4025974025974028e-05,
+      "loss": 0.5145,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.5788,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5064935064935066e-05,
+      "loss": 0.5306,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5584415584415587e-05,
+      "loss": 0.5125,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6103896103896105e-05,
+      "loss": 0.4689,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6623376623376627e-05,
+      "loss": 0.547,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.5291,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7662337662337664e-05,
+      "loss": 0.5471,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.5058,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8701298701298704e-05,
+      "loss": 0.6317,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9220779220779222e-05,
+      "loss": 0.5528,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.974025974025974e-05,
+      "loss": 0.5237,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.025974025974026e-05,
+      "loss": 0.4855,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0779220779220778e-05,
+      "loss": 0.5078,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.12987012987013e-05,
+      "loss": 0.5442,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1818181818181818e-05,
+      "loss": 0.5459,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.233766233766234e-05,
+      "loss": 0.5121,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.5818,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3376623376623376e-05,
+      "loss": 0.5621,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3896103896103898e-05,
+      "loss": 0.5046,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4415584415584417e-05,
+      "loss": 0.6467,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4935064935064935e-05,
+      "loss": 0.6309,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5454545454545457e-05,
+      "loss": 0.5307,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5974025974025975e-05,
+      "loss": 0.5275,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6493506493506497e-05,
+      "loss": 0.5076,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7012987012987016e-05,
+      "loss": 0.4868,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7532467532467534e-05,
+      "loss": 0.498,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8051948051948056e-05,
+      "loss": 0.5097,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8571428571428574e-05,
+      "loss": 0.5728,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9090909090909093e-05,
+      "loss": 0.6051,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9610389610389614e-05,
+      "loss": 0.5289,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0129870129870133e-05,
+      "loss": 0.4784,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.064935064935065e-05,
+      "loss": 0.553,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.116883116883117e-05,
+      "loss": 0.6398,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1688311688311695e-05,
+      "loss": 0.5603,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.220779220779221e-05,
+      "loss": 0.574,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.272727272727273e-05,
+      "loss": 0.5635,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3246753246753254e-05,
+      "loss": 0.6664,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.376623376623377e-05,
+      "loss": 0.5512,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4285714285714284e-05,
+      "loss": 0.5781,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4805194805194805e-05,
+      "loss": 0.6433,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.532467532467533e-05,
+      "loss": 0.6159,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.584415584415584e-05,
+      "loss": 0.5117,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.5382,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6883116883116886e-05,
+      "loss": 0.6954,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.740259740259741e-05,
+      "loss": 0.5334,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.792207792207792e-05,
+      "loss": 0.5543,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8441558441558445e-05,
+      "loss": 0.5431,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8961038961038966e-05,
+      "loss": 0.5465,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.948051948051948e-05,
+      "loss": 0.705,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 0.5507,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999998377018602e-05,
+      "loss": 0.5442,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999935080770425e-05,
+      "loss": 0.5857,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999985393183223e-05,
+      "loss": 0.5712,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999974032350314e-05,
+      "loss": 0.5801,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999959425596754e-05,
+      "loss": 0.5809,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9999415729462494e-05,
+      "loss": 0.4979,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999920474427775e-05,
+      "loss": 0.5822,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999896130075573e-05,
+      "loss": 0.6364,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999868539929155e-05,
+      "loss": 0.5549,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998377040332975e-05,
+      "loss": 0.7195,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9998036224380475e-05,
+      "loss": 0.5773,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.999766295198718e-05,
+      "loss": 0.5054,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999725722375892e-05,
+      "loss": 0.7297,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999681904035417e-05,
+      "loss": 0.6237,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9996348402484097e-05,
+      "loss": 0.5179,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999584531091254e-05,
+      "loss": 0.6476,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9995309766456015e-05,
+      "loss": 0.5733,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9994741769983683e-05,
+      "loss": 0.7115,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99941413224174e-05,
+      "loss": 0.6657,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9993508424731686e-05,
+      "loss": 0.6315,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999284307795372e-05,
+      "loss": 0.5486,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999214528316334e-05,
+      "loss": 0.6716,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999141504149307e-05,
+      "loss": 0.5824,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.999065235412806e-05,
+      "loss": 0.5687,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9989857222306145e-05,
+      "loss": 0.6212,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998902964731781e-05,
+      "loss": 0.6434,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99881696305062e-05,
+      "loss": 0.63,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9987277173267094e-05,
+      "loss": 0.5466,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9986352277048935e-05,
+      "loss": 0.5889,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9985394943352827e-05,
+      "loss": 0.5018,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9984405173732485e-05,
+      "loss": 0.6604,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9983382969794305e-05,
+      "loss": 0.6143,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998232833319729e-05,
+      "loss": 0.6032,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.99812412656531e-05,
+      "loss": 0.5684,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.998012176892604e-05,
+      "loss": 0.607,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9978969844833006e-05,
+      "loss": 0.6904,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.997778549524357e-05,
+      "loss": 0.5654,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99765687220799e-05,
+      "loss": 0.5444,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99753195273168e-05,
+      "loss": 0.4964,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997403791298169e-05,
+      "loss": 0.6358,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99727238811546e-05,
+      "loss": 0.586,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997137743396818e-05,
+      "loss": 0.6463,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99699985736077e-05,
+      "loss": 0.6591,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996858730231102e-05,
+      "loss": 0.6572,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9967143622368595e-05,
+      "loss": 0.5987,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99656675361235e-05,
+      "loss": 0.4924,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9964159045971406e-05,
+      "loss": 0.5582,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9962618154360545e-05,
+      "loss": 0.5476,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996104486379177e-05,
+      "loss": 0.6053,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9959439176818496e-05,
+      "loss": 0.5272,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995780109604672e-05,
+      "loss": 0.5695,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9956130624135025e-05,
+      "loss": 0.5387,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995442776379455e-05,
+      "loss": 0.5237,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9952692517789004e-05,
+      "loss": 0.5105,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995092488893466e-05,
+      "loss": 0.6233,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9949124880100354e-05,
+      "loss": 0.5598,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994729249420746e-05,
+      "loss": 0.5255,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994542773422991e-05,
+      "loss": 0.6103,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994353060319417e-05,
+      "loss": 0.5559,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9941601104179245e-05,
+      "loss": 0.5664,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993963924031669e-05,
+      "loss": 0.5473,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993764501479056e-05,
+      "loss": 0.5592,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993561843083746e-05,
+      "loss": 0.4979,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993355949174649e-05,
+      "loss": 0.5479,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.993146820085926e-05,
+      "loss": 0.5516,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9929344561569906e-05,
+      "loss": 0.5907,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992718857732506e-05,
+      "loss": 0.4603,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992500025162384e-05,
+      "loss": 0.5514,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.992277958801785e-05,
+      "loss": 0.534,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.99205265901112e-05,
+      "loss": 0.4596,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9918241261560446e-05,
+      "loss": 0.4973,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991592360607465e-05,
+      "loss": 0.5523,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991357362741532e-05,
+      "loss": 0.7569,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.991119132939642e-05,
+      "loss": 0.6503,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990877671588439e-05,
+      "loss": 0.6139,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9906329790798085e-05,
+      "loss": 0.4892,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990385055810883e-05,
+      "loss": 0.4562,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.990133902184037e-05,
+      "loss": 0.4952,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9898795186068884e-05,
+      "loss": 0.5797,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989621905492297e-05,
+      "loss": 0.5032,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9893610632583634e-05,
+      "loss": 0.5649,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.989096992328431e-05,
+      "loss": 0.4962,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9888296931310806e-05,
+      "loss": 0.5127,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988559166100134e-05,
+      "loss": 0.6262,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988285411674653e-05,
+      "loss": 0.6089,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.988008430298935e-05,
+      "loss": 0.5512,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987728222422515e-05,
+      "loss": 0.4607,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.987444788500166e-05,
+      "loss": 0.5287,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.987158128991897e-05,
+      "loss": 0.5012,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986868244362948e-05,
+      "loss": 0.6283,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.986575135083799e-05,
+      "loss": 0.5733,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.98627880163016e-05,
+      "loss": 0.5631,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985979244482974e-05,
+      "loss": 0.5192,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985676464128418e-05,
+      "loss": 0.4973,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.985370461057898e-05,
+      "loss": 0.5503,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9850612357680527e-05,
+      "loss": 0.5935,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9847487887607464e-05,
+      "loss": 0.4936,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.984433120543077e-05,
+      "loss": 0.5004,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9841142316273675e-05,
+      "loss": 0.598,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983792122531169e-05,
+      "loss": 0.5926,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9834667937772575e-05,
+      "loss": 0.469,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.983138245893636e-05,
+      "loss": 0.6319,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982806479413533e-05,
+      "loss": 0.5071,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9824714948753964e-05,
+      "loss": 0.5592,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.982133292822902e-05,
+      "loss": 0.6466,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981791873804945e-05,
+      "loss": 0.512,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9814472383756435e-05,
+      "loss": 0.5332,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.981099387094332e-05,
+      "loss": 0.5653,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9807483205255675e-05,
+      "loss": 0.4709,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9803940392391255e-05,
+      "loss": 0.469,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9800365438099965e-05,
+      "loss": 0.6292,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9796758348183894e-05,
+      "loss": 0.5164,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9793119128497286e-05,
+      "loss": 0.5267,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978944778494653e-05,
+      "loss": 0.7136,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.978574432349013e-05,
+      "loss": 0.5142,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9782008750138756e-05,
+      "loss": 0.6516,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977824107095517e-05,
+      "loss": 0.5824,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.977444129205423e-05,
+      "loss": 0.4845,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9770609419602924e-05,
+      "loss": 0.5729,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.976674545982031e-05,
+      "loss": 0.5599,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97628494189775e-05,
+      "loss": 0.5576,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975892130339772e-05,
+      "loss": 0.6129,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9754961119456224e-05,
+      "loss": 0.515,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.975096887358031e-05,
+      "loss": 0.6165,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974694457224932e-05,
+      "loss": 0.5927,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.974288822199462e-05,
+      "loss": 0.5912,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.97387998293996e-05,
+      "loss": 0.6112,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973467940109964e-05,
+      "loss": 0.566,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.973052694378211e-05,
+      "loss": 0.626,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.972634246418638e-05,
+      "loss": 0.5197,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9722125969103784e-05,
+      "loss": 0.5387,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9717877465377606e-05,
+      "loss": 0.5472,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.971359695990309e-05,
+      "loss": 0.4668,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970928445962743e-05,
+      "loss": 0.667,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970493997154972e-05,
+      "loss": 0.4388,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.970056350272099e-05,
+      "loss": 0.5436,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9696155060244166e-05,
+      "loss": 0.615,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.969171465127406e-05,
+      "loss": 0.5022,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9687242283017394e-05,
+      "loss": 0.4725,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9682737962732715e-05,
+      "loss": 0.5885,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967820169773046e-05,
+      "loss": 0.5274,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.967363349537291e-05,
+      "loss": 0.6182,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.966903336307416e-05,
+      "loss": 0.498,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9664401308300154e-05,
+      "loss": 0.5827,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9659737338568614e-05,
+      "loss": 0.5368,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965504146144909e-05,
+      "loss": 0.5695,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.965031368456289e-05,
+      "loss": 0.5837,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.964555401558312e-05,
+      "loss": 0.5837,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9640762462234625e-05,
+      "loss": 0.5104,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.963593903229402e-05,
+      "loss": 0.5839,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9631083733589636e-05,
+      "loss": 0.5267,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962619657400152e-05,
+      "loss": 0.532,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.962127756146146e-05,
+      "loss": 0.6273,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961632670395291e-05,
+      "loss": 0.6771,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.961134400951101e-05,
+      "loss": 0.5064,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.960632948622261e-05,
+      "loss": 0.4727,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9601283142226166e-05,
+      "loss": 0.5815,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959620498571181e-05,
+      "loss": 0.5043,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.959109502492128e-05,
+      "loss": 0.5384,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958595326814796e-05,
+      "loss": 0.6107,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.958077972373683e-05,
+      "loss": 0.5685,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957557440008445e-05,
+      "loss": 0.6008,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.957033730563896e-05,
+      "loss": 0.5422,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.956506844890007e-05,
+      "loss": 0.5228,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9559767838419046e-05,
+      "loss": 0.5225,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.955443548279866e-05,
+      "loss": 0.4088,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.954907139069325e-05,
+      "loss": 0.4711,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9543675570808616e-05,
+      "loss": 0.4121,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.953824803190208e-05,
+      "loss": 0.4987,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9532788782782445e-05,
+      "loss": 0.5003,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9527297832309965e-05,
+      "loss": 0.4264,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.952177518939635e-05,
+      "loss": 0.5848,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9516220863004744e-05,
+      "loss": 0.5817,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9510634862149714e-05,
+      "loss": 0.4763,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.950501719589725e-05,
+      "loss": 0.5078,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94993678733647e-05,
+      "loss": 0.4858,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.949368690372083e-05,
+      "loss": 0.4842,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.948797429618573e-05,
+      "loss": 0.5472,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9482230060030866e-05,
+      "loss": 0.6265,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.947645420457902e-05,
+      "loss": 0.4302,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.94706467392043e-05,
+      "loss": 0.4702,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.946480767333212e-05,
+      "loss": 0.5875,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945893701643917e-05,
+      "loss": 0.4451,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945303477805341e-05,
+      "loss": 0.4793,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944710096775408e-05,
+      "loss": 0.5687,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.944113559517163e-05,
+      "loss": 0.5151,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.943513866998776e-05,
+      "loss": 0.4787,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942911020193536e-05,
+      "loss": 0.505,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.942305020079852e-05,
+      "loss": 0.5215,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.941695867641251e-05,
+      "loss": 0.5851,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.941083563866377e-05,
+      "loss": 0.6024,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.940468109748986e-05,
+      "loss": 0.424,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9398495062879496e-05,
+      "loss": 0.5863,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.93922775448725e-05,
+      "loss": 0.5696,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.938602855355979e-05,
+      "loss": 0.5615,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937974809908335e-05,
+      "loss": 0.459,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.937343619163624e-05,
+      "loss": 0.5283,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9367092841462583e-05,
+      "loss": 0.5184,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9360718058857514e-05,
+      "loss": 0.581,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.935431185416718e-05,
+      "loss": 0.5015,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934787423778874e-05,
+      "loss": 0.523,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.934140522017032e-05,
+      "loss": 0.4759,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.933490481181101e-05,
+      "loss": 0.5608,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932837302326087e-05,
+      "loss": 0.4703,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.932180986512085e-05,
+      "loss": 0.6176,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.931521534804285e-05,
+      "loss": 0.5746,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930858948272964e-05,
+      "loss": 0.5095,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.930193227993487e-05,
+      "loss": 0.5712,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.929524375046308e-05,
+      "loss": 0.4942,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.928852390516961e-05,
+      "loss": 0.5358,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9281772754960646e-05,
+      "loss": 0.4513,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.927499031079317e-05,
+      "loss": 0.5557,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926817658367499e-05,
+      "loss": 0.4836,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.926133158466463e-05,
+      "loss": 0.5452,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.925445532487141e-05,
+      "loss": 0.4743,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9247547815455364e-05,
+      "loss": 0.4349,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.924060906762726e-05,
+      "loss": 0.6087,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.923363909264856e-05,
+      "loss": 0.5657,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9226637901831394e-05,
+      "loss": 0.5267,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9219605506538564e-05,
+      "loss": 0.5845,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.921254191818353e-05,
+      "loss": 0.4986,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.920544714823034e-05,
+      "loss": 0.4618,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9198321208193696e-05,
+      "loss": 0.5387,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.919116410963886e-05,
+      "loss": 0.5784,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.918397586418167e-05,
+      "loss": 0.6678,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.917675648348851e-05,
+      "loss": 0.616,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9169505979276304e-05,
+      "loss": 0.4794,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9162224363312484e-05,
+      "loss": 0.5049,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915491164741498e-05,
+      "loss": 0.5892,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914756784345219e-05,
+      "loss": 0.5677,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.914019296334298e-05,
+      "loss": 0.6145,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9132787019056625e-05,
+      "loss": 0.4475,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.912535002261286e-05,
+      "loss": 0.6545,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911788198608177e-05,
+      "loss": 0.5416,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911038292158384e-05,
+      "loss": 0.5993,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910285284128994e-05,
+      "loss": 0.5418,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.909529175742122e-05,
+      "loss": 0.5655,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908769968224918e-05,
+      "loss": 0.5841,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.908007662809564e-05,
+      "loss": 0.4949,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.907242260733265e-05,
+      "loss": 0.5678,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9064737632382554e-05,
+      "loss": 0.4739,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.905702171571793e-05,
+      "loss": 0.4468,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.904927486986156e-05,
+      "loss": 0.5498,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9041497107386424e-05,
+      "loss": 0.4612,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9033688440915695e-05,
+      "loss": 0.3967,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9025848883122696e-05,
+      "loss": 0.5758,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901797844673087e-05,
+      "loss": 0.5568,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.901007714451381e-05,
+      "loss": 0.5079,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9002144989295164e-05,
+      "loss": 0.5619,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.899418199394868e-05,
+      "loss": 0.419,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.898618817139815e-05,
+      "loss": 0.5037,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.897816353461739e-05,
+      "loss": 0.5302,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8970108096630254e-05,
+      "loss": 0.5113,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.896202187051057e-05,
+      "loss": 0.5775,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.89539048693821e-05,
+      "loss": 0.4279,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.894575710641862e-05,
+      "loss": 0.5142,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8937578594843786e-05,
+      "loss": 0.5765,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8929369347931166e-05,
+      "loss": 0.4883,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.892112937900422e-05,
+      "loss": 0.6234,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.891285870143627e-05,
+      "loss": 0.4076,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.890455732865045e-05,
+      "loss": 0.5835,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.889622527411976e-05,
+      "loss": 0.4768,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8887862551366956e-05,
+      "loss": 0.5948,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.8879469173964583e-05,
+      "loss": 0.475,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.887104515553493e-05,
+      "loss": 0.5291,
+      "step": 342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.886259050975004e-05,
+      "loss": 0.475,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8854105250331634e-05,
+      "loss": 0.5078,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.884558939105113e-05,
+      "loss": 0.4905,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.88370429457296e-05,
+      "loss": 0.5698,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8828465928237786e-05,
+      "loss": 0.4713,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8819858352496016e-05,
+      "loss": 0.5396,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.881122023247422e-05,
+      "loss": 0.5581,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.880255158219192e-05,
+      "loss": 0.4544,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.5557,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8785122747171547e-05,
+      "loss": 0.4763,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8776362590720154e-05,
+      "loss": 0.5243,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.876757196058155e-05,
+      "loss": 0.5749,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8758750871022766e-05,
+      "loss": 0.5123,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874989933636028e-05,
+      "loss": 0.5105,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.874101737095994e-05,
+      "loss": 0.5152,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.873210498923704e-05,
+      "loss": 0.5083,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.872316220565619e-05,
+      "loss": 0.4874,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.871418903473138e-05,
+      "loss": 0.4745,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8705185491025876e-05,
+      "loss": 0.476,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.869615158915227e-05,
+      "loss": 0.5503,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.868708734377244e-05,
+      "loss": 0.637,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8677992769597456e-05,
+      "loss": 0.6173,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.866886788138766e-05,
+      "loss": 0.4823,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865971269395257e-05,
+      "loss": 0.4883,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.865052722215088e-05,
+      "loss": 0.5758,
+      "step": 367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8641311480890456e-05,
+      "loss": 0.5365,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.863206548512826e-05,
+      "loss": 0.5036,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.862278924987037e-05,
+      "loss": 0.4875,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8613482790171954e-05,
+      "loss": 0.52,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8604146121137215e-05,
+      "loss": 0.4766,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8594779257919406e-05,
+      "loss": 0.5596,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858538221572075e-05,
+      "loss": 0.5437,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.857595500979248e-05,
+      "loss": 0.5154,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8566497655434795e-05,
+      "loss": 0.5373,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8557010167996785e-05,
+      "loss": 0.5488,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854749256287646e-05,
+      "loss": 0.4678,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8537944855520735e-05,
+      "loss": 0.5343,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.852836706142535e-05,
+      "loss": 0.4335,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.851875919613489e-05,
+      "loss": 0.4824,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8509121275242744e-05,
+      "loss": 0.5803,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.849945331439107e-05,
+      "loss": 0.5944,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.848975532927079e-05,
+      "loss": 0.4715,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8480027335621564e-05,
+      "loss": 0.5204,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.847026934923174e-05,
+      "loss": 0.5004,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.846048138593834e-05,
+      "loss": 0.5234,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.845066346162706e-05,
+      "loss": 0.4365,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8440815592232197e-05,
+      "loss": 0.5007,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8430937793736665e-05,
+      "loss": 0.4528,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8421030082171946e-05,
+      "loss": 0.4274,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.841109247361807e-05,
+      "loss": 0.5858,
+      "step": 392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84011249842036e-05,
+      "loss": 0.5586,
+      "step": 393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8391127630105566e-05,
+      "loss": 0.5323,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8381100427549513e-05,
+      "loss": 0.5186,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.837104339280939e-05,
+      "loss": 0.5664,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8360956542207576e-05,
+      "loss": 0.6716,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.835083989211484e-05,
+      "loss": 0.4722,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8340693458950326e-05,
+      "loss": 0.54,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.83305172591815e-05,
+      "loss": 0.6097,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.832031130932415e-05,
+      "loss": 0.4382,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8310075625942334e-05,
+      "loss": 0.5365,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8299810225648394e-05,
+      "loss": 0.4416,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.828951512510286e-05,
+      "loss": 0.5995,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.82791903410145e-05,
+      "loss": 0.4308,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.826883589014025e-05,
+      "loss": 0.4772,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.825845178928519e-05,
+      "loss": 0.3986,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.824803805530251e-05,
+      "loss": 0.5877,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.823759470509353e-05,
+      "loss": 0.4991,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8227121755607595e-05,
+      "loss": 0.5229,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8216619223842105e-05,
+      "loss": 0.5466,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8206087126842484e-05,
+      "loss": 0.4902,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.819552548170213e-05,
+      "loss": 0.4547,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8184934305562386e-05,
+      "loss": 0.4283,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.817431361561255e-05,
+      "loss": 0.4245,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.816366342908978e-05,
+      "loss": 0.4963,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.815298376327916e-05,
+      "loss": 0.578,
+      "step": 417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8142274635513565e-05,
+      "loss": 0.4404,
+      "step": 418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.813153606317373e-05,
+      "loss": 0.536,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.812076806368814e-05,
+      "loss": 0.5148,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8109970654533066e-05,
+      "loss": 0.5244,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8099143853232504e-05,
+      "loss": 0.5388,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.808828767735814e-05,
+      "loss": 0.5295,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.807740214452937e-05,
+      "loss": 0.4123,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.806648727241318e-05,
+      "loss": 0.5009,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8055543078724224e-05,
+      "loss": 0.4591,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.8044569581224725e-05,
+      "loss": 0.4696,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.803356679772446e-05,
+      "loss": 0.6017,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.802253474608074e-05,
+      "loss": 0.5985,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.801147344419839e-05,
+      "loss": 0.4544,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.800038291002968e-05,
+      "loss": 0.4763,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.798926316157435e-05,
+      "loss": 0.4556,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7978114216879554e-05,
+      "loss": 0.5222,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.796693609403981e-05,
+      "loss": 0.5888,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7955728811197006e-05,
+      "loss": 0.3966,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.7944492386540355e-05,
+      "loss": 0.5383,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.793322683830636e-05,
+      "loss": 0.5369,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.792193218477881e-05,
+      "loss": 0.5572,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.79106084442887e-05,
+      "loss": 0.5011,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.789925563521426e-05,
+      "loss": 0.4965,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.788787377598089e-05,
+      "loss": 0.4592,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.787646288506113e-05,
+      "loss": 0.43,
+      "step": 442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.786502298097464e-05,
+      "loss": 0.5452,
+      "step": 443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.785355408228818e-05,
+      "loss": 0.544,
+      "step": 444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.784205620761556e-05,
+      "loss": 0.5826,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.783052937561762e-05,
+      "loss": 0.4862,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.781897360500217e-05,
+      "loss": 0.5582,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.780738891452404e-05,
+      "loss": 0.4761,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.779577532298495e-05,
+      "loss": 0.5395,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.778413284923355e-05,
+      "loss": 0.4824,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.777246151216536e-05,
+      "loss": 0.4581,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.776076133072273e-05,
+      "loss": 0.4843,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.774903232389486e-05,
+      "loss": 0.4269,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7737274510717686e-05,
+      "loss": 0.5191,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.772548791027393e-05,
+      "loss": 0.5512,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.771367254169303e-05,
+      "loss": 0.5506,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.77018284241511e-05,
+      "loss": 0.4839,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.768995557687093e-05,
+      "loss": 0.5629,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7678054019121926e-05,
+      "loss": 0.458,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7666123770220094e-05,
+      "loss": 0.4668,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.765416484952802e-05,
+      "loss": 0.5874,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.764217727645479e-05,
+      "loss": 0.6277,
+      "step": 462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.763016107045602e-05,
+      "loss": 0.4448,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.761811625103379e-05,
+      "loss": 0.4731,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.760604283773662e-05,
+      "loss": 0.5338,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.759394085015942e-05,
+      "loss": 0.683,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.758181030794352e-05,
+      "loss": 0.5332,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7569651230776545e-05,
+      "loss": 0.4513,
+      "step": 468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.755746363839245e-05,
+      "loss": 0.4642,
+      "step": 469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.7545247550571466e-05,
+      "loss": 0.5065,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.753300298714009e-05,
+      "loss": 0.6196,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.752072996797102e-05,
+      "loss": 0.4861,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.750842851298313e-05,
+      "loss": 0.5863,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.749609864214145e-05,
+      "loss": 0.5436,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.748374037545714e-05,
+      "loss": 0.4978,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7471353732987435e-05,
+      "loss": 0.4821,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.745893873483562e-05,
+      "loss": 0.3653,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.744649540115101e-05,
+      "loss": 0.5194,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7434023752128907e-05,
+      "loss": 0.4082,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.742152380801057e-05,
+      "loss": 0.5249,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.740899558908315e-05,
+      "loss": 0.5432,
+      "step": 481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.739643911567975e-05,
+      "loss": 0.5635,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.738385440817926e-05,
+      "loss": 0.5366,
+      "step": 483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.737124148700645e-05,
+      "loss": 0.5511,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7358600372631835e-05,
+      "loss": 0.4457,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.734593108557173e-05,
+      "loss": 0.4634,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.733323364638814e-05,
+      "loss": 0.4635,
+      "step": 487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.732050807568878e-05,
+      "loss": 0.537,
+      "step": 488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7307754394126995e-05,
+      "loss": 0.6059,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7294972622401807e-05,
+      "loss": 0.5893,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.728216278125776e-05,
+      "loss": 0.5085,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7269324891485004e-05,
+      "loss": 0.4242,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725645897391919e-05,
+      "loss": 0.5455,
+      "step": 493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.724356504944148e-05,
+      "loss": 0.4501,
+      "step": 494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.723064313897845e-05,
+      "loss": 0.4816,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.721769326350213e-05,
+      "loss": 0.5642,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.720471544402993e-05,
+      "loss": 0.5614,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.719170970162461e-05,
+      "loss": 0.5259,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.717867605739423e-05,
+      "loss": 0.3752,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.716561453249218e-05,
+      "loss": 0.4771,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.715252514811706e-05,
+      "loss": 0.49,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7139407925512696e-05,
+      "loss": 0.4702,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.712626288596809e-05,
+      "loss": 0.5321,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7113090050817404e-05,
+      "loss": 0.5525,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7099889441439904e-05,
+      "loss": 0.5278,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7086661079259934e-05,
+      "loss": 0.5719,
+      "step": 506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.707340498574688e-05,
+      "loss": 0.4702,
+      "step": 507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.706012118241513e-05,
+      "loss": 0.562,
+      "step": 508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.7046809690824056e-05,
+      "loss": 0.5092,
+      "step": 509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.703347053257796e-05,
+      "loss": 0.5284,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.702010372932605e-05,
+      "loss": 0.4989,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.700670930276238e-05,
+      "loss": 0.5538,
+      "step": 512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6993287274625884e-05,
+      "loss": 0.4981,
+      "step": 513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.697983766670025e-05,
+      "loss": 0.4475,
+      "step": 514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6966360500813945e-05,
+      "loss": 0.4723,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.695285579884015e-05,
+      "loss": 0.4978,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.693932358269675e-05,
+      "loss": 0.554,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.692576387434629e-05,
+      "loss": 0.569,
+      "step": 518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.69121766957959e-05,
+      "loss": 0.5445,
+      "step": 519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6898562069097345e-05,
+      "loss": 0.4051,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.6884920016346893e-05,
+      "loss": 0.4815,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.687125055968535e-05,
+      "loss": 0.5236,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.685755372129799e-05,
+      "loss": 0.5094,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6843829523414524e-05,
+      "loss": 0.5529,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6830077988309066e-05,
+      "loss": 0.4887,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.681629913830011e-05,
+      "loss": 0.6015,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6802492995750476e-05,
+      "loss": 0.4497,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6788659583067256e-05,
+      "loss": 0.5144,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.677479892270185e-05,
+      "loss": 0.4443,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6760911037149834e-05,
+      "loss": 0.4414,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6746995948950986e-05,
+      "loss": 0.4908,
+      "step": 531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.673305368068924e-05,
+      "loss": 0.5238,
+      "step": 532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.671908425499264e-05,
+      "loss": 0.5608,
+      "step": 533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.670508769453331e-05,
+      "loss": 0.5362,
+      "step": 534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.669106402202739e-05,
+      "loss": 0.5022,
+      "step": 535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6677013260235054e-05,
+      "loss": 0.4535,
+      "step": 536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.666293543196043e-05,
+      "loss": 0.5125,
+      "step": 537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6648830560051553e-05,
+      "loss": 0.4886,
+      "step": 538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.6634698667400384e-05,
+      "loss": 0.5597,
+      "step": 539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.662053977694272e-05,
+      "loss": 0.5552,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.660635391165817e-05,
+      "loss": 0.4465,
+      "step": 541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.659214109457013e-05,
+      "loss": 0.5953,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.657790134874574e-05,
+      "loss": 0.5021,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.656363469729585e-05,
+      "loss": 0.4471,
+      "step": 544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.654934116337496e-05,
+      "loss": 0.4542,
+      "step": 545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.653502077018122e-05,
+      "loss": 0.4486,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6520673540956335e-05,
+      "loss": 0.4503,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.650629949898563e-05,
+      "loss": 0.4966,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6491898667597884e-05,
+      "loss": 0.5178,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.647747107016538e-05,
+      "loss": 0.5378,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.646301673010385e-05,
+      "loss": 0.3736,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.64485356708724e-05,
+      "loss": 0.5457,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.643402791597354e-05,
+      "loss": 0.4876,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6419493488953076e-05,
+      "loss": 0.4898,
+      "step": 554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.640493241340012e-05,
+      "loss": 0.498,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.639034471294702e-05,
+      "loss": 0.4739,
+      "step": 556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.637573041126934e-05,
+      "loss": 0.4243,
+      "step": 557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6361089532085826e-05,
+      "loss": 0.4907,
+      "step": 558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6346422099158354e-05,
+      "loss": 0.5058,
+      "step": 559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6331728136291884e-05,
+      "loss": 0.3693,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.631700766733446e-05,
+      "loss": 0.452,
+      "step": 561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6302260716177115e-05,
+      "loss": 0.4754,
+      "step": 562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.628748730675389e-05,
+      "loss": 0.6154,
+      "step": 563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6272687463041735e-05,
+      "loss": 0.4228,
+      "step": 564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.625786120906054e-05,
+      "loss": 0.5139,
+      "step": 565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.624300856887303e-05,
+      "loss": 0.4644,
+      "step": 566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622812956658477e-05,
+      "loss": 0.45,
+      "step": 567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6213224226344097e-05,
+      "loss": 0.4001,
+      "step": 568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.61982925723421e-05,
+      "loss": 0.5407,
+      "step": 569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.618333462881258e-05,
+      "loss": 0.5144,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6168350420032004e-05,
+      "loss": 0.5646,
+      "step": 571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.615333997031945e-05,
+      "loss": 0.4699,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.613830330403662e-05,
+      "loss": 0.4586,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.612324044558773e-05,
+      "loss": 0.5682,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.610815141941952e-05,
+      "loss": 0.4821,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6093036250021195e-05,
+      "loss": 0.5243,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.607789496192441e-05,
+      "loss": 0.5943,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.606272757970318e-05,
+      "loss": 0.391,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.604753412797388e-05,
+      "loss": 0.5144,
+      "step": 579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.603231463139521e-05,
+      "loss": 0.4946,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.601706911466813e-05,
+      "loss": 0.448,
+      "step": 581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.6001797602535826e-05,
+      "loss": 0.5044,
+      "step": 582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.598650011978368e-05,
+      "loss": 0.3948,
+      "step": 583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.597117669123922e-05,
+      "loss": 0.5032,
+      "step": 584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.595582734177208e-05,
+      "loss": 0.4679,
+      "step": 585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.594045209629398e-05,
+      "loss": 0.5023,
+      "step": 586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.592505097975865e-05,
+      "loss": 0.4072,
+      "step": 587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.590962401716182e-05,
+      "loss": 0.3938,
+      "step": 588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.589417123354116e-05,
+      "loss": 0.5037,
+      "step": 589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.587869265397625e-05,
+      "loss": 0.5356,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.586318830358854e-05,
+      "loss": 0.4492,
+      "step": 591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5847658207541304e-05,
+      "loss": 0.5302,
+      "step": 592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5832102391039595e-05,
+      "loss": 0.4834,
+      "step": 593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.581652087933021e-05,
+      "loss": 0.451,
+      "step": 594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.580091369770166e-05,
+      "loss": 0.5303,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.578528087148411e-05,
+      "loss": 0.5433,
+      "step": 596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.576962242604934e-05,
+      "loss": 0.5367,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5753938386810715e-05,
+      "loss": 0.4982,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.573822877922315e-05,
+      "loss": 0.497,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.572249362878303e-05,
+      "loss": 0.4415,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.570673296102823e-05,
+      "loss": 0.589,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.569094680153801e-05,
+      "loss": 0.5518,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5675135175933015e-05,
+      "loss": 0.6131,
+      "step": 603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.565929810987522e-05,
+      "loss": 0.5942,
+      "step": 604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5643435629067884e-05,
+      "loss": 0.5429,
+      "step": 605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.562754775925552e-05,
+      "loss": 0.54,
+      "step": 606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.561163452622384e-05,
+      "loss": 0.4638,
+      "step": 607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5595695955799745e-05,
+      "loss": 0.4643,
+      "step": 608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.557973207385122e-05,
+      "loss": 0.5324,
+      "step": 609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5563742906287355e-05,
+      "loss": 0.5073,
+      "step": 610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.554772847905826e-05,
+      "loss": 0.5931,
+      "step": 611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.553168881815508e-05,
+      "loss": 0.6033,
+      "step": 612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5515623949609856e-05,
+      "loss": 0.5567,
+      "step": 613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5499533899495584e-05,
+      "loss": 0.4609,
+      "step": 614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.548341869392612e-05,
+      "loss": 0.5222,
+      "step": 615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.546727835905614e-05,
+      "loss": 0.4794,
+      "step": 616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5451112921081106e-05,
+      "loss": 0.4886,
+      "step": 617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.543492240623722e-05,
+      "loss": 0.4034,
+      "step": 618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.541870684080139e-05,
+      "loss": 0.5108,
+      "step": 619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.540246625109117e-05,
+      "loss": 0.4848,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.538620066346475e-05,
+      "loss": 0.4527,
+      "step": 621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5369910104320865e-05,
+      "loss": 0.4262,
+      "step": 622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.535359460009879e-05,
+      "loss": 0.5448,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.533725417727829e-05,
+      "loss": 0.5545,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.6005,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5304498681963216e-05,
+      "loss": 0.4632,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5288083662630194e-05,
+      "loss": 0.463,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.527164383102178e-05,
+      "loss": 0.567,
+      "step": 628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.525517921381951e-05,
+      "loss": 0.3915,
+      "step": 629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.523868983774516e-05,
+      "loss": 0.524,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.522217572956067e-05,
+      "loss": 0.5808,
+      "step": 631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.520563691606813e-05,
+      "loss": 0.4882,
+      "step": 632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.518907342410974e-05,
+      "loss": 0.4451,
+      "step": 633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5172485280567725e-05,
+      "loss": 0.5678,
+      "step": 634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.515587251236434e-05,
+      "loss": 0.4416,
+      "step": 635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.513923514646179e-05,
+      "loss": 0.4353,
+      "step": 636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5122573209862226e-05,
+      "loss": 0.4951,
+      "step": 637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.510588672960765e-05,
+      "loss": 0.5311,
+      "step": 638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508917573277991e-05,
+      "loss": 0.5395,
+      "step": 639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5072440246500644e-05,
+      "loss": 0.4744,
+      "step": 640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505568029793124e-05,
+      "loss": 0.4652,
+      "step": 641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.5038895914272775e-05,
+      "loss": 0.514,
+      "step": 642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.502208712276599e-05,
+      "loss": 0.4281,
+      "step": 643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.500525395069126e-05,
+      "loss": 0.4386,
+      "step": 644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.498839642536848e-05,
+      "loss": 0.4394,
+      "step": 645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.497151457415713e-05,
+      "loss": 0.5385,
+      "step": 646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.495460842445612e-05,
+      "loss": 0.4855,
+      "step": 647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.493767800370382e-05,
+      "loss": 0.5326,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4920723339377996e-05,
+      "loss": 0.5587,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4903744458995746e-05,
+      "loss": 0.4308,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4886741390113483e-05,
+      "loss": 0.3771,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.486971416032686e-05,
+      "loss": 0.451,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.485266279727076e-05,
+      "loss": 0.5269,
+      "step": 653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.483558732861924e-05,
+      "loss": 0.3923,
+      "step": 654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4818487782085446e-05,
+      "loss": 0.4396,
+      "step": 655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.480136418542165e-05,
+      "loss": 0.5025,
+      "step": 656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.47842165664191e-05,
+      "loss": 0.435,
+      "step": 657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.476704495290808e-05,
+      "loss": 0.5078,
+      "step": 658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.474984937275781e-05,
+      "loss": 0.5105,
+      "step": 659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4732629853876383e-05,
+      "loss": 0.4975,
+      "step": 660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4715386424210754e-05,
+      "loss": 0.5334,
+      "step": 661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4698119111746696e-05,
+      "loss": 0.5511,
+      "step": 662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4680827944508735e-05,
+      "loss": 0.5033,
+      "step": 663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.466351295056011e-05,
+      "loss": 0.5337,
+      "step": 664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4646174158002744e-05,
+      "loss": 0.4916,
+      "step": 665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.462881159497716e-05,
+      "loss": 0.5126,
+      "step": 666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4611425289662487e-05,
+      "loss": 0.5015,
+      "step": 667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4594015270276364e-05,
+      "loss": 0.5908,
+      "step": 668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.457658156507495e-05,
+      "loss": 0.4719,
+      "step": 669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.45591242023528e-05,
+      "loss": 0.495,
+      "step": 670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4541643210442895e-05,
+      "loss": 0.4086,
+      "step": 671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.452413861771657e-05,
+      "loss": 0.4833,
+      "step": 672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.450661045258345e-05,
+      "loss": 0.4607,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.448905874349141e-05,
+      "loss": 0.4604,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.447148351892656e-05,
+      "loss": 0.5165,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.445388480741315e-05,
+      "loss": 0.569,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.443626263751358e-05,
+      "loss": 0.5806,
+      "step": 677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4418617037828284e-05,
+      "loss": 0.4693,
+      "step": 678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.440094803699576e-05,
+      "loss": 0.5037,
+      "step": 679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.438325566369245e-05,
+      "loss": 0.5094,
+      "step": 680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4365539946632755e-05,
+      "loss": 0.4847,
+      "step": 681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4347800914568954e-05,
+      "loss": 0.4603,
+      "step": 682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433003859629117e-05,
+      "loss": 0.5028,
+      "step": 683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431225302062732e-05,
+      "loss": 0.4397,
+      "step": 684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.429444421644304e-05,
+      "loss": 0.4376,
+      "step": 685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4276612212641715e-05,
+      "loss": 0.4982,
+      "step": 686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.425875703816434e-05,
+      "loss": 0.5015,
+      "step": 687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4240878721989525e-05,
+      "loss": 0.5613,
+      "step": 688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4222977293133456e-05,
+      "loss": 0.4167,
+      "step": 689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.420505278064982e-05,
+      "loss": 0.513,
+      "step": 690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.418710521362976e-05,
+      "loss": 0.5448,
+      "step": 691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.416913462120185e-05,
+      "loss": 0.616,
+      "step": 692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4151141032532026e-05,
+      "loss": 0.5791,
+      "step": 693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.413312447682354e-05,
+      "loss": 0.5131,
+      "step": 694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4115084983316934e-05,
+      "loss": 0.5542,
+      "step": 695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4097022581289976e-05,
+      "loss": 0.408,
+      "step": 696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.40789373000576e-05,
+      "loss": 0.5337,
+      "step": 697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4060829168971875e-05,
+      "loss": 0.4721,
+      "step": 698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.404269821742197e-05,
+      "loss": 0.4695,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4024544474834086e-05,
+      "loss": 0.4884,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.40063679706714e-05,
+      "loss": 0.4006,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.398816873443405e-05,
+      "loss": 0.441,
+      "step": 702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3969946795659044e-05,
+      "loss": 0.4471,
+      "step": 703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.395170218392026e-05,
+      "loss": 0.4182,
+      "step": 704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.393343492882837e-05,
+      "loss": 0.5503,
+      "step": 705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.391514506003077e-05,
+      "loss": 0.4295,
+      "step": 706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.389683260721159e-05,
+      "loss": 0.4451,
+      "step": 707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.38784976000916e-05,
+      "loss": 0.4594,
+      "step": 708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3860140068428174e-05,
+      "loss": 0.4335,
+      "step": 709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.384176004201524e-05,
+      "loss": 0.5331,
+      "step": 710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3823357550683246e-05,
+      "loss": 0.4503,
+      "step": 711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.380493262429909e-05,
+      "loss": 0.4396,
+      "step": 712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.378648529276608e-05,
+      "loss": 0.5117,
+      "step": 713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.376801558602389e-05,
+      "loss": 0.4402,
+      "step": 714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.374952353404853e-05,
+      "loss": 0.5874,
+      "step": 715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3731009166852234e-05,
+      "loss": 0.5402,
+      "step": 716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.371247251448349e-05,
+      "loss": 0.4879,
+      "step": 717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.369391360702693e-05,
+      "loss": 0.4516,
+      "step": 718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.367533247460332e-05,
+      "loss": 0.4059,
+      "step": 719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.365672914736951e-05,
+      "loss": 0.5416,
+      "step": 720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.363810365551831e-05,
+      "loss": 0.4434,
+      "step": 721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.361945602927859e-05,
+      "loss": 0.4817,
+      "step": 722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3600786298915075e-05,
+      "loss": 0.5187,
+      "step": 723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.3582094494728406e-05,
+      "loss": 0.4099,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3563380647055024e-05,
+      "loss": 0.5413,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3544644786267154e-05,
+      "loss": 0.461,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.352588694277276e-05,
+      "loss": 0.4338,
+      "step": 727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3507107147015466e-05,
+      "loss": 0.5911,
+      "step": 728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3488305429474533e-05,
+      "loss": 0.4516,
+      "step": 729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.346948182066479e-05,
+      "loss": 0.4739,
+      "step": 730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3450636351136614e-05,
+      "loss": 0.5487,
+      "step": 731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.343176905147585e-05,
+      "loss": 0.478,
+      "step": 732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3412879952303777e-05,
+      "loss": 0.5168,
+      "step": 733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.339396908427704e-05,
+      "loss": 0.4873,
+      "step": 734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.337503647808764e-05,
+      "loss": 0.4694,
+      "step": 735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.335608216446284e-05,
+      "loss": 0.4542,
+      "step": 736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.333710617416514e-05,
+      "loss": 0.4479,
+      "step": 737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.331810853799221e-05,
+      "loss": 0.5534,
+      "step": 738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3299089286776866e-05,
+      "loss": 0.5766,
+      "step": 739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3280048451387e-05,
+      "loss": 0.4225,
+      "step": 740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3260986062725535e-05,
+      "loss": 0.5127,
+      "step": 741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.324190215173037e-05,
+      "loss": 0.5232,
+      "step": 742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3222796749374343e-05,
+      "loss": 0.5194,
+      "step": 743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3203669886665155e-05,
+      "loss": 0.4674,
+      "step": 744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3184521594645364e-05,
+      "loss": 0.3973,
+      "step": 745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3165351904392276e-05,
+      "loss": 0.4646,
+      "step": 746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.314616084701795e-05,
+      "loss": 0.4843,
+      "step": 747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.3126948453669126e-05,
+      "loss": 0.4921,
+      "step": 748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.310771475552713e-05,
+      "loss": 0.4535,
+      "step": 749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.308845978380793e-05,
+      "loss": 0.4404,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.306918356976197e-05,
+      "loss": 0.4886,
+      "step": 751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.304988614467419e-05,
+      "loss": 0.5074,
+      "step": 752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.303056753986395e-05,
+      "loss": 0.5004,
+      "step": 753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.301122778668498e-05,
+      "loss": 0.3993,
+      "step": 754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.299186691652536e-05,
+      "loss": 0.5298,
+      "step": 755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.29724849608074e-05,
+      "loss": 0.4418,
+      "step": 756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.295308195098767e-05,
+      "loss": 0.4942,
+      "step": 757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2933657918556886e-05,
+      "loss": 0.4626,
+      "step": 758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2914212895039885e-05,
+      "loss": 0.4811,
+      "step": 759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.289474691199559e-05,
+      "loss": 0.6168,
+      "step": 760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.287526000101693e-05,
+      "loss": 0.3782,
+      "step": 761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.285575219373079e-05,
+      "loss": 0.4867,
+      "step": 762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.283622352179798e-05,
+      "loss": 0.5996,
+      "step": 763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.281667401691318e-05,
+      "loss": 0.4389,
+      "step": 764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.279710371080486e-05,
+      "loss": 0.56,
+      "step": 765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2777512635235275e-05,
+      "loss": 0.3742,
+      "step": 766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2757900822000366e-05,
+      "loss": 0.4034,
+      "step": 767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.273826830292975e-05,
+      "loss": 0.4369,
+      "step": 768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2718615109886624e-05,
+      "loss": 0.4873,
+      "step": 769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.269894127476778e-05,
+      "loss": 0.581,
+      "step": 770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.267924682950347e-05,
+      "loss": 0.4639,
+      "step": 771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2659531806057416e-05,
+      "loss": 0.4396,
+      "step": 772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.263979623642673e-05,
+      "loss": 0.4516,
+      "step": 773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2620040152641875e-05,
+      "loss": 0.4972,
+      "step": 774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2600263586766617e-05,
+      "loss": 0.5436,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.258046657089795e-05,
+      "loss": 0.4099,
+      "step": 776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.256064913716605e-05,
+      "loss": 0.4376,
+      "step": 777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2540811317734265e-05,
+      "loss": 0.4125,
+      "step": 778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.252095314479899e-05,
+      "loss": 0.4374,
+      "step": 779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.250107465058968e-05,
+      "loss": 0.6356,
+      "step": 780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.248117586736875e-05,
+      "loss": 0.4665,
+      "step": 781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.246125682743157e-05,
+      "loss": 0.4746,
+      "step": 782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2441317563106355e-05,
+      "loss": 0.5271,
+      "step": 783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.242135810675418e-05,
+      "loss": 0.4444,
+      "step": 784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.240137849076885e-05,
+      "loss": 0.5212,
+      "step": 785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.238137874757692e-05,
+      "loss": 0.5208,
+      "step": 786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.236135890963761e-05,
+      "loss": 0.4329,
+      "step": 787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.234131900944273e-05,
+      "loss": 0.4654,
+      "step": 788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.232125907951667e-05,
+      "loss": 0.4542,
+      "step": 789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2301179152416326e-05,
+      "loss": 0.4556,
+      "step": 790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.228107926073104e-05,
+      "loss": 0.5383,
+      "step": 791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2260959437082576e-05,
+      "loss": 0.5776,
+      "step": 792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.224081971412502e-05,
+      "loss": 0.4685,
+      "step": 793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.222066012454477e-05,
+      "loss": 0.4562,
+      "step": 794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.220048070106047e-05,
+      "loss": 0.541,
+      "step": 795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.218028147642294e-05,
+      "loss": 0.4888,
+      "step": 796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.216006248341516e-05,
+      "loss": 0.4791,
+      "step": 797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2139823754852164e-05,
+      "loss": 0.488,
+      "step": 798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.211956532358104e-05,
+      "loss": 0.4528,
+      "step": 799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.2099287222480844e-05,
+      "loss": 0.3793,
+      "step": 800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.207898948446256e-05,
+      "loss": 0.4615,
+      "step": 801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.205867214246903e-05,
+      "loss": 0.5663,
+      "step": 802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.203833522947494e-05,
+      "loss": 0.6092,
+      "step": 803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.20179787784867e-05,
+      "loss": 0.4427,
+      "step": 804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1997602822542466e-05,
+      "loss": 0.492,
+      "step": 805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.197720739471202e-05,
+      "loss": 0.5084,
+      "step": 806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.195679252809679e-05,
+      "loss": 0.4587,
+      "step": 807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19363582558297e-05,
+      "loss": 0.4175,
+      "step": 808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.19159046110752e-05,
+      "loss": 0.4976,
+      "step": 809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1895431627029174e-05,
+      "loss": 0.441,
+      "step": 810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.18749393369189e-05,
+      "loss": 0.3815,
+      "step": 811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.185442777400297e-05,
+      "loss": 0.4086,
+      "step": 812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1833896971571296e-05,
+      "loss": 0.4575,
+      "step": 813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1813346962944956e-05,
+      "loss": 0.4681,
+      "step": 814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.179277778147626e-05,
+      "loss": 0.4523,
+      "step": 815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.177218946054859e-05,
+      "loss": 0.4826,
+      "step": 816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1751582033576407e-05,
+      "loss": 0.4189,
+      "step": 817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.173095553400519e-05,
+      "loss": 0.4147,
+      "step": 818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.171030999531136e-05,
+      "loss": 0.525,
+      "step": 819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1689645451002244e-05,
+      "loss": 0.5071,
+      "step": 820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.166896193461601e-05,
+      "loss": 0.4686,
+      "step": 821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.164825947972163e-05,
+      "loss": 0.4643,
+      "step": 822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.1627538119918777e-05,
+      "loss": 0.489,
+      "step": 823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.160679788883787e-05,
+      "loss": 0.5085,
+      "step": 824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.15860388201399e-05,
+      "loss": 0.4603,
+      "step": 825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.156526094751644e-05,
+      "loss": 0.4718,
+      "step": 826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1544464304689593e-05,
+      "loss": 0.4265,
+      "step": 827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.152364892541194e-05,
+      "loss": 0.631,
+      "step": 828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1502814843466435e-05,
+      "loss": 0.5095,
+      "step": 829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1481962092666425e-05,
+      "loss": 0.4041,
+      "step": 830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.146109070685552e-05,
+      "loss": 0.5587,
+      "step": 831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1440200719907604e-05,
+      "loss": 0.4549,
+      "step": 832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.141929216572672e-05,
+      "loss": 0.3486,
+      "step": 833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.139836507824708e-05,
+      "loss": 0.4382,
+      "step": 834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.137741949143295e-05,
+      "loss": 0.4381,
+      "step": 835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1356455439278626e-05,
+      "loss": 0.402,
+      "step": 836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1335472955808385e-05,
+      "loss": 0.4856,
+      "step": 837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.131447207507638e-05,
+      "loss": 0.5101,
+      "step": 838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.129345283116669e-05,
+      "loss": 0.5207,
+      "step": 839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.127241525819313e-05,
+      "loss": 0.5223,
+      "step": 840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.125135939029929e-05,
+      "loss": 0.5078,
+      "step": 841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.123028526165846e-05,
+      "loss": 0.5544,
+      "step": 842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.120919290647356e-05,
+      "loss": 0.4907,
+      "step": 843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.118808235897708e-05,
+      "loss": 0.389,
+      "step": 844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1166953653431065e-05,
+      "loss": 0.4399,
+      "step": 845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.114580682412698e-05,
+      "loss": 0.4826,
+      "step": 846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1124641905385774e-05,
+      "loss": 0.447,
+      "step": 847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.110345893155768e-05,
+      "loss": 0.4583,
+      "step": 848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.108225793702229e-05,
+      "loss": 0.4597,
+      "step": 849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.1061038956188427e-05,
+      "loss": 0.4386,
+      "step": 850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.103980202349409e-05,
+      "loss": 0.4616,
+      "step": 851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.101854717340644e-05,
+      "loss": 0.5059,
+      "step": 852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0997274440421684e-05,
+      "loss": 0.4555,
+      "step": 853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.097598385906509e-05,
+      "loss": 0.4722,
+      "step": 854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.095467546389086e-05,
+      "loss": 0.433,
+      "step": 855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0933349289482134e-05,
+      "loss": 0.4563,
+      "step": 856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0912005370450895e-05,
+      "loss": 0.4889,
+      "step": 857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.089064374143793e-05,
+      "loss": 0.4575,
+      "step": 858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.086926443711275e-05,
+      "loss": 0.4884,
+      "step": 859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0847867492173586e-05,
+      "loss": 0.5973,
+      "step": 860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0826452941347277e-05,
+      "loss": 0.5556,
+      "step": 861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0805020819389236e-05,
+      "loss": 0.6166,
+      "step": 862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0783571161083395e-05,
+      "loss": 0.444,
+      "step": 863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0762104001242156e-05,
+      "loss": 0.5083,
+      "step": 864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.074061937470633e-05,
+      "loss": 0.5194,
+      "step": 865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.071911731634504e-05,
+      "loss": 0.6035,
+      "step": 866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069759786105576e-05,
+      "loss": 0.5395,
+      "step": 867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.067606104376414e-05,
+      "loss": 0.3995,
+      "step": 868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.065450689942404e-05,
+      "loss": 0.4451,
+      "step": 869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.063293546301744e-05,
+      "loss": 0.447,
+      "step": 870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.061134676955438e-05,
+      "loss": 0.5705,
+      "step": 871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.05897408540729e-05,
+      "loss": 0.4608,
+      "step": 872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.056811775163901e-05,
+      "loss": 0.4069,
+      "step": 873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.054647749734659e-05,
+      "loss": 0.4658,
+      "step": 874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.052482012631738e-05,
+      "loss": 0.4375,
+      "step": 875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.050314567370089e-05,
+      "loss": 0.4558,
+      "step": 876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.048145417467436e-05,
+      "loss": 0.4633,
+      "step": 877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.045974566444267e-05,
+      "loss": 0.4456,
+      "step": 878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0438020178238338e-05,
+      "loss": 0.4189,
+      "step": 879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0416277751321432e-05,
+      "loss": 0.4388,
+      "step": 880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0394518418979495e-05,
+      "loss": 0.4391,
+      "step": 881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.037274221652752e-05,
+      "loss": 0.3522,
+      "step": 882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0350949179307885e-05,
+      "loss": 0.3772,
+      "step": 883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0329139342690283e-05,
+      "loss": 0.4918,
+      "step": 884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0307312742071667e-05,
+      "loss": 0.4326,
+      "step": 885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.028546941287621e-05,
+      "loss": 0.4288,
+      "step": 886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0263609390555224e-05,
+      "loss": 0.4539,
+      "step": 887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.024173271058712e-05,
+      "loss": 0.3628,
+      "step": 888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0219839408477348e-05,
+      "loss": 0.4238,
+      "step": 889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.019792951975832e-05,
+      "loss": 0.4207,
+      "step": 890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0176003079989386e-05,
+      "loss": 0.5083,
+      "step": 891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0154060124756746e-05,
+      "loss": 0.5136,
+      "step": 892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0132100689673415e-05,
+      "loss": 0.3989,
+      "step": 893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.011012481037914e-05,
+      "loss": 0.4432,
+      "step": 894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0088132522540366e-05,
+      "loss": 0.4801,
+      "step": 895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0066123861850162e-05,
+      "loss": 0.5141,
+      "step": 896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.004409886402819e-05,
+      "loss": 0.4797,
+      "step": 897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0022057564820594e-05,
+      "loss": 0.5166,
+      "step": 898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4551,
+      "step": 899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9977926205365433e-05,
+      "loss": 0.414,
+      "step": 900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9955836216742244e-05,
+      "loss": 0.5003,
+      "step": 901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.993373006998207e-05,
+      "loss": 0.4925,
+      "step": 902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9911607800962777e-05,
+      "loss": 0.4726,
+      "step": 903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.98894694455884e-05,
+      "loss": 0.4159,
+      "step": 904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9867315039789084e-05,
+      "loss": 0.5027,
+      "step": 905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9845144619521006e-05,
+      "loss": 0.4846,
+      "step": 906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.982295822076635e-05,
+      "loss": 0.4039,
+      "step": 907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.980075587953323e-05,
+      "loss": 0.4676,
+      "step": 908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9778537631855632e-05,
+      "loss": 0.4594,
+      "step": 909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975630351379335e-05,
+      "loss": 0.4944,
+      "step": 910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9734053561431962e-05,
+      "loss": 0.4147,
+      "step": 911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9711787810882713e-05,
+      "loss": 0.5273,
+      "step": 912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.96895062982825e-05,
+      "loss": 0.5289,
+      "step": 913
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.966720905979382e-05,
+      "loss": 0.4834,
+      "step": 914
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.964489613160466e-05,
+      "loss": 0.4642,
+      "step": 915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9622567549928494e-05,
+      "loss": 0.3462,
+      "step": 916
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.960022335100419e-05,
+      "loss": 0.3678,
+      "step": 917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.957786357109598e-05,
+      "loss": 0.4697,
+      "step": 918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9555488246493353e-05,
+      "loss": 0.4159,
+      "step": 919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953309741351105e-05,
+      "loss": 0.3855,
+      "step": 920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.951069110848899e-05,
+      "loss": 0.3472,
+      "step": 921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9488269367792174e-05,
+      "loss": 0.4075,
+      "step": 922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9465832227810672e-05,
+      "loss": 0.575,
+      "step": 923
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.944337972495955e-05,
+      "loss": 0.4752,
+      "step": 924
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9420911895678803e-05,
+      "loss": 0.4765,
+      "step": 925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9398428776433293e-05,
+      "loss": 0.4882,
+      "step": 926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9375930403712714e-05,
+      "loss": 0.4465,
+      "step": 927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.93534168140315e-05,
+      "loss": 0.5087,
+      "step": 928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9330888043928783e-05,
+      "loss": 0.4328,
+      "step": 929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9308344129968355e-05,
+      "loss": 0.4736,
+      "step": 930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9285785108738545e-05,
+      "loss": 0.5137,
+      "step": 931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9263211016852247e-05,
+      "loss": 0.3672,
+      "step": 932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.924062189094678e-05,
+      "loss": 0.5514,
+      "step": 933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.921801776768388e-05,
+      "loss": 0.4554,
+      "step": 934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9195398683749613e-05,
+      "loss": 0.4571,
+      "step": 935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9172764675854342e-05,
+      "loss": 0.4967,
+      "step": 936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.915011578073263e-05,
+      "loss": 0.3699,
+      "step": 937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9127452035143215e-05,
+      "loss": 0.4328,
+      "step": 938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9104773475868936e-05,
+      "loss": 0.5568,
+      "step": 939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9082080139716674e-05,
+      "loss": 0.453,
+      "step": 940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9059372063517285e-05,
+      "loss": 0.4655,
+      "step": 941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9036649284125567e-05,
+      "loss": 0.4041,
+      "step": 942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.901391183842016e-05,
+      "loss": 0.4362,
+      "step": 943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8991159763303516e-05,
+      "loss": 0.515,
+      "step": 944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8968393095701824e-05,
+      "loss": 0.3929,
+      "step": 945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8945611872564973e-05,
+      "loss": 0.3816,
+      "step": 946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.892281613086646e-05,
+      "loss": 0.4493,
+      "step": 947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8900005907603346e-05,
+      "loss": 0.4992,
+      "step": 948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8877181239796206e-05,
+      "loss": 0.4364,
+      "step": 949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8854342164489046e-05,
+      "loss": 0.416,
+      "step": 950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8831488718749262e-05,
+      "loss": 0.5375,
+      "step": 951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8808620939667566e-05,
+      "loss": 0.4237,
+      "step": 952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8785738864357948e-05,
+      "loss": 0.4749,
+      "step": 953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8762842529957585e-05,
+      "loss": 0.3595,
+      "step": 954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.87399319736268e-05,
+      "loss": 0.4154,
+      "step": 955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8717007232549005e-05,
+      "loss": 0.4597,
+      "step": 956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8694068343930623e-05,
+      "loss": 0.4274,
+      "step": 957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8671115345001048e-05,
+      "loss": 0.4606,
+      "step": 958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8648148273012565e-05,
+      "loss": 0.5212,
+      "step": 959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8625167165240303e-05,
+      "loss": 0.4739,
+      "step": 960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8602172058982193e-05,
+      "loss": 0.4964,
+      "step": 961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8579162991558837e-05,
+      "loss": 0.4129,
+      "step": 962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.855614000031354e-05,
+      "loss": 0.499,
+      "step": 963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8533103122612184e-05,
+      "loss": 0.4879,
+      "step": 964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.851005239584319e-05,
+      "loss": 0.441,
+      "step": 965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8486987857417464e-05,
+      "loss": 0.4832,
+      "step": 966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.846390954476832e-05,
+      "loss": 0.5384,
+      "step": 967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8440817495351427e-05,
+      "loss": 0.5132,
+      "step": 968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.841771174664476e-05,
+      "loss": 0.5381,
+      "step": 969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8394592336148517e-05,
+      "loss": 0.4289,
+      "step": 970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8371459301385065e-05,
+      "loss": 0.4353,
+      "step": 971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.834831267989889e-05,
+      "loss": 0.4918,
+      "step": 972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8325152509256536e-05,
+      "loss": 0.4548,
+      "step": 973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8301978827046518e-05,
+      "loss": 0.4237,
+      "step": 974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8278791670879303e-05,
+      "loss": 0.468,
+      "step": 975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.825559107838721e-05,
+      "loss": 0.3896,
+      "step": 976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8232377087224364e-05,
+      "loss": 0.4945,
+      "step": 977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.820914973506664e-05,
+      "loss": 0.5119,
+      "step": 978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8185909059611613e-05,
+      "loss": 0.4295,
+      "step": 979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.816265509857845e-05,
+      "loss": 0.5335,
+      "step": 980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8139387889707902e-05,
+      "loss": 0.475,
+      "step": 981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8116107470762222e-05,
+      "loss": 0.4626,
+      "step": 982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.809281387952509e-05,
+      "loss": 0.4227,
+      "step": 983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8069507153801575e-05,
+      "loss": 0.4126,
+      "step": 984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8046187331418063e-05,
+      "loss": 0.4588,
+      "step": 985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.8022854450222182e-05,
+      "loss": 0.4096,
+      "step": 986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7999508548082772e-05,
+      "loss": 0.4419,
+      "step": 987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7976149662889792e-05,
+      "loss": 0.5224,
+      "step": 988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7952777832554283e-05,
+      "loss": 0.5077,
+      "step": 989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7929393095008292e-05,
+      "loss": 0.4902,
+      "step": 990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.790599548820481e-05,
+      "loss": 0.4816,
+      "step": 991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7882585050117715e-05,
+      "loss": 0.4628,
+      "step": 992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.785916181874172e-05,
+      "loss": 0.3747,
+      "step": 993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7835725832092286e-05,
+      "loss": 0.4601,
+      "step": 994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7812277128205587e-05,
+      "loss": 0.5229,
+      "step": 995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7788815745138436e-05,
+      "loss": 0.5621,
+      "step": 996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7765341720968213e-05,
+      "loss": 0.4388,
+      "step": 997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7741855093792835e-05,
+      "loss": 0.4338,
+      "step": 998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.771835590173065e-05,
+      "loss": 0.4621,
+      "step": 999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7694844182920416e-05,
+      "loss": 0.6323,
+      "step": 1000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.767131997552121e-05,
+      "loss": 0.3781,
+      "step": 1001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.764778331771238e-05,
+      "loss": 0.424,
+      "step": 1002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7624234247693496e-05,
+      "loss": 0.3857,
+      "step": 1003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.760067280368426e-05,
+      "loss": 0.3731,
+      "step": 1004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7577099023924443e-05,
+      "loss": 0.4758,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.755351294667386e-05,
+      "loss": 0.4164,
+      "step": 1006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7529914610212283e-05,
+      "loss": 0.4915,
+      "step": 1007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7506304052839357e-05,
+      "loss": 0.4483,
+      "step": 1008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.748268131287459e-05,
+      "loss": 0.3905,
+      "step": 1009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.745904642865724e-05,
+      "loss": 0.4194,
+      "step": 1010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.743539943854629e-05,
+      "loss": 0.466,
+      "step": 1011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7411740380920366e-05,
+      "loss": 0.4445,
+      "step": 1012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.738806929417768e-05,
+      "loss": 0.3929,
+      "step": 1013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.736438621673596e-05,
+      "loss": 0.4155,
+      "step": 1014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7340691187032403e-05,
+      "loss": 0.4198,
+      "step": 1015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7316984243523604e-05,
+      "loss": 0.3928,
+      "step": 1016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7293265424685483e-05,
+      "loss": 0.4438,
+      "step": 1017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7269534769013247e-05,
+      "loss": 0.5108,
+      "step": 1018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.724579231502132e-05,
+      "loss": 0.5252,
+      "step": 1019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.722203810124324e-05,
+      "loss": 0.4138,
+      "step": 1020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7198272166231673e-05,
+      "loss": 0.4684,
+      "step": 1021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.717449454855828e-05,
+      "loss": 0.4603,
+      "step": 1022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.715070528681369e-05,
+      "loss": 0.3783,
+      "step": 1023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.712690441960744e-05,
+      "loss": 0.4003,
+      "step": 1024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7103091985567896e-05,
+      "loss": 0.4431,
+      "step": 1025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7079268023342188e-05,
+      "loss": 0.5347,
+      "step": 1026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.705543257159617e-05,
+      "loss": 0.419,
+      "step": 1027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7031585669014333e-05,
+      "loss": 0.5081,
+      "step": 1028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7007727354299756e-05,
+      "loss": 0.4887,
+      "step": 1029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6983857666174035e-05,
+      "loss": 0.3913,
+      "step": 1030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6959976643377238e-05,
+      "loss": 0.4256,
+      "step": 1031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6936084324667824e-05,
+      "loss": 0.5119,
+      "step": 1032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6912180748822572e-05,
+      "loss": 0.4653,
+      "step": 1033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6888265954636542e-05,
+      "loss": 0.3675,
+      "step": 1034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6864339980923015e-05,
+      "loss": 0.4606,
+      "step": 1035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.684040286651338e-05,
+      "loss": 0.4312,
+      "step": 1036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6816454650257138e-05,
+      "loss": 0.3623,
+      "step": 1037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6792495371021798e-05,
+      "loss": 0.5024,
+      "step": 1038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.676852506769283e-05,
+      "loss": 0.4821,
+      "step": 1039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.674454377917358e-05,
+      "loss": 0.4253,
+      "step": 1040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.672055154438524e-05,
+      "loss": 0.3446,
+      "step": 1041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6696548402266764e-05,
+      "loss": 0.4193,
+      "step": 1042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6672534391774797e-05,
+      "loss": 0.4299,
+      "step": 1043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.664850955188363e-05,
+      "loss": 0.4342,
+      "step": 1044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.662447392158514e-05,
+      "loss": 0.4704,
+      "step": 1045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6600427539888705e-05,
+      "loss": 0.5306,
+      "step": 1046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6576370445821155e-05,
+      "loss": 0.4104,
+      "step": 1047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6552302678426704e-05,
+      "loss": 0.4599,
+      "step": 1048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.652822427676689e-05,
+      "loss": 0.4404,
+      "step": 1049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6504135279920515e-05,
+      "loss": 0.4365,
+      "step": 1050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.648003572698357e-05,
+      "loss": 0.4182,
+      "step": 1051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6455925657069177e-05,
+      "loss": 0.4792,
+      "step": 1052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6431805109307546e-05,
+      "loss": 0.5278,
+      "step": 1053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.6407674122845863e-05,
+      "loss": 0.4588,
+      "step": 1054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.638353273684827e-05,
+      "loss": 0.3727,
+      "step": 1055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6359380990495797e-05,
+      "loss": 0.4419,
+      "step": 1056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.633521892298627e-05,
+      "loss": 0.5314,
+      "step": 1057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.631104657353428e-05,
+      "loss": 0.4877,
+      "step": 1058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6286863981371098e-05,
+      "loss": 0.468,
+      "step": 1059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6262671185744624e-05,
+      "loss": 0.3833,
+      "step": 1060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6238468225919312e-05,
+      "loss": 0.4503,
+      "step": 1061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6214255141176127e-05,
+      "loss": 0.4522,
+      "step": 1062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6190031970812433e-05,
+      "loss": 0.5101,
+      "step": 1063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6165798754142002e-05,
+      "loss": 0.5231,
+      "step": 1064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.614155553049489e-05,
+      "loss": 0.3934,
+      "step": 1065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6117302339217397e-05,
+      "loss": 0.468,
+      "step": 1066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6093039219672004e-05,
+      "loss": 0.4271,
+      "step": 1067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6068766211237292e-05,
+      "loss": 0.3602,
+      "step": 1068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.6044483353307923e-05,
+      "loss": 0.4987,
+      "step": 1069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.60201906852945e-05,
+      "loss": 0.4072,
+      "step": 1070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5995888246623587e-05,
+      "loss": 0.425,
+      "step": 1071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.597157607673759e-05,
+      "loss": 0.4508,
+      "step": 1072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5947254215094692e-05,
+      "loss": 0.4123,
+      "step": 1073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.592292270116884e-05,
+      "loss": 0.4581,
+      "step": 1074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.589858157444963e-05,
+      "loss": 0.3969,
+      "step": 1075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5874230874442238e-05,
+      "loss": 0.5364,
+      "step": 1076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.584987064066742e-05,
+      "loss": 0.5663,
+      "step": 1077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5825500912661367e-05,
+      "loss": 0.4083,
+      "step": 1078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.58011217299757e-05,
+      "loss": 0.4189,
+      "step": 1079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.577673313217737e-05,
+      "loss": 0.4287,
+      "step": 1080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5752335158848635e-05,
+      "loss": 0.4363,
+      "step": 1081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5727927849586937e-05,
+      "loss": 0.3991,
+      "step": 1082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.570351124400489e-05,
+      "loss": 0.3971,
+      "step": 1083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5679085381730192e-05,
+      "loss": 0.4288,
+      "step": 1084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5654650302405565e-05,
+      "loss": 0.4011,
+      "step": 1085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.563020604568868e-05,
+      "loss": 0.4906,
+      "step": 1086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5605752651252117e-05,
+      "loss": 0.463,
+      "step": 1087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5581290158783283e-05,
+      "loss": 0.437,
+      "step": 1088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.555681860798434e-05,
+      "loss": 0.4605,
+      "step": 1089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.553233803857217e-05,
+      "loss": 0.4407,
+      "step": 1090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.550784849027827e-05,
+      "loss": 0.5779,
+      "step": 1091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.548335000284873e-05,
+      "loss": 0.4734,
+      "step": 1092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5458842616044138e-05,
+      "loss": 0.5017,
+      "step": 1093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5434326369639523e-05,
+      "loss": 0.5016,
+      "step": 1094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5409801303424296e-05,
+      "loss": 0.4406,
+      "step": 1095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.53852674572022e-05,
+      "loss": 0.3701,
+      "step": 1096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.536072487079119e-05,
+      "loss": 0.4359,
+      "step": 1097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5336173584023433e-05,
+      "loss": 0.4009,
+      "step": 1098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.531161363674522e-05,
+      "loss": 0.4721,
+      "step": 1099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.528704506881687e-05,
+      "loss": 0.4627,
+      "step": 1100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5262467920112728e-05,
+      "loss": 0.4579,
+      "step": 1101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.523788223052104e-05,
+      "loss": 0.4503,
+      "step": 1102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5213288039943934e-05,
+      "loss": 0.44,
+      "step": 1103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5188685388297315e-05,
+      "loss": 0.4562,
+      "step": 1104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5164074315510826e-05,
+      "loss": 0.4254,
+      "step": 1105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5139454861527798e-05,
+      "loss": 0.4079,
+      "step": 1106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5114827066305122e-05,
+      "loss": 0.425,
+      "step": 1107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5090190969813265e-05,
+      "loss": 0.4694,
+      "step": 1108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.506554661203615e-05,
+      "loss": 0.3781,
+      "step": 1109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.504089403297112e-05,
+      "loss": 0.4381,
+      "step": 1110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.501623327262884e-05,
+      "loss": 0.4836,
+      "step": 1111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4991564371033277e-05,
+      "loss": 0.4924,
+      "step": 1112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4966887368221588e-05,
+      "loss": 0.4871,
+      "step": 1113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4942202304244093e-05,
+      "loss": 0.3833,
+      "step": 1114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4917509219164196e-05,
+      "loss": 0.4862,
+      "step": 1115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.489280815305831e-05,
+      "loss": 0.3907,
+      "step": 1116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4868099146015806e-05,
+      "loss": 0.4262,
+      "step": 1117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.484338223813895e-05,
+      "loss": 0.4644,
+      "step": 1118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4818657469542817e-05,
+      "loss": 0.5781,
+      "step": 1119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4793924880355246e-05,
+      "loss": 0.4131,
+      "step": 1120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.476918451071677e-05,
+      "loss": 0.4225,
+      "step": 1121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4744436400780552e-05,
+      "loss": 0.4566,
+      "step": 1122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4719680590712302e-05,
+      "loss": 0.4743,
+      "step": 1123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.469491712069026e-05,
+      "loss": 0.4672,
+      "step": 1124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4670146030905063e-05,
+      "loss": 0.4621,
+      "step": 1125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.464536736155973e-05,
+      "loss": 0.4427,
+      "step": 1126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.462058115286959e-05,
+      "loss": 0.5348,
+      "step": 1127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4595787445062183e-05,
+      "loss": 0.4158,
+      "step": 1128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4570986278377243e-05,
+      "loss": 0.456,
+      "step": 1129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4546177693066606e-05,
+      "loss": 0.4704,
+      "step": 1130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.4521361729394144e-05,
+      "loss": 0.4248,
+      "step": 1131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4496538427635695e-05,
+      "loss": 0.4528,
+      "step": 1132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.447170782807903e-05,
+      "loss": 0.3969,
+      "step": 1133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4446869971023737e-05,
+      "loss": 0.5218,
+      "step": 1134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4422024896781205e-05,
+      "loss": 0.4691,
+      "step": 1135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.439717264567452e-05,
+      "loss": 0.4961,
+      "step": 1136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.437231325803843e-05,
+      "loss": 0.4512,
+      "step": 1137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4347446774219257e-05,
+      "loss": 0.3665,
+      "step": 1138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.432257323457484e-05,
+      "loss": 0.4508,
+      "step": 1139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4297692679474473e-05,
+      "loss": 0.4737,
+      "step": 1140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.427280514929883e-05,
+      "loss": 0.4531,
+      "step": 1141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4247910684439914e-05,
+      "loss": 0.4176,
+      "step": 1142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.422300932530098e-05,
+      "loss": 0.3736,
+      "step": 1143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.419810111229646e-05,
+      "loss": 0.4385,
+      "step": 1144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4173186085851928e-05,
+      "loss": 0.5374,
+      "step": 1145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.414826428640402e-05,
+      "loss": 0.4099,
+      "step": 1146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4123335754400336e-05,
+      "loss": 0.4313,
+      "step": 1147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.409840053029942e-05,
+      "loss": 0.4646,
+      "step": 1148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4073458654570688e-05,
+      "loss": 0.465,
+      "step": 1149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.404851016769433e-05,
+      "loss": 0.4226,
+      "step": 1150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4023555110161278e-05,
+      "loss": 0.3571,
+      "step": 1151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.399859352247313e-05,
+      "loss": 0.3844,
+      "step": 1152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.397362544514208e-05,
+      "loss": 0.5416,
+      "step": 1153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3948650918690847e-05,
+      "loss": 0.4308,
+      "step": 1154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.392366998365263e-05,
+      "loss": 0.4977,
+      "step": 1155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3898682680571013e-05,
+      "loss": 0.4236,
+      "step": 1156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3873689049999933e-05,
+      "loss": 0.4653,
+      "step": 1157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.384868913250358e-05,
+      "loss": 0.4379,
+      "step": 1158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.382368296865636e-05,
+      "loss": 0.4362,
+      "step": 1159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.379867059904281e-05,
+      "loss": 0.3921,
+      "step": 1160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3773652064257543e-05,
+      "loss": 0.4038,
+      "step": 1161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.374862740490517e-05,
+      "loss": 0.49,
+      "step": 1162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.372359666160025e-05,
+      "loss": 0.4178,
+      "step": 1163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3698559874967218e-05,
+      "loss": 0.4719,
+      "step": 1164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3673517085640302e-05,
+      "loss": 0.4357,
+      "step": 1165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3648468334263495e-05,
+      "loss": 0.4454,
+      "step": 1166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.362341366149046e-05,
+      "loss": 0.4007,
+      "step": 1167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3598353107984445e-05,
+      "loss": 0.3775,
+      "step": 1168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3573286714418272e-05,
+      "loss": 0.4747,
+      "step": 1169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3548214521474236e-05,
+      "loss": 0.4161,
+      "step": 1170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3523136569844035e-05,
+      "loss": 0.4336,
+      "step": 1171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3498052900228717e-05,
+      "loss": 0.4365,
+      "step": 1172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.4029,
+      "step": 1173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3447868569893263e-05,
+      "loss": 0.4487,
+      "step": 1174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.342276799062136e-05,
+      "loss": 0.4326,
+      "step": 1175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3397661856260688e-05,
+      "loss": 0.4095,
+      "step": 1176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3372550207558024e-05,
+      "loss": 0.4156,
+      "step": 1177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3347433085269106e-05,
+      "loss": 0.3514,
+      "step": 1178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.332231053015856e-05,
+      "loss": 0.4069,
+      "step": 1179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3297182582999823e-05,
+      "loss": 0.4305,
+      "step": 1180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3272049284575092e-05,
+      "loss": 0.4248,
+      "step": 1181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3246910675675238e-05,
+      "loss": 0.4651,
+      "step": 1182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3221766797099758e-05,
+      "loss": 0.3965,
+      "step": 1183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3196617689656696e-05,
+      "loss": 0.4684,
+      "step": 1184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3171463394162586e-05,
+      "loss": 0.4687,
+      "step": 1185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3146303951442383e-05,
+      "loss": 0.4663,
+      "step": 1186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3121139402329392e-05,
+      "loss": 0.3605,
+      "step": 1187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3095969787665217e-05,
+      "loss": 0.4569,
+      "step": 1188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3070795148299666e-05,
+      "loss": 0.5684,
+      "step": 1189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3045615525090713e-05,
+      "loss": 0.5205,
+      "step": 1190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3020430958904413e-05,
+      "loss": 0.5152,
+      "step": 1191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2995241490614858e-05,
+      "loss": 0.3836,
+      "step": 1192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2970047161104078e-05,
+      "loss": 0.4042,
+      "step": 1193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2944848011262e-05,
+      "loss": 0.4925,
+      "step": 1194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2919644081986386e-05,
+      "loss": 0.5038,
+      "step": 1195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2894435414182738e-05,
+      "loss": 0.4234,
+      "step": 1196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2869222048764254e-05,
+      "loss": 0.4366,
+      "step": 1197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2844004026651764e-05,
+      "loss": 0.4259,
+      "step": 1198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2818781388773636e-05,
+      "loss": 0.5058,
+      "step": 1199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.279355417606575e-05,
+      "loss": 0.364,
+      "step": 1200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.27683224294714e-05,
+      "loss": 0.4578,
+      "step": 1201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2743086189941247e-05,
+      "loss": 0.5212,
+      "step": 1202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.271784549843323e-05,
+      "loss": 0.453,
+      "step": 1203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.269260039591253e-05,
+      "loss": 0.4887,
+      "step": 1204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.266735092335148e-05,
+      "loss": 0.4965,
+      "step": 1205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2642097121729493e-05,
+      "loss": 0.5081,
+      "step": 1206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2616839032033028e-05,
+      "loss": 0.5854,
+      "step": 1207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2591576695255485e-05,
+      "loss": 0.4537,
+      "step": 1208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.256631015239718e-05,
+      "loss": 0.4634,
+      "step": 1209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2541039444465245e-05,
+      "loss": 0.4589,
+      "step": 1210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2515764612473553e-05,
+      "loss": 0.5212,
+      "step": 1211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.249048569744269e-05,
+      "loss": 0.321,
+      "step": 1212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2465202740399873e-05,
+      "loss": 0.4585,
+      "step": 1213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.243991578237886e-05,
+      "loss": 0.5032,
+      "step": 1214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.241462486441992e-05,
+      "loss": 0.5101,
+      "step": 1215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2389330027569748e-05,
+      "loss": 0.4426,
+      "step": 1216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2364031312881383e-05,
+      "loss": 0.4038,
+      "step": 1217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2338728761414176e-05,
+      "loss": 0.5358,
+      "step": 1218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2313422414233693e-05,
+      "loss": 0.4887,
+      "step": 1219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2288112312411667e-05,
+      "loss": 0.3734,
+      "step": 1220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.226279849702592e-05,
+      "loss": 0.456,
+      "step": 1221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2237481009160305e-05,
+      "loss": 0.391,
+      "step": 1222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2212159889904634e-05,
+      "loss": 0.368,
+      "step": 1223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2186835180354614e-05,
+      "loss": 0.4879,
+      "step": 1224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2161506921611773e-05,
+      "loss": 0.4322,
+      "step": 1225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2136175154783407e-05,
+      "loss": 0.3974,
+      "step": 1226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2110839920982503e-05,
+      "loss": 0.3989,
+      "step": 1227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2085501261327672e-05,
+      "loss": 0.4047,
+      "step": 1228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.206015921694309e-05,
+      "loss": 0.4466,
+      "step": 1229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.203481382895842e-05,
+      "loss": 0.3834,
+      "step": 1230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.200946513850876e-05,
+      "loss": 0.4185,
+      "step": 1231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.198411318673456e-05,
+      "loss": 0.5257,
+      "step": 1232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.1958758014781564e-05,
+      "loss": 0.3754,
+      "step": 1233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1933399663800747e-05,
+      "loss": 0.3548,
+      "step": 1234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1908038174948247e-05,
+      "loss": 0.4328,
+      "step": 1235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1882673589385273e-05,
+      "loss": 0.4637,
+      "step": 1236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.185730594827809e-05,
+      "loss": 0.4238,
+      "step": 1237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.18319352927979e-05,
+      "loss": 0.4428,
+      "step": 1238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1806561664120813e-05,
+      "loss": 0.4977,
+      "step": 1239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1781185103427744e-05,
+      "loss": 0.4429,
+      "step": 1240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1755805651904396e-05,
+      "loss": 0.3797,
+      "step": 1241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1730423350741128e-05,
+      "loss": 0.358,
+      "step": 1242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1705038241132954e-05,
+      "loss": 0.4037,
+      "step": 1243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1679650364279426e-05,
+      "loss": 0.3584,
+      "step": 1244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1654259761384602e-05,
+      "loss": 0.4922,
+      "step": 1245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.162886647365696e-05,
+      "loss": 0.3407,
+      "step": 1246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.160347054230934e-05,
+      "loss": 0.4036,
+      "step": 1247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.157807200855885e-05,
+      "loss": 0.4846,
+      "step": 1248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1552670913626846e-05,
+      "loss": 0.486,
+      "step": 1249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1527267298738833e-05,
+      "loss": 0.3524,
+      "step": 1250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1501861205124405e-05,
+      "loss": 0.3637,
+      "step": 1251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1476452674017173e-05,
+      "loss": 0.5161,
+      "step": 1252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1451041746654722e-05,
+      "loss": 0.4617,
+      "step": 1253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1425628464278508e-05,
+      "loss": 0.3469,
+      "step": 1254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1400212868133817e-05,
+      "loss": 0.4052,
+      "step": 1255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1374794999469683e-05,
+      "loss": 0.4504,
+      "step": 1256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.134937489953884e-05,
+      "loss": 0.5516,
+      "step": 1257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1323952609597642e-05,
+      "loss": 0.4713,
+      "step": 1258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1298528170905976e-05,
+      "loss": 0.4538,
+      "step": 1259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1273101624727257e-05,
+      "loss": 0.4509,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1247673012328276e-05,
+      "loss": 0.4761,
+      "step": 1261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1222242374979202e-05,
+      "loss": 0.4158,
+      "step": 1262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1196809753953498e-05,
+      "loss": 0.4111,
+      "step": 1263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1171375190527824e-05,
+      "loss": 0.4153,
+      "step": 1264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1145938725982e-05,
+      "loss": 0.4252,
+      "step": 1265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1120500401598946e-05,
+      "loss": 0.5109,
+      "step": 1266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.109506025866458e-05,
+      "loss": 0.4969,
+      "step": 1267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1069618338467784e-05,
+      "loss": 0.4846,
+      "step": 1268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.104417468230033e-05,
+      "loss": 0.4118,
+      "step": 1269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.1018729331456784e-05,
+      "loss": 0.4205,
+      "step": 1270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.099328232723448e-05,
+      "loss": 0.4065,
+      "step": 1271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.096783371093344e-05,
+      "loss": 0.4839,
+      "step": 1272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0942383523856293e-05,
+      "loss": 0.3817,
+      "step": 1273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0916931807308214e-05,
+      "loss": 0.4728,
+      "step": 1274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0891478602596872e-05,
+      "loss": 0.4048,
+      "step": 1275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.086602395103234e-05,
+      "loss": 0.3986,
+      "step": 1276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.084056789392704e-05,
+      "loss": 0.38,
+      "step": 1277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0815110472595688e-05,
+      "loss": 0.5145,
+      "step": 1278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.07896517283552e-05,
+      "loss": 0.3834,
+      "step": 1279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0764191702524646e-05,
+      "loss": 0.413,
+      "step": 1280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0738730436425182e-05,
+      "loss": 0.4216,
+      "step": 1281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.071326797137996e-05,
+      "loss": 0.4346,
+      "step": 1282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.068780434871408e-05,
+      "loss": 0.3643,
+      "step": 1283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.0662339609754543e-05,
+      "loss": 0.4364,
+      "step": 1284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.063687379583014e-05,
+      "loss": 0.4851,
+      "step": 1285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0611406948271403e-05,
+      "loss": 0.2894,
+      "step": 1286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.058593910841058e-05,
+      "loss": 0.4577,
+      "step": 1287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0560470317581473e-05,
+      "loss": 0.4158,
+      "step": 1288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0535000617119466e-05,
+      "loss": 0.4388,
+      "step": 1289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0509530048361417e-05,
+      "loss": 0.4745,
+      "step": 1290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.048405865264557e-05,
+      "loss": 0.4252,
+      "step": 1291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0458586471311537e-05,
+      "loss": 0.504,
+      "step": 1292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.043311354570019e-05,
+      "loss": 0.4089,
+      "step": 1293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0407639917153615e-05,
+      "loss": 0.4012,
+      "step": 1294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0382165627015042e-05,
+      "loss": 0.3895,
+      "step": 1295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0356690716628756e-05,
+      "loss": 0.3783,
+      "step": 1296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0331215227340086e-05,
+      "loss": 0.4031,
+      "step": 1297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0305739200495254e-05,
+      "loss": 0.3929,
+      "step": 1298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.028026267744139e-05,
+      "loss": 0.4364,
+      "step": 1299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.025478569952641e-05,
+      "loss": 0.3677,
+      "step": 1300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.022930830809898e-05,
+      "loss": 0.4435,
+      "step": 1301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0203830544508435e-05,
+      "loss": 0.4736,
+      "step": 1302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.017835245010471e-05,
+      "loss": 0.3681,
+      "step": 1303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.015287406623827e-05,
+      "loss": 0.3983,
+      "step": 1304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.012739543426007e-05,
+      "loss": 0.4694,
+      "step": 1305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.010191659552145e-05,
+      "loss": 0.4801,
+      "step": 1306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.007643759137408e-05,
+      "loss": 0.3786,
+      "step": 1307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0050958463169928e-05,
+      "loss": 0.3938,
+      "step": 1308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0025479252261142e-05,
+      "loss": 0.4381,
+      "step": 1309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.4025,
+      "step": 1310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997452074773886e-05,
+      "loss": 0.4013,
+      "step": 1311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9949041536830072e-05,
+      "loss": 0.4034,
+      "step": 1312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9923562408625922e-05,
+      "loss": 0.4623,
+      "step": 1313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9898083404478563e-05,
+      "loss": 0.355,
+      "step": 1314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.987260456573994e-05,
+      "loss": 0.4129,
+      "step": 1315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9847125933761735e-05,
+      "loss": 0.4309,
+      "step": 1316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9821647549895296e-05,
+      "loss": 0.4476,
+      "step": 1317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9796169455491568e-05,
+      "loss": 0.3531,
+      "step": 1318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.977069169190102e-05,
+      "loss": 0.4144,
+      "step": 1319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9745214300473595e-05,
+      "loss": 0.4567,
+      "step": 1320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719737322558615e-05,
+      "loss": 0.5224,
+      "step": 1321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9694260799504752e-05,
+      "loss": 0.3433,
+      "step": 1322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.966878477265992e-05,
+      "loss": 0.3549,
+      "step": 1323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.964330928337124e-05,
+      "loss": 0.4286,
+      "step": 1324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.961783437298496e-05,
+      "loss": 0.4803,
+      "step": 1325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9592360082846395e-05,
+      "loss": 0.4505,
+      "step": 1326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9566886454299815e-05,
+      "loss": 0.5434,
+      "step": 1327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.954141352868847e-05,
+      "loss": 0.4652,
+      "step": 1328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9515941347354435e-05,
+      "loss": 0.4657,
+      "step": 1329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.949046995163859e-05,
+      "loss": 0.434,
+      "step": 1330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9464999382880534e-05,
+      "loss": 0.4295,
+      "step": 1331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.943952968241854e-05,
+      "loss": 0.4362,
+      "step": 1332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9414060891589433e-05,
+      "loss": 0.4012,
+      "step": 1333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.93885930517286e-05,
+      "loss": 0.4211,
+      "step": 1334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9363126204169867e-05,
+      "loss": 0.4008,
+      "step": 1335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.933766039024546e-05,
+      "loss": 0.4026,
+      "step": 1336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.931219565128592e-05,
+      "loss": 0.4702,
+      "step": 1337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9286732028620054e-05,
+      "loss": 0.3801,
+      "step": 1338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9261269563574828e-05,
+      "loss": 0.4007,
+      "step": 1339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9235808297475357e-05,
+      "loss": 0.4387,
+      "step": 1340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9210348271644804e-05,
+      "loss": 0.3962,
+      "step": 1341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.918488952740432e-05,
+      "loss": 0.4328,
+      "step": 1342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.915943210607296e-05,
+      "loss": 0.4631,
+      "step": 1343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9133976048967674e-05,
+      "loss": 0.4393,
+      "step": 1344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9108521397403138e-05,
+      "loss": 0.4507,
+      "step": 1345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9083068192691792e-05,
+      "loss": 0.4775,
+      "step": 1346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9057616476143714e-05,
+      "loss": 0.5027,
+      "step": 1347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9032166289066562e-05,
+      "loss": 0.3933,
+      "step": 1348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9006717672765522e-05,
+      "loss": 0.4274,
+      "step": 1349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8981270668543222e-05,
+      "loss": 0.3531,
+      "step": 1350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.895582531769968e-05,
+      "loss": 0.4444,
+      "step": 1351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.893038166153222e-05,
+      "loss": 0.349,
+      "step": 1352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904939741335425e-05,
+      "loss": 0.4709,
+      "step": 1353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887949959840106e-05,
+      "loss": 0.3518,
+      "step": 1354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8854061274018004e-05,
+      "loss": 0.3345,
+      "step": 1355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8828624809472183e-05,
+      "loss": 0.4524,
+      "step": 1356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.880319024604651e-05,
+      "loss": 0.4088,
+      "step": 1357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.87777576250208e-05,
+      "loss": 0.385,
+      "step": 1358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.875232698767173e-05,
+      "loss": 0.3562,
+      "step": 1359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.872689837527275e-05,
+      "loss": 0.3447,
+      "step": 1360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8701471829094024e-05,
+      "loss": 0.4479,
+      "step": 1361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.867604739040236e-05,
+      "loss": 0.4506,
+      "step": 1362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8650625100461163e-05,
+      "loss": 0.4587,
+      "step": 1363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8625205000530323e-05,
+      "loss": 0.375,
+      "step": 1364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8599787131866193e-05,
+      "loss": 0.4099,
+      "step": 1365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8574371535721495e-05,
+      "loss": 0.3858,
+      "step": 1366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.854895825334528e-05,
+      "loss": 0.4271,
+      "step": 1367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8523547325982827e-05,
+      "loss": 0.3395,
+      "step": 1368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8498138794875605e-05,
+      "loss": 0.3923,
+      "step": 1369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8472732701261173e-05,
+      "loss": 0.3676,
+      "step": 1370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.844732908637316e-05,
+      "loss": 0.3438,
+      "step": 1371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8421927991441156e-05,
+      "loss": 0.407,
+      "step": 1372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8396529457690663e-05,
+      "loss": 0.4314,
+      "step": 1373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8371133526343036e-05,
+      "loss": 0.4554,
+      "step": 1374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8345740238615408e-05,
+      "loss": 0.4154,
+      "step": 1375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8320349635720584e-05,
+      "loss": 0.36,
+      "step": 1376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8294961758867057e-05,
+      "loss": 0.4501,
+      "step": 1377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.826957664925888e-05,
+      "loss": 0.4288,
+      "step": 1378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.824419434809561e-05,
+      "loss": 0.4465,
+      "step": 1379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8218814896572256e-05,
+      "loss": 0.4284,
+      "step": 1380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8193438335879197e-05,
+      "loss": 0.4272,
+      "step": 1381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8168064707202107e-05,
+      "loss": 0.3847,
+      "step": 1382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8142694051721917e-05,
+      "loss": 0.4297,
+      "step": 1383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8117326410614733e-05,
+      "loss": 0.3611,
+      "step": 1384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8091961825051757e-05,
+      "loss": 0.443,
+      "step": 1385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8066600336199253e-05,
+      "loss": 0.3463,
+      "step": 1386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8041241985218436e-05,
+      "loss": 0.3886,
+      "step": 1387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.801588681326545e-05,
+      "loss": 0.4283,
+      "step": 1388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7990534861491247e-05,
+      "loss": 0.4413,
+      "step": 1389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7965186171041584e-05,
+      "loss": 0.4278,
+      "step": 1390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7939840783056918e-05,
+      "loss": 0.4364,
+      "step": 1391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.791449873867233e-05,
+      "loss": 0.4501,
+      "step": 1392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7889160079017497e-05,
+      "loss": 0.4162,
+      "step": 1393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7863824845216603e-05,
+      "loss": 0.4063,
+      "step": 1394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7838493078388233e-05,
+      "loss": 0.4442,
+      "step": 1395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7813164819645393e-05,
+      "loss": 0.3719,
+      "step": 1396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7787840110095373e-05,
+      "loss": 0.4465,
+      "step": 1397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.77625189908397e-05,
+      "loss": 0.407,
+      "step": 1398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7737201502974083e-05,
+      "loss": 0.3552,
+      "step": 1399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7711887687588343e-05,
+      "loss": 0.4426,
+      "step": 1400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7686577585766314e-05,
+      "loss": 0.4152,
+      "step": 1401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.766127123858583e-05,
+      "loss": 0.4254,
+      "step": 1402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.763596868711862e-05,
+      "loss": 0.4186,
+      "step": 1403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7610669972430255e-05,
+      "loss": 0.4061,
+      "step": 1404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.758537513558008e-05,
+      "loss": 0.3993,
+      "step": 1405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7560084217621146e-05,
+      "loss": 0.3364,
+      "step": 1406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7534797259600137e-05,
+      "loss": 0.3983,
+      "step": 1407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7509514302557317e-05,
+      "loss": 0.3154,
+      "step": 1408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7484235387526457e-05,
+      "loss": 0.4569,
+      "step": 1409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7458960555534762e-05,
+      "loss": 0.4311,
+      "step": 1410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7433689847602816e-05,
+      "loss": 0.3951,
+      "step": 1411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7408423304744518e-05,
+      "loss": 0.416,
+      "step": 1412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7383160967966982e-05,
+      "loss": 0.4447,
+      "step": 1413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7357902878270513e-05,
+      "loss": 0.4862,
+      "step": 1414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7332649076648528e-05,
+      "loss": 0.4574,
+      "step": 1415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7307399604087473e-05,
+      "loss": 0.4151,
+      "step": 1416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.728215450156677e-05,
+      "loss": 0.4807,
+      "step": 1417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7256913810058766e-05,
+      "loss": 0.4472,
+      "step": 1418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7231677570528606e-05,
+      "loss": 0.4489,
+      "step": 1419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7206445823934255e-05,
+      "loss": 0.3738,
+      "step": 1420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7181218611226368e-05,
+      "loss": 0.4081,
+      "step": 1421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7155995973348243e-05,
+      "loss": 0.3673,
+      "step": 1422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7130777951235746e-05,
+      "loss": 0.4299,
+      "step": 1423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.710556458581726e-05,
+      "loss": 0.3332,
+      "step": 1424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.708035591801362e-05,
+      "loss": 0.4721,
+      "step": 1425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7055151988738003e-05,
+      "loss": 0.362,
+      "step": 1426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7029952838895925e-05,
+      "loss": 0.3586,
+      "step": 1427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7004758509385145e-05,
+      "loss": 0.4329,
+      "step": 1428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6979569041095587e-05,
+      "loss": 0.4081,
+      "step": 1429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6954384474909287e-05,
+      "loss": 0.4524,
+      "step": 1430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6929204851700344e-05,
+      "loss": 0.4668,
+      "step": 1431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.690403021233479e-05,
+      "loss": 0.2983,
+      "step": 1432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687886059767061e-05,
+      "loss": 0.5635,
+      "step": 1433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685369604855762e-05,
+      "loss": 0.4608,
+      "step": 1434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6828536605837417e-05,
+      "loss": 0.3714,
+      "step": 1435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6803382310343307e-05,
+      "loss": 0.4693,
+      "step": 1436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6778233202900253e-05,
+      "loss": 0.4112,
+      "step": 1437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675308932432477e-05,
+      "loss": 0.3936,
+      "step": 1438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6727950715424915e-05,
+      "loss": 0.4494,
+      "step": 1439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.670281741700018e-05,
+      "loss": 0.4889,
+      "step": 1440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.667768946984145e-05,
+      "loss": 0.4373,
+      "step": 1441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.66525669147309e-05,
+      "loss": 0.4686,
+      "step": 1442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.662744979244199e-05,
+      "loss": 0.4578,
+      "step": 1443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6602338143739322e-05,
+      "loss": 0.3479,
+      "step": 1444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6577232009378642e-05,
+      "loss": 0.3831,
+      "step": 1445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6552131430106744e-05,
+      "loss": 0.3862,
+      "step": 1446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.4331,
+      "step": 1447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6501947099771293e-05,
+      "loss": 0.469,
+      "step": 1448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6476863430155975e-05,
+      "loss": 0.3878,
+      "step": 1449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.645178547852577e-05,
+      "loss": 0.3877,
+      "step": 1450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6426713285581734e-05,
+      "loss": 0.3442,
+      "step": 1451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6401646892015565e-05,
+      "loss": 0.4462,
+      "step": 1452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.637658633850955e-05,
+      "loss": 0.5026,
+      "step": 1453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6351531665736505e-05,
+      "loss": 0.3636,
+      "step": 1454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.63264829143597e-05,
+      "loss": 0.4285,
+      "step": 1455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6301440125032792e-05,
+      "loss": 0.4283,
+      "step": 1456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6276403338399756e-05,
+      "loss": 0.3378,
+      "step": 1457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.625137259509484e-05,
+      "loss": 0.5072,
+      "step": 1458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6226347935742463e-05,
+      "loss": 0.3462,
+      "step": 1459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6201329400957194e-05,
+      "loss": 0.4032,
+      "step": 1460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6176317031343644e-05,
+      "loss": 0.472,
+      "step": 1461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6151310867496426e-05,
+      "loss": 0.4356,
+      "step": 1462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6126310950000074e-05,
+      "loss": 0.4256,
+      "step": 1463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.610131731942899e-05,
+      "loss": 0.4369,
+      "step": 1464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6076330016347376e-05,
+      "loss": 0.4454,
+      "step": 1465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6051349081309153e-05,
+      "loss": 0.4752,
+      "step": 1466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6026374554857923e-05,
+      "loss": 0.5108,
+      "step": 1467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6001406477526878e-05,
+      "loss": 0.4649,
+      "step": 1468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.597644488983873e-05,
+      "loss": 0.4556,
+      "step": 1469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5951489832305678e-05,
+      "loss": 0.375,
+      "step": 1470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.592654134542932e-05,
+      "loss": 0.5046,
+      "step": 1471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5901599469700584e-05,
+      "loss": 0.3824,
+      "step": 1472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5876664245599668e-05,
+      "loss": 0.4381,
+      "step": 1473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.585173571359599e-05,
+      "loss": 0.3963,
+      "step": 1474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5826813914148075e-05,
+      "loss": 0.4051,
+      "step": 1475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5801898887703546e-05,
+      "loss": 0.3555,
+      "step": 1476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5776990674699026e-05,
+      "loss": 0.4563,
+      "step": 1477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.575208931556009e-05,
+      "loss": 0.4204,
+      "step": 1478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.572719485070117e-05,
+      "loss": 0.4339,
+      "step": 1479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5702307320525537e-05,
+      "loss": 0.474,
+      "step": 1480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5677426765425166e-05,
+      "loss": 0.4084,
+      "step": 1481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.565255322578075e-05,
+      "loss": 0.436,
+      "step": 1482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5627686741961574e-05,
+      "loss": 0.4471,
+      "step": 1483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5602827354325486e-05,
+      "loss": 0.4283,
+      "step": 1484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5577975103218795e-05,
+      "loss": 0.3798,
+      "step": 1485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5553130028976273e-05,
+      "loss": 0.424,
+      "step": 1486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.552829217192098e-05,
+      "loss": 0.4124,
+      "step": 1487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5503461572364308e-05,
+      "loss": 0.3336,
+      "step": 1488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5478638270605863e-05,
+      "loss": 0.3563,
+      "step": 1489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5453822306933398e-05,
+      "loss": 0.5313,
+      "step": 1490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.542901372162276e-05,
+      "loss": 0.3748,
+      "step": 1491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5404212554937827e-05,
+      "loss": 0.3934,
+      "step": 1492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.537941884713042e-05,
+      "loss": 0.3765,
+      "step": 1493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5354632638440274e-05,
+      "loss": 0.3729,
+      "step": 1494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.532985396909494e-05,
+      "loss": 0.435,
+      "step": 1495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5305082879309742e-05,
+      "loss": 0.4453,
+      "step": 1496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5280319409287698e-05,
+      "loss": 0.4261,
+      "step": 1497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5255563599219454e-05,
+      "loss": 0.3621,
+      "step": 1498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5230815489283236e-05,
+      "loss": 0.4562,
+      "step": 1499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5206075119644762e-05,
+      "loss": 0.3921,
+      "step": 1500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.518134253045719e-05,
+      "loss": 0.3733,
+      "step": 1501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5156617761861055e-05,
+      "loss": 0.398,
+      "step": 1502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5131900853984194e-05,
+      "loss": 0.4283,
+      "step": 1503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5107191846941693e-05,
+      "loss": 0.403,
+      "step": 1504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5082490780835812e-05,
+      "loss": 0.3489,
+      "step": 1505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5057797695755914e-05,
+      "loss": 0.3588,
+      "step": 1506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5033112631778419e-05,
+      "loss": 0.4061,
+      "step": 1507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5008435628966728e-05,
+      "loss": 0.4128,
+      "step": 1508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.498376672737116e-05,
+      "loss": 0.3596,
+      "step": 1509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4959105967028882e-05,
+      "loss": 0.4155,
+      "step": 1510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4934453387963852e-05,
+      "loss": 0.4899,
+      "step": 1511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4909809030186742e-05,
+      "loss": 0.3835,
+      "step": 1512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4885172933694883e-05,
+      "loss": 0.3662,
+      "step": 1513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4860545138472212e-05,
+      "loss": 0.362,
+      "step": 1514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4835925684489173e-05,
+      "loss": 0.3559,
+      "step": 1515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4811314611702688e-05,
+      "loss": 0.4203,
+      "step": 1516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4786711960056077e-05,
+      "loss": 0.4107,
+      "step": 1517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4762117769478965e-05,
+      "loss": 0.3706,
+      "step": 1518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4737532079887277e-05,
+      "loss": 0.4199,
+      "step": 1519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4712954931183135e-05,
+      "loss": 0.4369,
+      "step": 1520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4688386363254789e-05,
+      "loss": 0.4001,
+      "step": 1521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.466382641597657e-05,
+      "loss": 0.3525,
+      "step": 1522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4639275129208824e-05,
+      "loss": 0.4013,
+      "step": 1523
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4614732542797812e-05,
+      "loss": 0.352,
+      "step": 1524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4590198696575705e-05,
+      "loss": 0.4188,
+      "step": 1525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4565673630360482e-05,
+      "loss": 0.38,
+      "step": 1526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4541157383955864e-05,
+      "loss": 0.3034,
+      "step": 1527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4516649997151268e-05,
+      "loss": 0.4518,
+      "step": 1528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4492151509721737e-05,
+      "loss": 0.4171,
+      "step": 1529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4467661961427839e-05,
+      "loss": 0.3683,
+      "step": 1530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4443181392015664e-05,
+      "loss": 0.3852,
+      "step": 1531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4418709841216725e-05,
+      "loss": 0.3851,
+      "step": 1532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4394247348747885e-05,
+      "loss": 0.3651,
+      "step": 1533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4369793954311323e-05,
+      "loss": 0.3876,
+      "step": 1534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4345349697594437e-05,
+      "loss": 0.388,
+      "step": 1535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4320914618269815e-05,
+      "loss": 0.4123,
+      "step": 1536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4296488755995115e-05,
+      "loss": 0.4046,
+      "step": 1537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.427207215041307e-05,
+      "loss": 0.4739,
+      "step": 1538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4247664841151375e-05,
+      "loss": 0.5016,
+      "step": 1539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.422326686782263e-05,
+      "loss": 0.3764,
+      "step": 1540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4198878270024307e-05,
+      "loss": 0.3476,
+      "step": 1541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4174499087338643e-05,
+      "loss": 0.3729,
+      "step": 1542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4150129359332589e-05,
+      "loss": 0.4212,
+      "step": 1543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4125769125557765e-05,
+      "loss": 0.4311,
+      "step": 1544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4101418425550377e-05,
+      "loss": 0.4601,
+      "step": 1545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4077077298831159e-05,
+      "loss": 0.3286,
+      "step": 1546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4052745784905306e-05,
+      "loss": 0.3821,
+      "step": 1547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.402842392326242e-05,
+      "loss": 0.4975,
+      "step": 1548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4004111753376416e-05,
+      "loss": 0.4053,
+      "step": 1549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3979809314705504e-05,
+      "loss": 0.4087,
+      "step": 1550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3955516646692084e-05,
+      "loss": 0.4754,
+      "step": 1551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3931233788762708e-05,
+      "loss": 0.3925,
+      "step": 1552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3906960780328001e-05,
+      "loss": 0.424,
+      "step": 1553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.388269766078261e-05,
+      "loss": 0.4208,
+      "step": 1554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3858444469505114e-05,
+      "loss": 0.4719,
+      "step": 1555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3834201245858001e-05,
+      "loss": 0.3697,
+      "step": 1556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.380996802918757e-05,
+      "loss": 0.3617,
+      "step": 1557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3785744858823883e-05,
+      "loss": 0.4599,
+      "step": 1558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3761531774080688e-05,
+      "loss": 0.4444,
+      "step": 1559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3737328814255384e-05,
+      "loss": 0.3256,
+      "step": 1560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3713136018628905e-05,
+      "loss": 0.4298,
+      "step": 1561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3688953426465726e-05,
+      "loss": 0.4301,
+      "step": 1562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3664781077013736e-05,
+      "loss": 0.4661,
+      "step": 1563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.3640619009504208e-05,
+      "loss": 0.5,
+      "step": 1564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3616467263151734e-05,
+      "loss": 0.4147,
+      "step": 1565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.359232587715415e-05,
+      "loss": 0.4755,
+      "step": 1566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3568194890692463e-05,
+      "loss": 0.3907,
+      "step": 1567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3544074342930826e-05,
+      "loss": 0.5497,
+      "step": 1568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3519964273016433e-05,
+      "loss": 0.3634,
+      "step": 1569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3495864720079489e-05,
+      "loss": 0.3871,
+      "step": 1570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.347177572323311e-05,
+      "loss": 0.3831,
+      "step": 1571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3447697321573299e-05,
+      "loss": 0.4758,
+      "step": 1572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3423629554178854e-05,
+      "loss": 0.4122,
+      "step": 1573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3399572460111304e-05,
+      "loss": 0.3779,
+      "step": 1574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3375526078414867e-05,
+      "loss": 0.4598,
+      "step": 1575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3351490448116376e-05,
+      "loss": 0.3688,
+      "step": 1576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3327465608225208e-05,
+      "loss": 0.3813,
+      "step": 1577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.330345159773324e-05,
+      "loss": 0.5386,
+      "step": 1578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3279448455614765e-05,
+      "loss": 0.4021,
+      "step": 1579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3255456220826428e-05,
+      "loss": 0.4061,
+      "step": 1580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3231474932307179e-05,
+      "loss": 0.4116,
+      "step": 1581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3207504628978207e-05,
+      "loss": 0.3852,
+      "step": 1582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3183545349742868e-05,
+      "loss": 0.4402,
+      "step": 1583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3159597133486628e-05,
+      "loss": 0.4701,
+      "step": 1584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3135660019076997e-05,
+      "loss": 0.419,
+      "step": 1585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3111734045363461e-05,
+      "loss": 0.354,
+      "step": 1586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3087819251177436e-05,
+      "loss": 0.5029,
+      "step": 1587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3063915675332184e-05,
+      "loss": 0.4279,
+      "step": 1588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3040023356622763e-05,
+      "loss": 0.3758,
+      "step": 1589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.301614233382597e-05,
+      "loss": 0.3895,
+      "step": 1590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2992272645700257e-05,
+      "loss": 0.4635,
+      "step": 1591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2968414330985677e-05,
+      "loss": 0.382,
+      "step": 1592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2944567428403838e-05,
+      "loss": 0.3727,
+      "step": 1593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2920731976657814e-05,
+      "loss": 0.3517,
+      "step": 1594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2896908014432109e-05,
+      "loss": 0.4522,
+      "step": 1595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2873095580392561e-05,
+      "loss": 0.3556,
+      "step": 1596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2849294713186315e-05,
+      "loss": 0.4149,
+      "step": 1597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2825505451441728e-05,
+      "loss": 0.3563,
+      "step": 1598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2801727833768335e-05,
+      "loss": 0.4478,
+      "step": 1599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2777961898756765e-05,
+      "loss": 0.4181,
+      "step": 1600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2754207684978688e-05,
+      "loss": 0.3918,
+      "step": 1601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.273046523098675e-05,
+      "loss": 0.3927,
+      "step": 1602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2706734575314523e-05,
+      "loss": 0.4673,
+      "step": 1603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2683015756476404e-05,
+      "loss": 0.355,
+      "step": 1604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.26593088129676e-05,
+      "loss": 0.3905,
+      "step": 1605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2635613783264042e-05,
+      "loss": 0.4196,
+      "step": 1606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2611930705822324e-05,
+      "loss": 0.3817,
+      "step": 1607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2588259619079634e-05,
+      "loss": 0.4316,
+      "step": 1608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2564600561453712e-05,
+      "loss": 0.3362,
+      "step": 1609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2540953571342771e-05,
+      "loss": 0.3478,
+      "step": 1610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.251731868712542e-05,
+      "loss": 0.4032,
+      "step": 1611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.249369594716065e-05,
+      "loss": 0.4411,
+      "step": 1612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.2470085389787725e-05,
+      "loss": 0.4149,
+      "step": 1613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.244648705332614e-05,
+      "loss": 0.4401,
+      "step": 1614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2422900976075558e-05,
+      "loss": 0.4054,
+      "step": 1615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2399327196315751e-05,
+      "loss": 0.4692,
+      "step": 1616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.237576575230651e-05,
+      "loss": 0.4984,
+      "step": 1617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2352216682287623e-05,
+      "loss": 0.5073,
+      "step": 1618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2328680024478794e-05,
+      "loss": 0.4625,
+      "step": 1619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2305155817079588e-05,
+      "loss": 0.4301,
+      "step": 1620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2281644098269352e-05,
+      "loss": 0.2723,
+      "step": 1621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2258144906207173e-05,
+      "loss": 0.4316,
+      "step": 1622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.223465827903179e-05,
+      "loss": 0.4612,
+      "step": 1623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2211184254861572e-05,
+      "loss": 0.4231,
+      "step": 1624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2187722871794416e-05,
+      "loss": 0.4684,
+      "step": 1625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.216427416790772e-05,
+      "loss": 0.3676,
+      "step": 1626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2140838181258281e-05,
+      "loss": 0.5033,
+      "step": 1627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2117414949882292e-05,
+      "loss": 0.4163,
+      "step": 1628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2094004511795197e-05,
+      "loss": 0.3559,
+      "step": 1629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2070606904991714e-05,
+      "loss": 0.4533,
+      "step": 1630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.204722216744572e-05,
+      "loss": 0.3857,
+      "step": 1631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2023850337110212e-05,
+      "loss": 0.4708,
+      "step": 1632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.2000491451917236e-05,
+      "loss": 0.3418,
+      "step": 1633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1977145549777826e-05,
+      "loss": 0.3195,
+      "step": 1634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1953812668581945e-05,
+      "loss": 0.3946,
+      "step": 1635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1930492846198432e-05,
+      "loss": 0.3925,
+      "step": 1636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1907186120474914e-05,
+      "loss": 0.4311,
+      "step": 1637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1883892529237784e-05,
+      "loss": 0.3757,
+      "step": 1638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1860612110292102e-05,
+      "loss": 0.4078,
+      "step": 1639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1837344901421561e-05,
+      "loss": 0.3239,
+      "step": 1640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1814090940388396e-05,
+      "loss": 0.3808,
+      "step": 1641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1790850264933362e-05,
+      "loss": 0.347,
+      "step": 1642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1767622912775642e-05,
+      "loss": 0.4259,
+      "step": 1643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1744408921612798e-05,
+      "loss": 0.3223,
+      "step": 1644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1721208329120699e-05,
+      "loss": 0.4737,
+      "step": 1645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1698021172953484e-05,
+      "loss": 0.4712,
+      "step": 1646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.167484749074347e-05,
+      "loss": 0.4365,
+      "step": 1647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1651687320101113e-05,
+      "loss": 0.4206,
+      "step": 1648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1628540698614943e-05,
+      "loss": 0.53,
+      "step": 1649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.160540766385149e-05,
+      "loss": 0.3143,
+      "step": 1650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1582288253355242e-05,
+      "loss": 0.4438,
+      "step": 1651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1559182504648573e-05,
+      "loss": 0.3881,
+      "step": 1652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1536090455231692e-05,
+      "loss": 0.3958,
+      "step": 1653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1513012142582544e-05,
+      "loss": 0.3757,
+      "step": 1654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1489947604156817e-05,
+      "loss": 0.4067,
+      "step": 1655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1466896877387826e-05,
+      "loss": 0.3774,
+      "step": 1656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1443859999686469e-05,
+      "loss": 0.4185,
+      "step": 1657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.142083700844117e-05,
+      "loss": 0.3111,
+      "step": 1658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1397827941017816e-05,
+      "loss": 0.3667,
+      "step": 1659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1374832834759695e-05,
+      "loss": 0.4254,
+      "step": 1660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1351851726987441e-05,
+      "loss": 0.3581,
+      "step": 1661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1328884654998959e-05,
+      "loss": 0.3417,
+      "step": 1662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1305931656069382e-05,
+      "loss": 0.408,
+      "step": 1663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.1282992767451e-05,
+      "loss": 0.4056,
+      "step": 1664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.126006802637321e-05,
+      "loss": 0.343,
+      "step": 1665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1237157470042429e-05,
+      "loss": 0.4427,
+      "step": 1666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.121426113564206e-05,
+      "loss": 0.4596,
+      "step": 1667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1191379060332441e-05,
+      "loss": 0.4495,
+      "step": 1668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1168511281250748e-05,
+      "loss": 0.3916,
+      "step": 1669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1145657835510961e-05,
+      "loss": 0.4093,
+      "step": 1670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.11228187602038e-05,
+      "loss": 0.4583,
+      "step": 1671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1099994092396655e-05,
+      "loss": 0.306,
+      "step": 1672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1077183869133543e-05,
+      "loss": 0.414,
+      "step": 1673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1054388127435027e-05,
+      "loss": 0.3654,
+      "step": 1674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1031606904298173e-05,
+      "loss": 0.4333,
+      "step": 1675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.1008840236696487e-05,
+      "loss": 0.343,
+      "step": 1676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.098608816157985e-05,
+      "loss": 0.4291,
+      "step": 1677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.096335071587444e-05,
+      "loss": 0.382,
+      "step": 1678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0940627936482717e-05,
+      "loss": 0.3142,
+      "step": 1679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0917919860283332e-05,
+      "loss": 0.4235,
+      "step": 1680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0895226524131069e-05,
+      "loss": 0.4585,
+      "step": 1681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.087254796485679e-05,
+      "loss": 0.4072,
+      "step": 1682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0849884219267384e-05,
+      "loss": 0.4355,
+      "step": 1683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0827235324145671e-05,
+      "loss": 0.2829,
+      "step": 1684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0804601316250394e-05,
+      "loss": 0.4624,
+      "step": 1685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0781982232316121e-05,
+      "loss": 0.3868,
+      "step": 1686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0759378109053217e-05,
+      "loss": 0.3702,
+      "step": 1687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0736788983147753e-05,
+      "loss": 0.3629,
+      "step": 1688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.071421489126145e-05,
+      "loss": 0.4579,
+      "step": 1689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0691655870031653e-05,
+      "loss": 0.3996,
+      "step": 1690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0669111956071218e-05,
+      "loss": 0.3754,
+      "step": 1691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0646583185968504e-05,
+      "loss": 0.403,
+      "step": 1692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0624069596287289e-05,
+      "loss": 0.3842,
+      "step": 1693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0601571223566707e-05,
+      "loss": 0.3925,
+      "step": 1694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.05790881043212e-05,
+      "loss": 0.403,
+      "step": 1695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0556620275040458e-05,
+      "loss": 0.39,
+      "step": 1696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0534167772189336e-05,
+      "loss": 0.3813,
+      "step": 1697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0511730632207838e-05,
+      "loss": 0.4396,
+      "step": 1698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0489308891511022e-05,
+      "loss": 0.4174,
+      "step": 1699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0466902586488954e-05,
+      "loss": 0.4257,
+      "step": 1700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0444511753506657e-05,
+      "loss": 0.3935,
+      "step": 1701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.042213642890403e-05,
+      "loss": 0.4131,
+      "step": 1702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0399776648995813e-05,
+      "loss": 0.4273,
+      "step": 1703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0377432450071511e-05,
+      "loss": 0.329,
+      "step": 1704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0355103868395344e-05,
+      "loss": 0.4391,
+      "step": 1705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0332790940206183e-05,
+      "loss": 0.4889,
+      "step": 1706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0310493701717496e-05,
+      "loss": 0.404,
+      "step": 1707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0288212189117299e-05,
+      "loss": 0.3949,
+      "step": 1708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0265946438568046e-05,
+      "loss": 0.3774,
+      "step": 1709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0243696486206653e-05,
+      "loss": 0.3335,
+      "step": 1710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0221462368144378e-05,
+      "loss": 0.3984,
+      "step": 1711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0199244120466778e-05,
+      "loss": 0.3871,
+      "step": 1712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0177041779233657e-05,
+      "loss": 0.4003,
+      "step": 1713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0154855380478999e-05,
+      "loss": 0.358,
+      "step": 1714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.013268496021092e-05,
+      "loss": 0.4138,
+      "step": 1715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0110530554411597e-05,
+      "loss": 0.386,
+      "step": 1716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0088392199037225e-05,
+      "loss": 0.4126,
+      "step": 1717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0066269930017936e-05,
+      "loss": 0.3544,
+      "step": 1718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0044163783257761e-05,
+      "loss": 0.381,
+      "step": 1719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0022073794634574e-05,
+      "loss": 0.425,
+      "step": 1720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.4111,
+      "step": 1721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.977942435179412e-06,
+      "loss": 0.3885,
+      "step": 1722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.955901135971819e-06,
+      "loss": 0.411,
+      "step": 1723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.933876138149843e-06,
+      "loss": 0.486,
+      "step": 1724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.911867477459642e-06,
+      "loss": 0.4497,
+      "step": 1725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.889875189620866e-06,
+      "loss": 0.3658,
+      "step": 1726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.86789931032659e-06,
+      "loss": 0.4438,
+      "step": 1727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.845939875243253e-06,
+      "loss": 0.3837,
+      "step": 1728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.823996920010614e-06,
+      "loss": 0.3993,
+      "step": 1729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.80207048024168e-06,
+      "loss": 0.4043,
+      "step": 1730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.780160591522654e-06,
+      "loss": 0.4169,
+      "step": 1731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.75826728941288e-06,
+      "loss": 0.3499,
+      "step": 1732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.736390609444783e-06,
+      "loss": 0.4144,
+      "step": 1733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.714530587123798e-06,
+      "loss": 0.3058,
+      "step": 1734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.69268725792834e-06,
+      "loss": 0.4805,
+      "step": 1735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.670860657309723e-06,
+      "loss": 0.2796,
+      "step": 1736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.649050820692119e-06,
+      "loss": 0.4348,
+      "step": 1737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.627257783472482e-06,
+      "loss": 0.4264,
+      "step": 1738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.605481581020517e-06,
+      "loss": 0.4131,
+      "step": 1739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.58372224867858e-06,
+      "loss": 0.4179,
+      "step": 1740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.561979821761669e-06,
+      "loss": 0.3719,
+      "step": 1741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.54025433555734e-06,
+      "loss": 0.3794,
+      "step": 1742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.518545825325644e-06,
+      "loss": 0.4605,
+      "step": 1743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.496854326299105e-06,
+      "loss": 0.3901,
+      "step": 1744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.475179873682621e-06,
+      "loss": 0.4091,
+      "step": 1745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.453522502653414e-06,
+      "loss": 0.3776,
+      "step": 1746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.431882248360998e-06,
+      "loss": 0.406,
+      "step": 1747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.410259145927104e-06,
+      "loss": 0.3557,
+      "step": 1748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.388653230445625e-06,
+      "loss": 0.4586,
+      "step": 1749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.367064536982565e-06,
+      "loss": 0.4573,
+      "step": 1750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.34549310057597e-06,
+      "loss": 0.382,
+      "step": 1751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.323938956235872e-06,
+      "loss": 0.4113,
+      "step": 1752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.30240213894425e-06,
+      "loss": 0.3905,
+      "step": 1753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.280882683654964e-06,
+      "loss": 0.4391,
+      "step": 1754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.259380625293681e-06,
+      "loss": 0.3967,
+      "step": 1755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.237895998757849e-06,
+      "loss": 0.4051,
+      "step": 1756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.216428838916612e-06,
+      "loss": 0.3513,
+      "step": 1757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.194979180610772e-06,
+      "loss": 0.2781,
+      "step": 1758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.17354705865273e-06,
+      "loss": 0.317,
+      "step": 1759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.152132507826417e-06,
+      "loss": 0.4057,
+      "step": 1760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.130735562887249e-06,
+      "loss": 0.4111,
+      "step": 1761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.109356258562077e-06,
+      "loss": 0.3277,
+      "step": 1762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.087994629549107e-06,
+      "loss": 0.3904,
+      "step": 1763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.066650710517874e-06,
+      "loss": 0.366,
+      "step": 1764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.045324536109148e-06,
+      "loss": 0.3589,
+      "step": 1765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.024016140934923e-06,
+      "loss": 0.4564,
+      "step": 1766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.002725559578325e-06,
+      "loss": 0.4008,
+      "step": 1767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.981452826593574e-06,
+      "loss": 0.3269,
+      "step": 1768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.960197976505915e-06,
+      "loss": 0.4248,
+      "step": 1769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.938961043811578e-06,
+      "loss": 0.4431,
+      "step": 1770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.91774206297771e-06,
+      "loss": 0.4064,
+      "step": 1771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.896541068442323e-06,
+      "loss": 0.4012,
+      "step": 1772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.875358094614232e-06,
+      "loss": 0.3846,
+      "step": 1773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.854193175873017e-06,
+      "loss": 0.3008,
+      "step": 1774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.83304634656894e-06,
+      "loss": 0.3448,
+      "step": 1775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.811917641022923e-06,
+      "loss": 0.4262,
+      "step": 1776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.790807093526445e-06,
+      "loss": 0.386,
+      "step": 1777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.769714738341544e-06,
+      "loss": 0.3435,
+      "step": 1778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.748640609700713e-06,
+      "loss": 0.454,
+      "step": 1779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.727584741806877e-06,
+      "loss": 0.3375,
+      "step": 1780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.706547168833315e-06,
+      "loss": 0.4246,
+      "step": 1781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.685527924923623e-06,
+      "loss": 0.4241,
+      "step": 1782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.66452704419163e-06,
+      "loss": 0.3623,
+      "step": 1783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.643544560721383e-06,
+      "loss": 0.4026,
+      "step": 1784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.622580508567052e-06,
+      "loss": 0.4348,
+      "step": 1785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.601634921752919e-06,
+      "loss": 0.3898,
+      "step": 1786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.580707834273278e-06,
+      "loss": 0.3948,
+      "step": 1787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.559799280092408e-06,
+      "loss": 0.3988,
+      "step": 1788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.538909293144485e-06,
+      "loss": 0.3846,
+      "step": 1789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.51803790733358e-06,
+      "loss": 0.3737,
+      "step": 1790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.497185156533567e-06,
+      "loss": 0.3877,
+      "step": 1791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.476351074588068e-06,
+      "loss": 0.3992,
+      "step": 1792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.45553569531041e-06,
+      "loss": 0.3967,
+      "step": 1793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.434739052483578e-06,
+      "loss": 0.3895,
+      "step": 1794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.413961179860117e-06,
+      "loss": 0.502,
+      "step": 1795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.393202111162139e-06,
+      "loss": 0.3607,
+      "step": 1796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.372461880081227e-06,
+      "loss": 0.3391,
+      "step": 1797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.351740520278385e-06,
+      "loss": 0.3831,
+      "step": 1798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.331038065383995e-06,
+      "loss": 0.4829,
+      "step": 1799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.310354548997764e-06,
+      "loss": 0.3556,
+      "step": 1800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.289690004688646e-06,
+      "loss": 0.4029,
+      "step": 1801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.269044465994815e-06,
+      "loss": 0.4931,
+      "step": 1802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.248417966423598e-06,
+      "loss": 0.4039,
+      "step": 1803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.227810539451418e-06,
+      "loss": 0.3711,
+      "step": 1804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.207222218523745e-06,
+      "loss": 0.4409,
+      "step": 1805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.186653037055044e-06,
+      "loss": 0.4205,
+      "step": 1806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.166103028428716e-06,
+      "loss": 0.3714,
+      "step": 1807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.145572225997032e-06,
+      "loss": 0.4124,
+      "step": 1808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.12506066308111e-06,
+      "loss": 0.4499,
+      "step": 1809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.104568372970834e-06,
+      "loss": 0.4697,
+      "step": 1810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.08409538892481e-06,
+      "loss": 0.3775,
+      "step": 1811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.063641744170307e-06,
+      "loss": 0.341,
+      "step": 1812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.043207471903216e-06,
+      "loss": 0.4364,
+      "step": 1813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.022792605287977e-06,
+      "loss": 0.3849,
+      "step": 1814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.00239717745754e-06,
+      "loss": 0.3726,
+      "step": 1815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.982021221513303e-06,
+      "loss": 0.3269,
+      "step": 1816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.961664770525064e-06,
+      "loss": 0.3634,
+      "step": 1817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.941327857530967e-06,
+      "loss": 0.4352,
+      "step": 1818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.921010515537449e-06,
+      "loss": 0.4432,
+      "step": 1819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.900712777519162e-06,
+      "loss": 0.3765,
+      "step": 1820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.880434676418966e-06,
+      "loss": 0.407,
+      "step": 1821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.860176245147844e-06,
+      "loss": 0.3808,
+      "step": 1822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.83993751658485e-06,
+      "loss": 0.4375,
+      "step": 1823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.819718523577065e-06,
+      "loss": 0.3907,
+      "step": 1824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.799519298939544e-06,
+      "loss": 0.4413,
+      "step": 1825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.779339875455241e-06,
+      "loss": 0.4408,
+      "step": 1826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.759180285874986e-06,
+      "loss": 0.4012,
+      "step": 1827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.739040562917428e-06,
+      "loss": 0.3566,
+      "step": 1828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.718920739268958e-06,
+      "loss": 0.3039,
+      "step": 1829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.698820847583674e-06,
+      "loss": 0.3686,
+      "step": 1830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.678740920483339e-06,
+      "loss": 0.35,
+      "step": 1831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.65868099055728e-06,
+      "loss": 0.4444,
+      "step": 1832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.6386410903624e-06,
+      "loss": 0.3645,
+      "step": 1833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.618621252423084e-06,
+      "loss": 0.4234,
+      "step": 1834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5986215092311565e-06,
+      "loss": 0.3929,
+      "step": 1835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5786418932458284e-06,
+      "loss": 0.4356,
+      "step": 1836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.558682436893647e-06,
+      "loss": 0.4573,
+      "step": 1837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.538743172568444e-06,
+      "loss": 0.3411,
+      "step": 1838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.5188241326312596e-06,
+      "loss": 0.4218,
+      "step": 1839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.4989253494103306e-06,
+      "loss": 0.476,
+      "step": 1840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.479046855201018e-06,
+      "loss": 0.443,
+      "step": 1841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.459188682265743e-06,
+      "loss": 0.4385,
+      "step": 1842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.439350862833945e-06,
+      "loss": 0.3918,
+      "step": 1843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.419533429102057e-06,
+      "loss": 0.4142,
+      "step": 1844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.399736413233387e-06,
+      "loss": 0.4378,
+      "step": 1845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.379959847358125e-06,
+      "loss": 0.3703,
+      "step": 1846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.360203763573275e-06,
+      "loss": 0.4433,
+      "step": 1847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.340468193942589e-06,
+      "loss": 0.412,
+      "step": 1848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.32075317049653e-06,
+      "loss": 0.3552,
+      "step": 1849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.301058725232226e-06,
+      "loss": 0.4991,
+      "step": 1850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.2813848901133765e-06,
+      "loss": 0.338,
+      "step": 1851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.26173169707026e-06,
+      "loss": 0.4318,
+      "step": 1852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.24209917799964e-06,
+      "loss": 0.4194,
+      "step": 1853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.222487364764732e-06,
+      "loss": 0.4137,
+      "step": 1854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.202896289195145e-06,
+      "loss": 0.4071,
+      "step": 1855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.183325983086826e-06,
+      "loss": 0.3616,
+      "step": 1856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.163776478202023e-06,
+      "loss": 0.3763,
+      "step": 1857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1442478062692135e-06,
+      "loss": 0.3678,
+      "step": 1858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.1247399989830726e-06,
+      "loss": 0.3381,
+      "step": 1859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.105253088004409e-06,
+      "loss": 0.4343,
+      "step": 1860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.085787104960113e-06,
+      "loss": 0.3489,
+      "step": 1861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.066342081443125e-06,
+      "loss": 0.3743,
+      "step": 1862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.046918049012337e-06,
+      "loss": 0.3574,
+      "step": 1863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.027515039192603e-06,
+      "loss": 0.3901,
+      "step": 1864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.0081330834746465e-06,
+      "loss": 0.4112,
+      "step": 1865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.988772213315021e-06,
+      "loss": 0.3891,
+      "step": 1866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.9694324601360565e-06,
+      "loss": 0.4226,
+      "step": 1867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.950113855325824e-06,
+      "loss": 0.3626,
+      "step": 1868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.930816430238034e-06,
+      "loss": 0.4414,
+      "step": 1869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.911540216192072e-06,
+      "loss": 0.3781,
+      "step": 1870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8922852444728675e-06,
+      "loss": 0.394,
+      "step": 1871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.873051546330882e-06,
+      "loss": 0.393,
+      "step": 1872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.853839152982047e-06,
+      "loss": 0.3137,
+      "step": 1873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8346480956077235e-06,
+      "loss": 0.4008,
+      "step": 1874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.815478405354646e-06,
+      "loss": 0.4168,
+      "step": 1875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.796330113334851e-06,
+      "loss": 0.3765,
+      "step": 1876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.777203250625666e-06,
+      "loss": 0.3887,
+      "step": 1877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.758097848269634e-06,
+      "loss": 0.4501,
+      "step": 1878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.73901393727447e-06,
+      "loss": 0.3857,
+      "step": 1879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.719951548613004e-06,
+      "loss": 0.4172,
+      "step": 1880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.700910713223146e-06,
+      "loss": 0.3836,
+      "step": 1881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.681891462007801e-06,
+      "loss": 0.3612,
+      "step": 1882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.662893825834873e-06,
+      "loss": 0.3488,
+      "step": 1883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6439178355371684e-06,
+      "loss": 0.3422,
+      "step": 1884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.624963521912357e-06,
+      "loss": 0.3451,
+      "step": 1885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6060309157229566e-06,
+      "loss": 0.4107,
+      "step": 1886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.587120047696232e-06,
+      "loss": 0.3638,
+      "step": 1887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.568230948524154e-06,
+      "loss": 0.4352,
+      "step": 1888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5493636488633915e-06,
+      "loss": 0.398,
+      "step": 1889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.530518179335217e-06,
+      "loss": 0.3162,
+      "step": 1890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5116945705254775e-06,
+      "loss": 0.2733,
+      "step": 1891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.49289285298454e-06,
+      "loss": 0.4471,
+      "step": 1892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.4741130572272494e-06,
+      "loss": 0.4216,
+      "step": 1893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.455355213732852e-06,
+      "loss": 0.3833,
+      "step": 1894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.436619352944984e-06,
+      "loss": 0.4171,
+      "step": 1895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.417905505271602e-06,
+      "loss": 0.3571,
+      "step": 1896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3992137010849296e-06,
+      "loss": 0.4755,
+      "step": 1897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.380543970721418e-06,
+      "loss": 0.4162,
+      "step": 1898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.361896344481693e-06,
+      "loss": 0.4304,
+      "step": 1899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.343270852630503e-06,
+      "loss": 0.3812,
+      "step": 1900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.3246675253966795e-06,
+      "loss": 0.3174,
+      "step": 1901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.306086392973072e-06,
+      "loss": 0.4537,
+      "step": 1902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.287527485516514e-06,
+      "loss": 0.4096,
+      "step": 1903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.268990833147768e-06,
+      "loss": 0.4567,
+      "step": 1904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.250476465951481e-06,
+      "loss": 0.351,
+      "step": 1905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.231984413976116e-06,
+      "loss": 0.4298,
+      "step": 1906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.213514707233932e-06,
+      "loss": 0.3394,
+      "step": 1907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.195067375700922e-06,
+      "loss": 0.401,
+      "step": 1908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.176642449316763e-06,
+      "loss": 0.4134,
+      "step": 1909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.158239957984766e-06,
+      "loss": 0.4468,
+      "step": 1910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.139859931571832e-06,
+      "loss": 0.3542,
+      "step": 1911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.1215023999084035e-06,
+      "loss": 0.3023,
+      "step": 1912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.103167392788411e-06,
+      "loss": 0.3864,
+      "step": 1913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.084854939969231e-06,
+      "loss": 0.3878,
+      "step": 1914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.066565071171635e-06,
+      "loss": 0.4287,
+      "step": 1915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.048297816079738e-06,
+      "loss": 0.3511,
+      "step": 1916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0300532043409555e-06,
+      "loss": 0.3515,
+      "step": 1917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0118312655659595e-06,
+      "loss": 0.3421,
+      "step": 1918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9936320293286044e-06,
+      "loss": 0.3829,
+      "step": 1919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.97545552516592e-06,
+      "loss": 0.3836,
+      "step": 1920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9573017825780335e-06,
+      "loss": 0.4459,
+      "step": 1921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9391708310281295e-06,
+      "loss": 0.4609,
+      "step": 1922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.921062699942408e-06,
+      "loss": 0.5153,
+      "step": 1923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9029774187100365e-06,
+      "loss": 0.3073,
+      "step": 1924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.884915016683072e-06,
+      "loss": 0.4477,
+      "step": 1925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.866875523176467e-06,
+      "loss": 0.3914,
+      "step": 1926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.848858967467979e-06,
+      "loss": 0.3814,
+      "step": 1927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8308653787981514e-06,
+      "loss": 0.4096,
+      "step": 1928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.81289478637024e-06,
+      "loss": 0.3252,
+      "step": 1929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.794947219350187e-06,
+      "loss": 0.3389,
+      "step": 1930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7770227068665485e-06,
+      "loss": 0.3632,
+      "step": 1931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.759121278010482e-06,
+      "loss": 0.4486,
+      "step": 1932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.741242961835671e-06,
+      "loss": 0.442,
+      "step": 1933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.723387787358292e-06,
+      "loss": 0.4245,
+      "step": 1934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.70555578355696e-06,
+      "loss": 0.4175,
+      "step": 1935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.687746979372693e-06,
+      "loss": 0.3367,
+      "step": 1936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.669961403708834e-06,
+      "loss": 0.461,
+      "step": 1937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.652199085431049e-06,
+      "loss": 0.3789,
+      "step": 1938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.634460053367252e-06,
+      "loss": 0.4447,
+      "step": 1939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.61674433630756e-06,
+      "loss": 0.3856,
+      "step": 1940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.599051963004252e-06,
+      "loss": 0.4049,
+      "step": 1941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.581382962171722e-06,
+      "loss": 0.4246,
+      "step": 1942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.563737362486428e-06,
+      "loss": 0.3747,
+      "step": 1943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.546115192586852e-06,
+      "loss": 0.3892,
+      "step": 1944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.528516481073445e-06,
+      "loss": 0.3504,
+      "step": 1945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.510941256508593e-06,
+      "loss": 0.3228,
+      "step": 1946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.493389547416555e-06,
+      "loss": 0.4045,
+      "step": 1947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.475861382283429e-06,
+      "loss": 0.413,
+      "step": 1948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.458356789557109e-06,
+      "loss": 0.4332,
+      "step": 1949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.440875797647207e-06,
+      "loss": 0.3761,
+      "step": 1950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4234184349250584e-06,
+      "loss": 0.3977,
+      "step": 1951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.405984729723637e-06,
+      "loss": 0.4051,
+      "step": 1952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3885747103375176e-06,
+      "loss": 0.4248,
+      "step": 1953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.371188405022845e-06,
+      "loss": 0.3838,
+      "step": 1954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.353825841997265e-06,
+      "loss": 0.4217,
+      "step": 1955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3364870494398934e-06,
+      "loss": 0.3765,
+      "step": 1956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.31917205549127e-06,
+      "loss": 0.325,
+      "step": 1957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.301880888253308e-06,
+      "loss": 0.3146,
+      "step": 1958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.28461357578925e-06,
+      "loss": 0.4158,
+      "step": 1959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2673701461236224e-06,
+      "loss": 0.3849,
+      "step": 1960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.250150627242196e-06,
+      "loss": 0.3678,
+      "step": 1961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.232955047091921e-06,
+      "loss": 0.4239,
+      "step": 1962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.215783433580906e-06,
+      "loss": 0.3347,
+      "step": 1963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.198635814578361e-06,
+      "loss": 0.3867,
+      "step": 1964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.181512217914557e-06,
+      "loss": 0.3519,
+      "step": 1965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1644126713807675e-06,
+      "loss": 0.4967,
+      "step": 1966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.147337202729245e-06,
+      "loss": 0.3503,
+      "step": 1967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1302858396731504e-06,
+      "loss": 0.3473,
+      "step": 1968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.113258609886524e-06,
+      "loss": 0.3683,
+      "step": 1969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.096255541004256e-06,
+      "loss": 0.4393,
+      "step": 1970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.079276660622005e-06,
+      "loss": 0.4179,
+      "step": 1971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.06232199629618e-06,
+      "loss": 0.3481,
+      "step": 1972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.045391575543888e-06,
+      "loss": 0.4134,
+      "step": 1973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.028485425842878e-06,
+      "loss": 0.3831,
+      "step": 1974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.011603574631521e-06,
+      "loss": 0.4583,
+      "step": 1975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.994746049308747e-06,
+      "loss": 0.322,
+      "step": 1976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.97791287723401e-06,
+      "loss": 0.2951,
+      "step": 1977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9611040857272304e-06,
+      "loss": 0.4738,
+      "step": 1978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.944319702068771e-06,
+      "loss": 0.3419,
+      "step": 1979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.927559753499365e-06,
+      "loss": 0.3834,
+      "step": 1980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9108242672201e-06,
+      "loss": 0.3969,
+      "step": 1981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8941132703923596e-06,
+      "loss": 0.3847,
+      "step": 1982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.877426790137783e-06,
+      "loss": 0.4065,
+      "step": 1983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.860764853538215e-06,
+      "loss": 0.3885,
+      "step": 1984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.844127487635664e-06,
+      "loss": 0.4619,
+      "step": 1985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.827514719432282e-06,
+      "loss": 0.4315,
+      "step": 1986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.810926575890267e-06,
+      "loss": 0.4172,
+      "step": 1987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79436308393187e-06,
+      "loss": 0.3876,
+      "step": 1988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.777824270439333e-06,
+      "loss": 0.3342,
+      "step": 1989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.761310162254842e-06,
+      "loss": 0.3518,
+      "step": 1990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.744820786180486e-06,
+      "loss": 0.4185,
+      "step": 1991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.728356168978225e-06,
+      "loss": 0.3751,
+      "step": 1992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.711916337369813e-06,
+      "loss": 0.3829,
+      "step": 1993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.695501318036794e-06,
+      "loss": 0.3233,
+      "step": 1994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.3841,
+      "step": 1995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.662745822721715e-06,
+      "loss": 0.449,
+      "step": 1996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6464053999012124e-06,
+      "loss": 0.3484,
+      "step": 1997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.630089895679139e-06,
+      "loss": 0.4135,
+      "step": 1998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.613799336535254e-06,
+      "loss": 0.3315,
+      "step": 1999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.597533748908831e-06,
+      "loss": 0.3802,
+      "step": 2000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5812931591986165e-06,
+      "loss": 0.3977,
+      "step": 2001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.565077593762785e-06,
+      "loss": 0.3354,
+      "step": 2002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5488870789189e-06,
+      "loss": 0.3562,
+      "step": 2003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.532721640943869e-06,
+      "loss": 0.401,
+      "step": 2004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5165813060738865e-06,
+      "loss": 0.3432,
+      "step": 2005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.500466100504419e-06,
+      "loss": 0.3312,
+      "step": 2006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.484376050390151e-06,
+      "loss": 0.4208,
+      "step": 2007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.468311181844929e-06,
+      "loss": 0.4179,
+      "step": 2008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.45227152094174e-06,
+      "loss": 0.3712,
+      "step": 2009
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.436257093712659e-06,
+      "loss": 0.3675,
+      "step": 2010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4202679261487845e-06,
+      "loss": 0.4104,
+      "step": 2011
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.404304044200261e-06,
+      "loss": 0.3122,
+      "step": 2012
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.38836547377616e-06,
+      "loss": 0.4605,
+      "step": 2013
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.372452240744485e-06,
+      "loss": 0.3835,
+      "step": 2014
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.356564370932122e-06,
+      "loss": 0.426,
+      "step": 2015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.34070189012479e-06,
+      "loss": 0.4013,
+      "step": 2016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.324864824066992e-06,
+      "loss": 0.4573,
+      "step": 2017
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3090531984619944e-06,
+      "loss": 0.366,
+      "step": 2018
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.293267038971771e-06,
+      "loss": 0.4087,
+      "step": 2019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.277506371216969e-06,
+      "loss": 0.3359,
+      "step": 2020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.261771220776856e-06,
+      "loss": 0.3826,
+      "step": 2021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.246061613189288e-06,
+      "loss": 0.489,
+      "step": 2022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2303775739506726e-06,
+      "loss": 0.5034,
+      "step": 2023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2147191285159e-06,
+      "loss": 0.2808,
+      "step": 2024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.199086302298348e-06,
+      "loss": 0.4124,
+      "step": 2025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.183479120669796e-06,
+      "loss": 0.3581,
+      "step": 2026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.167897608960408e-06,
+      "loss": 0.3851,
+      "step": 2027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.152341792458696e-06,
+      "loss": 0.3932,
+      "step": 2028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.136811696411462e-06,
+      "loss": 0.4319,
+      "step": 2029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.121307346023753e-06,
+      "loss": 0.4431,
+      "step": 2030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.105828766458844e-06,
+      "loss": 0.367,
+      "step": 2031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.090375982838184e-06,
+      "loss": 0.405,
+      "step": 2032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.074949020241352e-06,
+      "loss": 0.3654,
+      "step": 2033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0595479037060245e-06,
+      "loss": 0.4464,
+      "step": 2034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.04417265822793e-06,
+      "loss": 0.2841,
+      "step": 2035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.028823308760792e-06,
+      "loss": 0.2746,
+      "step": 2036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.013499880216329e-06,
+      "loss": 0.3726,
+      "step": 2037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.998202397464182e-06,
+      "loss": 0.3396,
+      "step": 2038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.982930885331875e-06,
+      "loss": 0.4503,
+      "step": 2039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.967685368604795e-06,
+      "loss": 0.3186,
+      "step": 2040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.952465872026126e-06,
+      "loss": 0.37,
+      "step": 2041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.937272420296829e-06,
+      "loss": 0.3776,
+      "step": 2042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.922105038075597e-06,
+      "loss": 0.423,
+      "step": 2043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.906963749978805e-06,
+      "loss": 0.3897,
+      "step": 2044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.891848580580484e-06,
+      "loss": 0.4538,
+      "step": 2045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.876759554412275e-06,
+      "loss": 0.4677,
+      "step": 2046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.861696695963388e-06,
+      "loss": 0.3259,
+      "step": 2047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.846660029680553e-06,
+      "loss": 0.4163,
+      "step": 2048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.831649579968004e-06,
+      "loss": 0.4582,
+      "step": 2049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.816665371187425e-06,
+      "loss": 0.429,
+      "step": 2050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8017074276579056e-06,
+      "loss": 0.4497,
+      "step": 2051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7867757736559105e-06,
+      "loss": 0.3982,
+      "step": 2052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7718704334152347e-06,
+      "loss": 0.2972,
+      "step": 2053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.756991431126971e-06,
+      "loss": 0.3987,
+      "step": 2054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7421387909394604e-06,
+      "loss": 0.3406,
+      "step": 2055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7273125369582653e-06,
+      "loss": 0.3844,
+      "step": 2056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7125126932461154e-06,
+      "loss": 0.3843,
+      "step": 2057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6977392838228854e-06,
+      "loss": 0.4928,
+      "step": 2058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6829923326655424e-06,
+      "loss": 0.3901,
+      "step": 2059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6682718637081214e-06,
+      "loss": 0.4172,
+      "step": 2060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653577900841656e-06,
+      "loss": 0.3781,
+      "step": 2061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6389104679141806e-06,
+      "loss": 0.3392,
+      "step": 2062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6242695887306667e-06,
+      "loss": 0.3836,
+      "step": 2063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6096552870529887e-06,
+      "loss": 0.4145,
+      "step": 2064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5950675865998874e-06,
+      "loss": 0.3916,
+      "step": 2065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5805065110469306e-06,
+      "loss": 0.3733,
+      "step": 2066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.565972084026468e-06,
+      "loss": 0.4072,
+      "step": 2067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5514643291276075e-06,
+      "loss": 0.3712,
+      "step": 2068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.536983269896159e-06,
+      "loss": 0.4648,
+      "step": 2069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5225289298346234e-06,
+      "loss": 0.364,
+      "step": 2070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5081013324021184e-06,
+      "loss": 0.267,
+      "step": 2071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4937005010143764e-06,
+      "loss": 0.3945,
+      "step": 2072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4793264590436635e-06,
+      "loss": 0.3609,
+      "step": 2073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4649792298187922e-06,
+      "loss": 0.3858,
+      "step": 2074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4506588366250447e-06,
+      "loss": 0.3538,
+      "step": 2075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4363653027041545e-06,
+      "loss": 0.4213,
+      "step": 2076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.42209865125426e-06,
+      "loss": 0.3188,
+      "step": 2077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.407858905429877e-06,
+      "loss": 0.4422,
+      "step": 2078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.39364608834184e-06,
+      "loss": 0.36,
+      "step": 2079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3794602230572893e-06,
+      "loss": 0.4263,
+      "step": 2080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3653013325996197e-06,
+      "loss": 0.4058,
+      "step": 2081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3511694399484516e-06,
+      "loss": 0.3698,
+      "step": 2082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3370645680395785e-06,
+      "loss": 0.4081,
+      "step": 2083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.322986739764946e-06,
+      "loss": 0.4201,
+      "step": 2084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3089359779726117e-06,
+      "loss": 0.4256,
+      "step": 2085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2949123054666953e-06,
+      "loss": 0.3425,
+      "step": 2086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2809157450073603e-06,
+      "loss": 0.4555,
+      "step": 2087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.266946319310762e-06,
+      "loss": 0.2999,
+      "step": 2088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.253004051049018e-06,
+      "loss": 0.3263,
+      "step": 2089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2390889628501763e-06,
+      "loss": 0.3774,
+      "step": 2090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.225201077298159e-06,
+      "loss": 0.4321,
+      "step": 2091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2113404169327444e-06,
+      "loss": 0.4087,
+      "step": 2092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1975070042495337e-06,
+      "loss": 0.4463,
+      "step": 2093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.183700861699892e-06,
+      "loss": 0.4551,
+      "step": 2094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1699220116909356e-06,
+      "loss": 0.4408,
+      "step": 2095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1561704765854806e-06,
+      "loss": 0.3744,
+      "step": 2096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.142446278702014e-06,
+      "loss": 0.2488,
+      "step": 2097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1287494403146532e-06,
+      "loss": 0.3529,
+      "step": 2098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1150799836531085e-06,
+      "loss": 0.3423,
+      "step": 2099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.10143793090266e-06,
+      "loss": 0.4029,
+      "step": 2100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0878233042041008e-06,
+      "loss": 0.331,
+      "step": 2101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.074236125653718e-06,
+      "loss": 0.4459,
+      "step": 2102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0606764173032544e-06,
+      "loss": 0.3869,
+      "step": 2103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0471442011598573e-06,
+      "loss": 0.3848,
+      "step": 2104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.033639499186063e-06,
+      "loss": 0.3959,
+      "step": 2105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.020162333299752e-06,
+      "loss": 0.3974,
+      "step": 2106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.006712725374117e-06,
+      "loss": 0.4191,
+      "step": 2107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9932906972376184e-06,
+      "loss": 0.372,
+      "step": 2108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9798962706739632e-06,
+      "loss": 0.4412,
+      "step": 2109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.966529467422048e-06,
+      "loss": 0.376,
+      "step": 2110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9531903091759464e-06,
+      "loss": 0.3406,
+      "step": 2111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9398788175848713e-06,
+      "loss": 0.459,
+      "step": 2112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.926595014253124e-06,
+      "loss": 0.3541,
+      "step": 2113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9133389207400674e-06,
+      "loss": 0.3432,
+      "step": 2114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.900110558560103e-06,
+      "loss": 0.3666,
+      "step": 2115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8869099491826036e-06,
+      "loss": 0.3402,
+      "step": 2116
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.873737114031918e-06,
+      "loss": 0.348,
+      "step": 2117
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8605920744873138e-06,
+      "loss": 0.3567,
+      "step": 2118
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8474748518829452e-06,
+      "loss": 0.4263,
+      "step": 2119
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8343854675078206e-06,
+      "loss": 0.3671,
+      "step": 2120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8213239426057715e-06,
+      "loss": 0.3359,
+      "step": 2121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8082902983754025e-06,
+      "loss": 0.3688,
+      "step": 2122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.795284555970077e-06,
+      "loss": 0.3723,
+      "step": 2123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7823067364978753e-06,
+      "loss": 0.3615,
+      "step": 2124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7693568610215572e-06,
+      "loss": 0.4027,
+      "step": 2125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7564349505585262e-06,
+      "loss": 0.3689,
+      "step": 2126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.743541026080809e-06,
+      "loss": 0.3704,
+      "step": 2127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7306751085149997e-06,
+      "loss": 0.3241,
+      "step": 2128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7178372187422454e-06,
+      "loss": 0.3329,
+      "step": 2129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.705027377598202e-06,
+      "loss": 0.3021,
+      "step": 2130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.692245605873003e-06,
+      "loss": 0.425,
+      "step": 2131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.679491924311226e-06,
+      "loss": 0.4011,
+      "step": 2132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6667663536118604e-06,
+      "loss": 0.4034,
+      "step": 2133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.654068914428276e-06,
+      "loss": 0.444,
+      "step": 2134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6413996273681665e-06,
+      "loss": 0.3547,
+      "step": 2135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.628758512993559e-06,
+      "loss": 0.407,
+      "step": 2136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6161455918207446e-06,
+      "loss": 0.3892,
+      "step": 2137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6035608843202597e-06,
+      "loss": 0.3433,
+      "step": 2138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5910044109168487e-06,
+      "loss": 0.3879,
+      "step": 2139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5784761919894383e-06,
+      "loss": 0.3625,
+      "step": 2140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5659762478710938e-06,
+      "loss": 0.3967,
+      "step": 2141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.55350459884899e-06,
+      "loss": 0.341,
+      "step": 2142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5410612651643817e-06,
+      "loss": 0.3294,
+      "step": 2143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5286462670125687e-06,
+      "loss": 0.3835,
+      "step": 2144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5162596245428626e-06,
+      "loss": 0.5468,
+      "step": 2145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.503901357858558e-06,
+      "loss": 0.3401,
+      "step": 2146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.491571487016882e-06,
+      "loss": 0.3789,
+      "step": 2147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4792700320289887e-06,
+      "loss": 0.3393,
+      "step": 2148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.466997012859915e-06,
+      "loss": 0.4125,
+      "step": 2149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.454752449428539e-06,
+      "loss": 0.3344,
+      "step": 2150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4425363616075615e-06,
+      "loss": 0.3309,
+      "step": 2151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.430348769223463e-06,
+      "loss": 0.3639,
+      "step": 2152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.418189692056483e-06,
+      "loss": 0.4542,
+      "step": 2153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.406059149840574e-06,
+      "loss": 0.3454,
+      "step": 2154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.393957162263383e-06,
+      "loss": 0.3785,
+      "step": 2155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3818837489662096e-06,
+      "loss": 0.3633,
+      "step": 2156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3698389295439796e-06,
+      "loss": 0.3877,
+      "step": 2157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3578227235452157e-06,
+      "loss": 0.3992,
+      "step": 2158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3458351504719844e-06,
+      "loss": 0.3639,
+      "step": 2159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3338762297799033e-06,
+      "loss": 0.4452,
+      "step": 2160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3219459808780755e-06,
+      "loss": 0.3541,
+      "step": 2161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.310044423129072e-06,
+      "loss": 0.3132,
+      "step": 2162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.298171575848902e-06,
+      "loss": 0.354,
+      "step": 2163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.286327458306978e-06,
+      "loss": 0.3944,
+      "step": 2164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2745120897260755e-06,
+      "loss": 0.3001,
+      "step": 2165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2627254892823224e-06,
+      "loss": 0.3444,
+      "step": 2166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.250967676105149e-06,
+      "loss": 0.3926,
+      "step": 2167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.239238669277266e-06,
+      "loss": 0.3894,
+      "step": 2168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.227538487834642e-06,
+      "loss": 0.369,
+      "step": 2169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2158671507664487e-06,
+      "loss": 0.4216,
+      "step": 2170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2042246770150523e-06,
+      "loss": 0.3816,
+      "step": 2171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.192611085475964e-06,
+      "loss": 0.3588,
+      "step": 2172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1810263949978317e-06,
+      "loss": 0.3038,
+      "step": 2173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1694706243823904e-06,
+      "loss": 0.3862,
+      "step": 2174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1579437923844405e-06,
+      "loss": 0.3428,
+      "step": 2175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1464459177118193e-06,
+      "loss": 0.3591,
+      "step": 2176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.134977019025366e-06,
+      "loss": 0.4029,
+      "step": 2177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1235371149388808e-06,
+      "loss": 0.3849,
+      "step": 2178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1121262240191174e-06,
+      "loss": 0.3697,
+      "step": 2179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.100744364785745e-06,
+      "loss": 0.3791,
+      "step": 2180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.089391555711304e-06,
+      "loss": 0.4075,
+      "step": 2181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.078067815221196e-06,
+      "loss": 0.4094,
+      "step": 2182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0667731616936406e-06,
+      "loss": 0.4423,
+      "step": 2183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0555076134596484e-06,
+      "loss": 0.3593,
+      "step": 2184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0442711888029976e-06,
+      "loss": 0.4039,
+      "step": 2185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.033063905960193e-06,
+      "loss": 0.3435,
+      "step": 2186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0218857831204475e-06,
+      "loss": 0.3548,
+      "step": 2187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0107368384256485e-06,
+      "loss": 0.3277,
+      "step": 2188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999617089970327e-06,
+      "loss": 0.3952,
+      "step": 2189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9885265558016197e-06,
+      "loss": 0.3358,
+      "step": 2190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.977465253919264e-06,
+      "loss": 0.468,
+      "step": 2191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.966433202275546e-06,
+      "loss": 0.4024,
+      "step": 2192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9554304187752795e-06,
+      "loss": 0.4302,
+      "step": 2193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.944456921275779e-06,
+      "loss": 0.373,
+      "step": 2194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.933512727586826e-06,
+      "loss": 0.4008,
+      "step": 2195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9225978554706403e-06,
+      "loss": 0.4427,
+      "step": 2196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.911712322641861e-06,
+      "loss": 0.3069,
+      "step": 2197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9008561467675046e-06,
+      "loss": 0.4112,
+      "step": 2198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8900293454669394e-06,
+      "loss": 0.3408,
+      "step": 2199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.879231936311865e-06,
+      "loss": 0.3641,
+      "step": 2200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8684639368262792e-06,
+      "loss": 0.4168,
+      "step": 2201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.857725364486438e-06,
+      "loss": 0.3358,
+      "step": 2202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8470162367208467e-06,
+      "loss": 0.392,
+      "step": 2203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.83633657091022e-06,
+      "loss": 0.459,
+      "step": 2204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8256863843874594e-06,
+      "loss": 0.3609,
+      "step": 2205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8150656944376167e-06,
+      "loss": 0.3683,
+      "step": 2206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8044745182978762e-06,
+      "loss": 0.3504,
+      "step": 2207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7939128731575218e-06,
+      "loss": 0.3375,
+      "step": 2208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7833807761579035e-06,
+      "loss": 0.2928,
+      "step": 2209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.772878244392413e-06,
+      "loss": 0.4396,
+      "step": 2210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7624052949064752e-06,
+      "loss": 0.4395,
+      "step": 2211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.751961944697489e-06,
+      "loss": 0.4262,
+      "step": 2212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.741548210714814e-06,
+      "loss": 0.3262,
+      "step": 2213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7311641098597554e-06,
+      "loss": 0.3389,
+      "step": 2214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7208096589855027e-06,
+      "loss": 0.3177,
+      "step": 2215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7104848748971448e-06,
+      "loss": 0.3172,
+      "step": 2216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7001897743516148e-06,
+      "loss": 0.3485,
+      "step": 2217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6899243740576654e-06,
+      "loss": 0.4029,
+      "step": 2218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6796886906758536e-06,
+      "loss": 0.3254,
+      "step": 2219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6694827408185065e-06,
+      "loss": 0.44,
+      "step": 2220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.659306541049681e-06,
+      "loss": 0.3726,
+      "step": 2221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6491601078851683e-06,
+      "loss": 0.2991,
+      "step": 2222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6390434577924352e-06,
+      "loss": 0.4237,
+      "step": 2223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6289566071906193e-06,
+      "loss": 0.3628,
+      "step": 2224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.618899572450494e-06,
+      "loss": 0.4151,
+      "step": 2225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6088723698944342e-06,
+      "loss": 0.356,
+      "step": 2226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5988750157964083e-06,
+      "loss": 0.3285,
+      "step": 2227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5889075263819332e-06,
+      "loss": 0.4248,
+      "step": 2228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5789699178280572e-06,
+      "loss": 0.4262,
+      "step": 2229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5690622062633388e-06,
+      "loss": 0.4224,
+      "step": 2230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5591844077678065e-06,
+      "loss": 0.3736,
+      "step": 2231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5493365383729475e-06,
+      "loss": 0.3463,
+      "step": 2232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5395186140616636e-06,
+      "loss": 0.3679,
+      "step": 2233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5297306507682662e-06,
+      "loss": 0.4668,
+      "step": 2234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5199726643784395e-06,
+      "loss": 0.4171,
+      "step": 2235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5102446707292107e-06,
+      "loss": 0.408,
+      "step": 2236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5005466856089368e-06,
+      "loss": 0.3356,
+      "step": 2237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4908787247572653e-06,
+      "loss": 0.3018,
+      "step": 2238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.481240803865114e-06,
+      "loss": 0.3894,
+      "step": 2239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4716329385746541e-06,
+      "loss": 0.4113,
+      "step": 2240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4620551444792685e-06,
+      "loss": 0.3206,
+      "step": 2241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4525074371235426e-06,
+      "loss": 0.3169,
+      "step": 2242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4429898320032231e-06,
+      "loss": 0.4151,
+      "step": 2243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.433502344565212e-06,
+      "loss": 0.4042,
+      "step": 2244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4240449902075182e-06,
+      "loss": 0.3835,
+      "step": 2245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4146177842792573e-06,
+      "loss": 0.4001,
+      "step": 2246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.405220742080604e-06,
+      "loss": 0.408,
+      "step": 2247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3958538788627874e-06,
+      "loss": 0.367,
+      "step": 2248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.386517209828051e-06,
+      "loss": 0.4137,
+      "step": 2249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.377210750129634e-06,
+      "loss": 0.3792,
+      "step": 2250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3679345148717516e-06,
+      "loss": 0.3812,
+      "step": 2251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.35868851910955e-06,
+      "loss": 0.3907,
+      "step": 2252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3494727778491213e-06,
+      "loss": 0.3408,
+      "step": 2253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3402873060474364e-06,
+      "loss": 0.3759,
+      "step": 2254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3311321186123437e-06,
+      "loss": 0.3952,
+      "step": 2255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3220072304025443e-06,
+      "loss": 0.3873,
+      "step": 2256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3129126562275674e-06,
+      "loss": 0.3828,
+      "step": 2257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3038484108477256e-06,
+      "loss": 0.3696,
+      "step": 2258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2948145089741294e-06,
+      "loss": 0.4037,
+      "step": 2259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.285810965268628e-06,
+      "loss": 0.3956,
+      "step": 2260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2768377943438103e-06,
+      "loss": 0.4051,
+      "step": 2261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.267895010762965e-06,
+      "loss": 0.3988,
+      "step": 2262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2589826290400642e-06,
+      "loss": 0.3753,
+      "step": 2263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2501006636397327e-06,
+      "loss": 0.3482,
+      "step": 2264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2412491289772378e-06,
+      "loss": 0.4047,
+      "step": 2265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2324280394184563e-06,
+      "loss": 0.39,
+      "step": 2266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.223637409279852e-06,
+      "loss": 0.4003,
+      "step": 2267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2148772528284502e-06,
+      "loss": 0.3554,
+      "step": 2268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.3407,
+      "step": 2269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1974484178080804e-06,
+      "loss": 0.356,
+      "step": 2270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1887797675257806e-06,
+      "loss": 0.3712,
+      "step": 2271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1801416475039896e-06,
+      "loss": 0.3497,
+      "step": 2272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1715340717622193e-06,
+      "loss": 0.3294,
+      "step": 2273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1629570542704017e-06,
+      "loss": 0.3763,
+      "step": 2274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1544106089488815e-06,
+      "loss": 0.3861,
+      "step": 2275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1458947496683749e-06,
+      "loss": 0.414,
+      "step": 2276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1374094902499656e-06,
+      "loss": 0.4321,
+      "step": 2277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.128954844465071e-06,
+      "loss": 0.3954,
+      "step": 2278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.120530826035422e-06,
+      "loss": 0.3852,
+      "step": 2279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1121374486330483e-06,
+      "loss": 0.3852,
+      "step": 2280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1037747258802445e-06,
+      "loss": 0.395,
+      "step": 2281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0954426713495491e-06,
+      "loss": 0.3028,
+      "step": 2282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0871412985637386e-06,
+      "loss": 0.3786,
+      "step": 2283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0788706209957822e-06,
+      "loss": 0.416,
+      "step": 2284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0706306520688358e-06,
+      "loss": 0.3961,
+      "step": 2285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0624214051562177e-06,
+      "loss": 0.3695,
+      "step": 2286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0542428935813808e-06,
+      "loss": 0.3377,
+      "step": 2287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0460951306179012e-06,
+      "loss": 0.3402,
+      "step": 2288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0379781294894432e-06,
+      "loss": 0.3749,
+      "step": 2289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0298919033697462e-06,
+      "loss": 0.3473,
+      "step": 2290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0218364653826106e-06,
+      "loss": 0.3334,
+      "step": 2291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0138118286018583e-06,
+      "loss": 0.4291,
+      "step": 2292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0058180060513267e-06,
+      "loss": 0.3727,
+      "step": 2293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.978550107048401e-07,
+      "loss": 0.3898,
+      "step": 2294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.89922855486194e-07,
+      "loss": 0.3849,
+      "step": 2295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.82021553269128e-07,
+      "loss": 0.3926,
+      "step": 2296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.741511168773087e-07,
+      "loss": 0.3249,
+      "step": 2297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.663115590843053e-07,
+      "loss": 0.4061,
+      "step": 2298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.585028926135798e-07,
+      "loss": 0.38,
+      "step": 2299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.507251301384501e-07,
+      "loss": 0.4463,
+      "step": 2300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.429782842820767e-07,
+      "loss": 0.3318,
+      "step": 2301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.352623676174488e-07,
+      "loss": 0.3204,
+      "step": 2302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.275773926673559e-07,
+      "loss": 0.3699,
+      "step": 2303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.199233719043698e-07,
+      "loss": 0.3911,
+      "step": 2304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.123003177508227e-07,
+      "loss": 0.44,
+      "step": 2305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.047082425787934e-07,
+      "loss": 0.3826,
+      "step": 2306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.971471587100722e-07,
+      "loss": 0.4476,
+      "step": 2307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.896170784161607e-07,
+      "loss": 0.4214,
+      "step": 2308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.821180139182384e-07,
+      "loss": 0.3786,
+      "step": 2309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.746499773871475e-07,
+      "loss": 0.4243,
+      "step": 2310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.672129809433749e-07,
+      "loss": 0.383,
+      "step": 2311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.598070366570299e-07,
+      "loss": 0.3223,
+      "step": 2312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.524321565478133e-07,
+      "loss": 0.3744,
+      "step": 2313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.45088352585024e-07,
+      "loss": 0.3861,
+      "step": 2314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.377756366875189e-07,
+      "loss": 0.4009,
+      "step": 2315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.304940207237e-07,
+      "loss": 0.3075,
+      "step": 2316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.232435165114938e-07,
+      "loss": 0.3301,
+      "step": 2317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.160241358183363e-07,
+      "loss": 0.3845,
+      "step": 2318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.088358903611438e-07,
+      "loss": 0.3468,
+      "step": 2319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.016787918063063e-07,
+      "loss": 0.3616,
+      "step": 2320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.94552851769661e-07,
+      "loss": 0.364,
+      "step": 2321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.874580818164789e-07,
+      "loss": 0.3735,
+      "step": 2322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.803944934614383e-07,
+      "loss": 0.3902,
+      "step": 2323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.733620981686107e-07,
+      "loss": 0.3773,
+      "step": 2324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.663609073514422e-07,
+      "loss": 0.3057,
+      "step": 2325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.593909323727389e-07,
+      "loss": 0.302,
+      "step": 2326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.524521845446364e-07,
+      "loss": 0.3688,
+      "step": 2327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.455446751285955e-07,
+      "loss": 0.385,
+      "step": 2328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.386684153353774e-07,
+      "loss": 0.4194,
+      "step": 2329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.318234163250193e-07,
+      "loss": 0.3484,
+      "step": 2330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.250096892068326e-07,
+      "loss": 0.3721,
+      "step": 2331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.182272450393646e-07,
+      "loss": 0.3771,
+      "step": 2332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.114760948303967e-07,
+      "loss": 0.4074,
+      "step": 2333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.04756249536922e-07,
+      "loss": 0.3986,
+      "step": 2334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.980677200651253e-07,
+      "loss": 0.3665,
+      "step": 2335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.914105172703634e-07,
+      "loss": 0.367,
+      "step": 2336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.847846519571533e-07,
+      "loss": 0.3394,
+      "step": 2337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.781901348791509e-07,
+      "loss": 0.3865,
+      "step": 2338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.716269767391348e-07,
+      "loss": 0.3665,
+      "step": 2339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.650951881889888e-07,
+      "loss": 0.3903,
+      "step": 2340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.585947798296844e-07,
+      "loss": 0.4164,
+      "step": 2341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.52125762211262e-07,
+      "loss": 0.3997,
+      "step": 2342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.456881458328234e-07,
+      "loss": 0.387,
+      "step": 2343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.392819411424888e-07,
+      "loss": 0.3774,
+      "step": 2344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.329071585374191e-07,
+      "loss": 0.4512,
+      "step": 2345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.265638083637604e-07,
+      "loss": 0.3937,
+      "step": 2346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.202519009166596e-07,
+      "loss": 0.3567,
+      "step": 2347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.13971446440218e-07,
+      "loss": 0.2694,
+      "step": 2348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.07722455127504e-07,
+      "loss": 0.3613,
+      "step": 2349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.01504937120505e-07,
+      "loss": 0.3628,
+      "step": 2350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.953189025101447e-07,
+      "loss": 0.3145,
+      "step": 2351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.891643613362363e-07,
+      "loss": 0.3257,
+      "step": 2352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.830413235874921e-07,
+      "loss": 0.4134,
+      "step": 2353
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.769497992014827e-07,
+      "loss": 0.4324,
+      "step": 2354
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.708897980646443e-07,
+      "loss": 0.3863,
+      "step": 2355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.64861330012243e-07,
+      "loss": 0.3829,
+      "step": 2356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.588644048283675e-07,
+      "loss": 0.3634,
+      "step": 2357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.528990322459193e-07,
+      "loss": 0.3783,
+      "step": 2358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.469652219465871e-07,
+      "loss": 0.4207,
+      "step": 2359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.410629835608338e-07,
+      "loss": 0.4628,
+      "step": 2360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.351923266678838e-07,
+      "loss": 0.3875,
+      "step": 2361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.293532607957042e-07,
+      "loss": 0.3802,
+      "step": 2362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.235457954209855e-07,
+      "loss": 0.4123,
+      "step": 2363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.177699399691416e-07,
+      "loss": 0.2639,
+      "step": 2364
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.120257038142762e-07,
+      "loss": 0.3623,
+      "step": 2365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.063130962791762e-07,
+      "loss": 0.3726,
+      "step": 2366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.00632126635301e-07,
+      "loss": 0.4314,
+      "step": 2367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.949828041027571e-07,
+      "loss": 0.3486,
+      "step": 2368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.893651378502884e-07,
+      "loss": 0.3727,
+      "step": 2369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.837791369952638e-07,
+      "loss": 0.3625,
+      "step": 2370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.78224810603658e-07,
+      "loss": 0.4223,
+      "step": 2371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7270216769004007e-07,
+      "loss": 0.3731,
+      "step": 2372
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.67211217217558e-07,
+      "loss": 0.3979,
+      "step": 2373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.617519680979232e-07,
+      "loss": 0.31,
+      "step": 2374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.563244291913904e-07,
+      "loss": 0.3437,
+      "step": 2375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5092860930676e-07,
+      "loss": 0.3315,
+      "step": 2376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.4556451720134233e-07,
+      "loss": 0.4199,
+      "step": 2377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.402321615809602e-07,
+      "loss": 0.3981,
+      "step": 2378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.349315510999308e-07,
+      "loss": 0.4188,
+      "step": 2379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.296626943610438e-07,
+      "loss": 0.3684,
+      "step": 2380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2442559991555666e-07,
+      "loss": 0.3179,
+      "step": 2381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1922027626317474e-07,
+      "loss": 0.3434,
+      "step": 2382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.140467318520425e-07,
+      "loss": 0.3783,
+      "step": 2383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0890497507872553e-07,
+      "loss": 0.3649,
+      "step": 2384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0379501428820187e-07,
+      "loss": 0.4107,
+      "step": 2385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9871685777384204e-07,
+      "loss": 0.3517,
+      "step": 2386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.936705137773955e-07,
+      "loss": 0.417,
+      "step": 2387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8865599048899085e-07,
+      "loss": 0.413,
+      "step": 2388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8367329604710015e-07,
+      "loss": 0.3144,
+      "step": 2389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7872243853855016e-07,
+      "loss": 0.2752,
+      "step": 2390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.738034259984846e-07,
+      "loss": 0.3948,
+      "step": 2391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.689162664103729e-07,
+      "loss": 0.3177,
+      "step": 2392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6406096770598365e-07,
+      "loss": 0.3256,
+      "step": 2393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.592375377653734e-07,
+      "loss": 0.4188,
+      "step": 2394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.544459844168824e-07,
+      "loss": 0.3335,
+      "step": 2395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.496863154371122e-07,
+      "loss": 0.3446,
+      "step": 2396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449585385509169e-07,
+      "loss": 0.3682,
+      "step": 2397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4026266143138977e-07,
+      "loss": 0.3506,
+      "step": 2398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.355986916998544e-07,
+      "loss": 0.3669,
+      "step": 2399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.309666369258424e-07,
+      "loss": 0.387,
+      "step": 2400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2636650462709587e-07,
+      "loss": 0.3461,
+      "step": 2401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.217983022695426e-07,
+      "loss": 0.4529,
+      "step": 2402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1726203726728966e-07,
+      "loss": 0.2854,
+      "step": 2403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1275771698261235e-07,
+      "loss": 0.3388,
+      "step": 2404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.082853487259407e-07,
+      "loss": 0.3782,
+      "step": 2405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.038449397558396e-07,
+      "loss": 0.344,
+      "step": 2406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.994364972790131e-07,
+      "loss": 0.3309,
+      "step": 2407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.950600284502825e-07,
+      "loss": 0.367,
+      "step": 2408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9071554037257477e-07,
+      "loss": 0.3662,
+      "step": 2409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8640304009690977e-07,
+      "loss": 0.3335,
+      "step": 2410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8212253462239994e-07,
+      "loss": 0.3359,
+      "step": 2411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7787403089622353e-07,
+      "loss": 0.405,
+      "step": 2412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.736575358136251e-07,
+      "loss": 0.3514,
+      "step": 2413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.694730562178949e-07,
+      "loss": 0.3787,
+      "step": 2414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.653205989003671e-07,
+      "loss": 0.3206,
+      "step": 2415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.612001706004019e-07,
+      "loss": 0.3199,
+      "step": 2416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5711177800538114e-07,
+      "loss": 0.3006,
+      "step": 2417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.530554277506836e-07,
+      "loss": 0.3187,
+      "step": 2418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.490311264196965e-07,
+      "loss": 0.3832,
+      "step": 2419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.45038880543782e-07,
+      "loss": 0.3085,
+      "step": 2420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.410786966022816e-07,
+      "loss": 0.3341,
+      "step": 2421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3715058102250277e-07,
+      "loss": 0.3442,
+      "step": 2422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.332545401796993e-07,
+      "loss": 0.3524,
+      "step": 2423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.293905803970775e-07,
+      "loss": 0.3417,
+      "step": 2424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2555870794577217e-07,
+      "loss": 0.3487,
+      "step": 2425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2175892904483743e-07,
+      "loss": 0.3093,
+      "step": 2426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1799124986124464e-07,
+      "loss": 0.3902,
+      "step": 2427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1425567650987132e-07,
+      "loss": 0.3538,
+      "step": 2428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1055221505347888e-07,
+      "loss": 0.4747,
+      "step": 2429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0688087150271707e-07,
+      "loss": 0.3107,
+      "step": 2430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0324165181610845e-07,
+      "loss": 0.3071,
+      "step": 2431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9963456190004172e-07,
+      "loss": 0.4234,
+      "step": 2432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9605960760875175e-07,
+      "loss": 0.3715,
+      "step": 2433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9251679474432628e-07,
+      "loss": 0.3092,
+      "step": 2434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.890061290566836e-07,
+      "loss": 0.4268,
+      "step": 2435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8552761624357041e-07,
+      "loss": 0.3764,
+      "step": 2436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8208126195054632e-07,
+      "loss": 0.4222,
+      "step": 2437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.786670717709793e-07,
+      "loss": 0.3904,
+      "step": 2438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7528505124603911e-07,
+      "loss": 0.3622,
+      "step": 2439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.719352058646817e-07,
+      "loss": 0.3243,
+      "step": 2440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.686175410636426e-07,
+      "loss": 0.4003,
+      "step": 2441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6533206222743015e-07,
+      "loss": 0.2536,
+      "step": 2442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.620787746883168e-07,
+      "loss": 0.3586,
+      "step": 2443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588576837263278e-07,
+      "loss": 0.3732,
+      "step": 2444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5566879456923034e-07,
+      "loss": 0.3364,
+      "step": 2445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.525121123925355e-07,
+      "loss": 0.3065,
+      "step": 2446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4938764231948067e-07,
+      "loss": 0.3521,
+      "step": 2447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4629538942102062e-07,
+      "loss": 0.3076,
+      "step": 2448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4323535871582307e-07,
+      "loss": 0.4389,
+      "step": 2449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.402075551702642e-07,
+      "loss": 0.3902,
+      "step": 2450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3721198369840872e-07,
+      "loss": 0.3089,
+      "step": 2451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3424864916201652e-07,
+      "loss": 0.3581,
+      "step": 2452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.313175563705249e-07,
+      "loss": 0.3497,
+      "step": 2453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.284187100810419e-07,
+      "loss": 0.4075,
+      "step": 2454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2555211499833964e-07,
+      "loss": 0.2975,
+      "step": 2455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.227177757748499e-07,
+      "loss": 0.2869,
+      "step": 2456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.19915697010653e-07,
+      "loss": 0.3911,
+      "step": 2457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.171458832534711e-07,
+      "loss": 0.3894,
+      "step": 2458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1440833899865721e-07,
+      "loss": 0.3776,
+      "step": 2459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1170306868919955e-07,
+      "loss": 0.4019,
+      "step": 2460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0903007671569931e-07,
+      "loss": 0.3123,
+      "step": 2461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0638936741637074e-07,
+      "loss": 0.3188,
+      "step": 2462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0378094507703884e-07,
+      "loss": 0.3918,
+      "step": 2463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0120481393112169e-07,
+      "loss": 0.381,
+      "step": 2464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.866097815963482e-08,
+      "loss": 0.4046,
+      "step": 2465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.614944189117569e-08,
+      "loss": 0.4555,
+      "step": 2466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.367020920191927e-08,
+      "loss": 0.3919,
+      "step": 2467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.122328411561576e-08,
+      "loss": 0.4136,
+      "step": 2468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.880867060357956e-08,
+      "loss": 0.3577,
+      "step": 2469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.642637258468257e-08,
+      "loss": 0.2694,
+      "step": 2470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.407639392534972e-08,
+      "loss": 0.3707,
+      "step": 2471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.175873843955462e-08,
+      "loss": 0.3475,
+      "step": 2472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.947340988880614e-08,
+      "loss": 0.3674,
+      "step": 2473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.722041198215291e-08,
+      "loss": 0.392,
+      "step": 2474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.499974837616553e-08,
+      "loss": 0.3522,
+      "step": 2475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.281142267494323e-08,
+      "loss": 0.4086,
+      "step": 2476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.065543843009393e-08,
+      "loss": 0.3541,
+      "step": 2477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.853179914074304e-08,
+      "loss": 0.3849,
+      "step": 2478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.644050825351578e-08,
+      "loss": 0.394,
+      "step": 2479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.438156916254157e-08,
+      "loss": 0.3486,
+      "step": 2480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.235498520943628e-08,
+      "loss": 0.3687,
+      "step": 2481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.03607596833089e-08,
+      "loss": 0.3797,
+      "step": 2482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.839889582075264e-08,
+      "loss": 0.4394,
+      "step": 2483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.6469396805833855e-08,
+      "loss": 0.3179,
+      "step": 2484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.457226577009422e-08,
+      "loss": 0.4367,
+      "step": 2485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.270750579254191e-08,
+      "loss": 0.3223,
+      "step": 2486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0875119899647106e-08,
+      "loss": 0.3822,
+      "step": 2487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.90751110653398e-08,
+      "loss": 0.3072,
+      "step": 2488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7307482210998676e-08,
+      "loss": 0.4018,
+      "step": 2489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.557223620545559e-08,
+      "loss": 0.3325,
+      "step": 2490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3869375864979965e-08,
+      "loss": 0.3241,
+      "step": 2491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2198903953283296e-08,
+      "loss": 0.4665,
+      "step": 2492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0560823181508005e-08,
+      "loss": 0.4372,
+      "step": 2493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8955136208231884e-08,
+      "loss": 0.3117,
+      "step": 2494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7381845639454795e-08,
+      "loss": 0.3545,
+      "step": 2495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.584095402859644e-08,
+      "loss": 0.356,
+      "step": 2496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.433246387649858e-08,
+      "loss": 0.3955,
+      "step": 2497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2856377631407255e-08,
+      "loss": 0.3859,
+      "step": 2498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1412697688988356e-08,
+      "loss": 0.419,
+      "step": 2499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.000142639230319e-08,
+      "loss": 0.3467,
+      "step": 2500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8622566031821785e-08,
+      "loss": 0.3594,
+      "step": 2501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7276118845407372e-08,
+      "loss": 0.3457,
+      "step": 2502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5962087018318592e-08,
+      "loss": 0.4135,
+      "step": 2503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.468047268320506e-08,
+      "loss": 0.3377,
+      "step": 2504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3431277920105132e-08,
+      "loss": 0.4223,
+      "step": 2505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.221450475643483e-08,
+      "loss": 0.3093,
+      "step": 2506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.103015516699669e-08,
+      "loss": 0.3691,
+      "step": 2507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.987823107396647e-08,
+      "loss": 0.3345,
+      "step": 2508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8758734346899787e-08,
+      "loss": 0.4174,
+      "step": 2509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.767166680271215e-08,
+      "loss": 0.3849,
+      "step": 2510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6617030205701156e-08,
+      "loss": 0.3832,
+      "step": 2511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5594826267517627e-08,
+      "loss": 0.3329,
+      "step": 2512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4605056647178928e-08,
+      "loss": 0.3766,
+      "step": 2513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3647722951066755e-08,
+      "loss": 0.3201,
+      "step": 2514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2722826732913806e-08,
+      "loss": 0.501,
+      "step": 2515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1830369493808225e-08,
+      "loss": 0.3944,
+      "step": 2516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0970352682193597e-08,
+      "loss": 0.3361,
+      "step": 2517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0142777693860074e-08,
+      "loss": 0.3063,
+      "step": 2518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.347645871946586e-09,
+      "loss": 0.4163,
+      "step": 2519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.584958506938635e-09,
+      "loss": 0.3663,
+      "step": 2520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.854716836661613e-09,
+      "loss": 0.4427,
+      "step": 2521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.156922046285264e-09,
+      "loss": 0.4262,
+      "step": 2522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.491575268317008e-09,
+      "loss": 0.3778,
+      "step": 2523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.858677582601946e-09,
+      "loss": 0.3459,
+      "step": 2524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.258230016320642e-09,
+      "loss": 0.3492,
+      "step": 2525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.69023354399134e-09,
+      "loss": 0.3649,
+      "step": 2526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.154689087461083e-09,
+      "loss": 0.3344,
+      "step": 2527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.651597515905714e-09,
+      "loss": 0.4369,
+      "step": 2528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.180959645834314e-09,
+      "loss": 0.3856,
+      "step": 2529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.742776241086986e-09,
+      "loss": 0.3856,
+      "step": 2530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.33704801282153e-09,
+      "loss": 0.3426,
+      "step": 2531
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.963775619533426e-09,
+      "loss": 0.3974,
+      "step": 2532
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6229596670314096e-09,
+      "loss": 0.4207,
+      "step": 2533
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3146007084552382e-09,
+      "loss": 0.3537,
+      "step": 2534
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0386992442690259e-09,
+      "loss": 0.3307,
+      "step": 2535
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.952557222501434e-10,
+      "loss": 0.3862,
+      "step": 2536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.842705375069813e-10,
+      "loss": 0.3566,
+      "step": 2537
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.057440324634065e-10,
+      "loss": 0.4515,
+      "step": 2538
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5967649686320373e-10,
+      "loss": 0.3491,
+      "step": 2539
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.460681677745157e-10,
+      "loss": 0.3722,
+      "step": 2540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.491922957874152e-11,
+      "loss": 0.3274,
+      "step": 2541
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.622981397941814e-11,
+      "loss": 0.3585,
+      "step": 2542
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3071,
+      "step": 2543
+    },
+    {
+      "epoch": 1.0,
+      "step": 2543,
+      "total_flos": 1.097140348541722e+22,
+      "train_loss": 0.4539335052059056,
+      "train_runtime": 9679.0162,
+      "train_samples_per_second": 33.106,
+      "train_steps_per_second": 0.263
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2543,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 1.097140348541722e+22,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}