diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13476 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2241,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 2.4788,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 2.1742,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 2.3452,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 2.4186,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 2.0701,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 2.1071,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 1.7251,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.7863,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 1.8111,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 1.7325,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 1.7824,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.7706,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 1.724,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 1.7277,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 1.7394,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.6897,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.6762,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 1.6358,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 1.5994,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 1.6197,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 1.5702,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 1.6441,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 1.6177,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 1.493,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 1.6278,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 1.583,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 1.5539,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 1.4853,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 1.5598,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 1.4251,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 1.5344,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 1.5459,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 1.4635,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 1.4967,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.147058823529411e-05,
+      "loss": 1.4587,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.294117647058824e-05,
+      "loss": 1.4996,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.441176470588235e-05,
+      "loss": 1.57,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.588235294117647e-05,
+      "loss": 1.5212,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.735294117647059e-05,
+      "loss": 1.4115,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 1.5373,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0294117647058825e-05,
+      "loss": 1.4239,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.176470588235295e-05,
+      "loss": 1.4767,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.323529411764705e-05,
+      "loss": 1.5196,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.470588235294118e-05,
+      "loss": 1.3984,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.61764705882353e-05,
+      "loss": 1.62,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.764705882352942e-05,
+      "loss": 1.3954,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.911764705882354e-05,
+      "loss": 1.4746,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.058823529411765e-05,
+      "loss": 1.4916,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.205882352941177e-05,
+      "loss": 1.4281,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.352941176470589e-05,
+      "loss": 1.4423,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.4676,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.647058823529411e-05,
+      "loss": 1.4503,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.794117647058824e-05,
+      "loss": 1.4357,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.941176470588235e-05,
+      "loss": 1.4552,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.088235294117648e-05,
+      "loss": 1.3965,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.23529411764706e-05,
+      "loss": 1.4523,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.382352941176471e-05,
+      "loss": 1.5022,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.529411764705883e-05,
+      "loss": 1.3563,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.676470588235295e-05,
+      "loss": 1.4362,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.823529411764706e-05,
+      "loss": 1.4223,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.970588235294118e-05,
+      "loss": 1.3593,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.11764705882353e-05,
+      "loss": 1.4579,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.264705882352942e-05,
+      "loss": 1.4635,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.411764705882353e-05,
+      "loss": 1.3896,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.558823529411765e-05,
+      "loss": 1.4864,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.705882352941177e-05,
+      "loss": 1.3461,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.852941176470589e-05,
+      "loss": 1.4454,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 1.4292,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999994774591761e-05,
+      "loss": 1.3922,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999979098377963e-05,
+      "loss": 1.4472,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999952971391373e-05,
+      "loss": 1.4291,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9999163936866e-05,
+      "loss": 1.438,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9998693653401e-05,
+      "loss": 1.372,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999811886450166e-05,
+      "loss": 1.4195,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99974395713694e-05,
+      "loss": 1.3846,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999665577542406e-05,
+      "loss": 1.4186,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99957674783039e-05,
+      "loss": 1.4086,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99947746818656e-05,
+      "loss": 1.3691,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999367738818428e-05,
+      "loss": 1.4126,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999247559955345e-05,
+      "loss": 1.3972,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999116931848504e-05,
+      "loss": 1.3279,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99897585477094e-05,
+      "loss": 1.4977,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998824329017527e-05,
+      "loss": 1.4041,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998662354904977e-05,
+      "loss": 1.3215,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998489932771845e-05,
+      "loss": 1.4292,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99830706297852e-05,
+      "loss": 1.3729,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99811374590723e-05,
+      "loss": 1.3613,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99790998196204e-05,
+      "loss": 1.3718,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997695771568849e-05,
+      "loss": 1.3346,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99747111517539e-05,
+      "loss": 1.366,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997236013251234e-05,
+      "loss": 1.4396,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996990466287783e-05,
+      "loss": 1.3603,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996734474798268e-05,
+      "loss": 1.4104,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996468039317754e-05,
+      "loss": 1.4893,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996191160403137e-05,
+      "loss": 1.3518,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995903838633134e-05,
+      "loss": 1.3501,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995606074608299e-05,
+      "loss": 1.4145,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995297868951006e-05,
+      "loss": 1.3133,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994979222305453e-05,
+      "loss": 1.4353,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994650135337667e-05,
+      "loss": 1.3907,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994310608735491e-05,
+      "loss": 1.2955,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993960643208591e-05,
+      "loss": 1.5013,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993600239488454e-05,
+      "loss": 1.394,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993229398328382e-05,
+      "loss": 1.3382,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992848120503493e-05,
+      "loss": 1.4251,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992456406810719e-05,
+      "loss": 1.3131,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99205425806881e-05,
+      "loss": 1.4099,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991641675118316e-05,
+      "loss": 1.4179,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991218658821608e-05,
+      "loss": 1.2979,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990785210062856e-05,
+      "loss": 1.4283,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99034132974804e-05,
+      "loss": 1.4327,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989887018804942e-05,
+      "loss": 1.2713,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989422278183146e-05,
+      "loss": 1.3652,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988947108854037e-05,
+      "loss": 1.4147,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988461511810796e-05,
+      "loss": 1.3238,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987965488068398e-05,
+      "loss": 1.3624,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987459038663616e-05,
+      "loss": 1.3728,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986942164655012e-05,
+      "loss": 1.3642,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986414867122936e-05,
+      "loss": 1.339,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985877147169525e-05,
+      "loss": 1.3893,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985329005918702e-05,
+      "loss": 1.3028,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984770444516174e-05,
+      "loss": 1.4613,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984201464129425e-05,
+      "loss": 1.3835,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983622065947714e-05,
+      "loss": 1.3367,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983032251182081e-05,
+      "loss": 1.3554,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982432021065333e-05,
+      "loss": 1.3255,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98182137685205e-05,
+      "loss": 1.3448,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98120031981858e-05,
+      "loss": 1.3574,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980568851263029e-05,
+      "loss": 1.3003,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979926972505273e-05,
+      "loss": 1.4174,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979274684886943e-05,
+      "loss": 1.4257,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978611989771424e-05,
+      "loss": 1.2841,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977938888543862e-05,
+      "loss": 1.318,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977255382611144e-05,
+      "loss": 1.3566,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97656147340191e-05,
+      "loss": 1.367,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975857162366546e-05,
+      "loss": 1.37,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975142450977174e-05,
+      "loss": 1.3842,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974417340727659e-05,
+      "loss": 1.3769,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973681833133598e-05,
+      "loss": 1.4379,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972935929732325e-05,
+      "loss": 1.299,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972179632082898e-05,
+      "loss": 1.4021,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971412941766105e-05,
+      "loss": 1.3592,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970635860384449e-05,
+      "loss": 1.3067,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969848389562162e-05,
+      "loss": 1.3637,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969050530945184e-05,
+      "loss": 1.3433,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968242286201171e-05,
+      "loss": 1.2948,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967423657019485e-05,
+      "loss": 1.4001,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966594645111196e-05,
+      "loss": 1.3833,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965755252209073e-05,
+      "loss": 1.3291,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964905480067586e-05,
+      "loss": 1.446,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964045330462896e-05,
+      "loss": 1.3754,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963174805192857e-05,
+      "loss": 1.2877,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962293906077006e-05,
+      "loss": 1.3607,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961402634956573e-05,
+      "loss": 1.2588,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960500993694454e-05,
+      "loss": 1.3904,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959588984175228e-05,
+      "loss": 1.3443,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958666608305145e-05,
+      "loss": 1.3054,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95773386801212e-05,
+      "loss": 1.3815,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956790765245732e-05,
+      "loss": 1.3279,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955837301977223e-05,
+      "loss": 1.2615,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954873480199482e-05,
+      "loss": 1.3867,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953899301927057e-05,
+      "loss": 1.2997,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95291476919614e-05,
+      "loss": 1.3289,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951919884064564e-05,
+      "loss": 1.3931,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950914648611802e-05,
+      "loss": 1.3387,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949899064938959e-05,
+      "loss": 1.2911,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948873135168773e-05,
+      "loss": 1.3589,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947836861445603e-05,
+      "loss": 1.3822,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946790245935429e-05,
+      "loss": 1.2447,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945733290825852e-05,
+      "loss": 1.3806,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944665998326079e-05,
+      "loss": 1.2989,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943588370666926e-05,
+      "loss": 1.3644,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942500410100808e-05,
+      "loss": 1.3536,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941402118901744e-05,
+      "loss": 1.301,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940293499365339e-05,
+      "loss": 1.3621,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939174553808793e-05,
+      "loss": 1.3381,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938045284570878e-05,
+      "loss": 1.3471,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936905694011959e-05,
+      "loss": 1.2256,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93575578451396e-05,
+      "loss": 1.3558,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934595558480384e-05,
+      "loss": 1.3343,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933425018336293e-05,
+      "loss": 1.3473,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932244166528302e-05,
+      "loss": 1.3312,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93105300552459e-05,
+      "loss": 1.286,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929851537814871e-05,
+      "loss": 1.3196,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928639765910417e-05,
+      "loss": 1.2862,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927417692344024e-05,
+      "loss": 1.3695,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926185319670027e-05,
+      "loss": 1.3424,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924942650464287e-05,
+      "loss": 1.3259,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923689687324181e-05,
+      "loss": 1.3039,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92242643286861e-05,
+      "loss": 1.3714,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921152889737984e-05,
+      "loss": 1.3822,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919869060594214e-05,
+      "loss": 1.2961,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918574948120711e-05,
+      "loss": 1.3482,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917270555022384e-05,
+      "loss": 1.2113,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915955884025626e-05,
+      "loss": 1.3763,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914630937878315e-05,
+      "loss": 1.342,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913295719349805e-05,
+      "loss": 1.3451,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911950231230919e-05,
+      "loss": 1.3275,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910594476333948e-05,
+      "loss": 1.3738,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909228457492641e-05,
+      "loss": 1.2818,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907852177562202e-05,
+      "loss": 1.2912,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906465639419278e-05,
+      "loss": 1.2939,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905068845961962e-05,
+      "loss": 1.3576,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90366180010978e-05,
+      "loss": 1.2457,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902244504803688e-05,
+      "loss": 1.281,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900816963006063e-05,
+      "loss": 1.4009,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899379177700703e-05,
+      "loss": 1.3041,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897931151892811e-05,
+      "loss": 1.3251,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896472888609e-05,
+      "loss": 1.3169,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895004390897278e-05,
+      "loss": 1.3042,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893525661827043e-05,
+      "loss": 1.324,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892036704489084e-05,
+      "loss": 1.2761,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.890537521995563e-05,
+      "loss": 1.3923,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889028117480013e-05,
+      "loss": 1.265,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88750849409734e-05,
+      "loss": 1.3626,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885978655023804e-05,
+      "loss": 1.3316,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884438603457019e-05,
+      "loss": 1.3046,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882888342615944e-05,
+      "loss": 1.2845,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881327875740876e-05,
+      "loss": 1.3052,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879757206093445e-05,
+      "loss": 1.3196,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87817633695661e-05,
+      "loss": 1.3122,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876585271634645e-05,
+      "loss": 1.3242,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874984013453135e-05,
+      "loss": 1.2824,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873372565758971e-05,
+      "loss": 1.3733,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871750931920344e-05,
+      "loss": 1.2279,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87011911532673e-05,
+      "loss": 1.3603,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868477119388896e-05,
+      "loss": 1.3438,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866824947538879e-05,
+      "loss": 1.2087,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865162603229988e-05,
+      "loss": 1.321,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863490089936796e-05,
+      "loss": 1.3286,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861807411155126e-05,
+      "loss": 1.3005,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860114570402054e-05,
+      "loss": 1.3387,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858411571215893e-05,
+      "loss": 1.2462,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856698417156189e-05,
+      "loss": 1.2877,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854975111803713e-05,
+      "loss": 1.3004,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853241658760457e-05,
+      "loss": 1.3326,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851498061649618e-05,
+      "loss": 1.3021,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849744324115602e-05,
+      "loss": 1.3047,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847980449824003e-05,
+      "loss": 1.3134,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846206442461608e-05,
+      "loss": 1.3068,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844422305736382e-05,
+      "loss": 1.2949,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842628043377464e-05,
+      "loss": 1.3063,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840823659135153e-05,
+      "loss": 1.3447,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839009156780908e-05,
+      "loss": 1.3203,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837184540107333e-05,
+      "loss": 1.2414,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835349812928179e-05,
+      "loss": 1.4108,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83350497907832e-05,
+      "loss": 1.3049,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831650042413763e-05,
+      "loss": 1.3052,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82978500681163e-05,
+      "loss": 1.2255,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827909876170147e-05,
+      "loss": 1.3081,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826024654408645e-05,
+      "loss": 1.22,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824129345467546e-05,
+      "loss": 1.3085,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822223953308354e-05,
+      "loss": 1.3079,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820308481913648e-05,
+      "loss": 1.2943,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818382935287078e-05,
+      "loss": 1.3881,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816447317453352e-05,
+      "loss": 1.2608,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814501632458226e-05,
+      "loss": 1.3446,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8125458843685e-05,
+      "loss": 1.2717,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810580077272005e-05,
+      "loss": 1.3419,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808604215277599e-05,
+      "loss": 1.2727,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80661830251516e-05,
+      "loss": 1.3,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804622343135565e-05,
+      "loss": 1.3355,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802616341310699e-05,
+      "loss": 1.2751,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800600301233432e-05,
+      "loss": 1.3762,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798574227117616e-05,
+      "loss": 1.2399,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796538123198077e-05,
+      "loss": 1.327,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794491993730606e-05,
+      "loss": 1.2904,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792435842991945e-05,
+      "loss": 1.2915,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790369675279789e-05,
+      "loss": 1.2446,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788293494912762e-05,
+      "loss": 1.2928,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786207306230423e-05,
+      "loss": 1.279,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784111113593243e-05,
+      "loss": 1.2988,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782004921382612e-05,
+      "loss": 1.2702,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779888734000813e-05,
+      "loss": 1.326,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777762555871023e-05,
+      "loss": 1.3681,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775626391437301e-05,
+      "loss": 1.2298,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773480245164582e-05,
+      "loss": 1.3702,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771324121538659e-05,
+      "loss": 1.2935,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769158025066185e-05,
+      "loss": 1.3036,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766981960274653e-05,
+      "loss": 1.3071,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764795931712396e-05,
+      "loss": 1.2876,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76259994394857e-05,
+      "loss": 1.1742,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760394001573148e-05,
+      "loss": 1.3212,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758178109196907e-05,
+      "loss": 1.3155,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75595227145143e-05,
+      "loss": 1.2527,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753716492989076e-05,
+      "loss": 1.3951,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751470778482988e-05,
+      "loss": 1.2679,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749215132627078e-05,
+      "loss": 1.3106,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746949560136013e-05,
+      "loss": 1.3018,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744674065745209e-05,
+      "loss": 1.3096,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742388654210822e-05,
+      "loss": 1.3182,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740093330309734e-05,
+      "loss": 1.2122,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737788098839549e-05,
+      "loss": 1.3048,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735472964618575e-05,
+      "loss": 1.2845,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733147932485821e-05,
+      "loss": 1.3337,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730813007300984e-05,
+      "loss": 1.2567,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72846819394444e-05,
+      "loss": 1.3662,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726113497317229e-05,
+      "loss": 1.2151,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723748922341055e-05,
+      "loss": 1.2647,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721374473958263e-05,
+      "loss": 1.3183,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71899015713184e-05,
+      "loss": 1.2283,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716595976845396e-05,
+      "loss": 1.3352,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.714191938103159e-05,
+      "loss": 1.3045,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711778045929962e-05,
+      "loss": 1.2833,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709354305371236e-05,
+      "loss": 1.2972,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706920721492994e-05,
+      "loss": 1.3364,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704477299381821e-05,
+      "loss": 1.2811,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702024044144871e-05,
+      "loss": 1.2237,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699560960909847e-05,
+      "loss": 1.281,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697088054824994e-05,
+      "loss": 1.2569,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694605331059093e-05,
+      "loss": 1.3094,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69211279480144e-05,
+      "loss": 1.3494,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689610451261841e-05,
+      "loss": 1.2855,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687098305670605e-05,
+      "loss": 1.2861,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684576363278526e-05,
+      "loss": 1.3211,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682044629356875e-05,
+      "loss": 1.2666,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67950310919739e-05,
+      "loss": 1.1526,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676951808112262e-05,
+      "loss": 1.2728,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674390731434129e-05,
+      "loss": 1.2594,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671819884516057e-05,
+      "loss": 1.3079,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669239272731538e-05,
+      "loss": 1.2872,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66664890147447e-05,
+      "loss": 1.2966,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664048776159153e-05,
+      "loss": 1.2602,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661438902220274e-05,
+      "loss": 1.2837,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.658819285112895e-05,
+      "loss": 1.3598,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656189930312443e-05,
+      "loss": 1.3031,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653550843314701e-05,
+      "loss": 1.2805,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650902029635789e-05,
+      "loss": 1.3301,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648243494812162e-05,
+      "loss": 1.285,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64557524440059e-05,
+      "loss": 1.2572,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642897283978156e-05,
+      "loss": 1.2886,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64020961914223e-05,
+      "loss": 1.3211,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637512255510475e-05,
+      "loss": 1.2536,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634805198720815e-05,
+      "loss": 1.2892,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632088454431448e-05,
+      "loss": 1.3401,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629362028320809e-05,
+      "loss": 1.2617,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626625926087574e-05,
+      "loss": 1.3269,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623880153450643e-05,
+      "loss": 1.2647,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621124716149132e-05,
+      "loss": 1.3258,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618359619942354e-05,
+      "loss": 1.3031,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615584870609808e-05,
+      "loss": 1.0721,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612800473951179e-05,
+      "loss": 1.3374,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610006435786306e-05,
+      "loss": 1.2969,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607202761955188e-05,
+      "loss": 1.2555,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604389458317958e-05,
+      "loss": 1.2956,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601566530754882e-05,
+      "loss": 1.3503,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598733985166341e-05,
+      "loss": 1.2656,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595891827472815e-05,
+      "loss": 1.4059,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593040063614879e-05,
+      "loss": 1.1727,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590178699553186e-05,
+      "loss": 1.2831,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587307741268452e-05,
+      "loss": 1.2973,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584427194761452e-05,
+      "loss": 1.2972,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581537066052994e-05,
+      "loss": 1.2537,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578637361183921e-05,
+      "loss": 1.3121,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575728086215092e-05,
+      "loss": 1.3309,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572809247227366e-05,
+      "loss": 1.2258,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569880850321589e-05,
+      "loss": 1.3473,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566942901618593e-05,
+      "loss": 1.191,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563995407259169e-05,
+      "loss": 1.2859,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561038373404062e-05,
+      "loss": 1.3234,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558071806233955e-05,
+      "loss": 1.2424,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555095711949459e-05,
+      "loss": 1.359,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552110096771095e-05,
+      "loss": 1.3321,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549114966939288e-05,
+      "loss": 1.1005,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546110328714349e-05,
+      "loss": 1.3162,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54309618837646e-05,
+      "loss": 1.2853,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54007255222567e-05,
+      "loss": 1.2703,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537039426581867e-05,
+      "loss": 1.3211,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533996817784785e-05,
+      "loss": 1.2235,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.53094473219397e-05,
+      "loss": 1.2583,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527883176188779e-05,
+      "loss": 1.325,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524812156168363e-05,
+      "loss": 1.2334,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52173167855166e-05,
+      "loss": 1.3312,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518641749777366e-05,
+      "loss": 1.2864,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515542376303941e-05,
+      "loss": 1.2713,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512433564609578e-05,
+      "loss": 1.3,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509315321192202e-05,
+      "loss": 1.3054,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506187652569453e-05,
+      "loss": 1.2117,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503050565278668e-05,
+      "loss": 1.3333,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499904065876873e-05,
+      "loss": 1.2562,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496748160940761e-05,
+      "loss": 1.2044,
+      "step": 381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493582857066694e-05,
+      "loss": 1.3498,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49040816087067e-05,
+      "loss": 1.3053,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487224078988325e-05,
+      "loss": 1.2576,
+      "step": 384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484030618074911e-05,
+      "loss": 1.291,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480827784805278e-05,
+      "loss": 1.2896,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477615585873876e-05,
+      "loss": 1.2442,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474394027994722e-05,
+      "loss": 1.3277,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471163117901398e-05,
+      "loss": 1.2422,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467922862347037e-05,
+      "loss": 1.2427,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464673268104298e-05,
+      "loss": 1.3466,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461414341965366e-05,
+      "loss": 1.2386,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458146090741928e-05,
+      "loss": 1.3285,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454868521265164e-05,
+      "loss": 1.2177,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451581640385728e-05,
+      "loss": 1.239,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448285454973738e-05,
+      "loss": 1.2831,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444979971918759e-05,
+      "loss": 1.3271,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441665198129791e-05,
+      "loss": 1.2665,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438341140535256e-05,
+      "loss": 1.251,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435007806082972e-05,
+      "loss": 1.2748,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431665201740155e-05,
+      "loss": 1.2347,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428313334493392e-05,
+      "loss": 1.335,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424952211348635e-05,
+      "loss": 1.3679,
+      "step": 403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42158183933118e-05,
+      "loss": 1.2648,
+      "step": 404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418202225485653e-05,
+      "loss": 1.2194,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414813376876002e-05,
+      "loss": 1.2709,
+      "step": 406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41141530058547e-05,
+      "loss": 1.2606,
+      "step": 407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408008003716593e-05,
+      "loss": 1.3222,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404591493391181e-05,
+      "loss": 1.219,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401165776750293e-05,
+      "loss": 1.3101,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397730860954242e-05,
+      "loss": 1.3294,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394286753182557e-05,
+      "loss": 1.202,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390833460633989e-05,
+      "loss": 1.2348,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387370990526485e-05,
+      "loss": 1.2443,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38389935009717e-05,
+      "loss": 1.2166,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380418546602338e-05,
+      "loss": 1.292,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376928587317439e-05,
+      "loss": 1.2403,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37342947953706e-05,
+      "loss": 1.288,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369921230574906e-05,
+      "loss": 1.32,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366403847763787e-05,
+      "loss": 1.2059,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362877338455611e-05,
+      "loss": 1.2986,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359341710021356e-05,
+      "loss": 1.2622,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355796969851066e-05,
+      "loss": 1.2562,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352243125353824e-05,
+      "loss": 1.2654,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348680183957748e-05,
+      "loss": 1.2919,
+      "step": 425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345108153109963e-05,
+      "loss": 1.2695,
+      "step": 426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341527040276601e-05,
+      "loss": 1.2974,
+      "step": 427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337936852942769e-05,
+      "loss": 1.3071,
+      "step": 428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334337598612548e-05,
+      "loss": 1.218,
+      "step": 429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330729284808966e-05,
+      "loss": 1.3467,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327111919073987e-05,
+      "loss": 1.2196,
+      "step": 431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3234855089685e-05,
+      "loss": 1.2372,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31985006207229e-05,
+      "loss": 1.3527,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316205585984035e-05,
+      "loss": 1.1242,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312552088321286e-05,
+      "loss": 1.2803,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308889576720453e-05,
+      "loss": 1.2757,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.305218058836778e-05,
+      "loss": 1.2415,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301537542344337e-05,
+      "loss": 1.2892,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297848034936006e-05,
+      "loss": 1.3609,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294149544323463e-05,
+      "loss": 1.1765,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290442078237154e-05,
+      "loss": 1.3365,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28672564442629e-05,
+      "loss": 1.2323,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283000250658823e-05,
+      "loss": 1.2178,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279265904721437e-05,
+      "loss": 1.3311,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275522614419522e-05,
+      "loss": 1.2823,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271770387577167e-05,
+      "loss": 1.224,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26800923203714e-05,
+      "loss": 1.2668,
+      "step": 447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26423915566087e-05,
+      "loss": 1.2517,
+      "step": 448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260460166328431e-05,
+      "loss": 1.2555,
+      "step": 449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256672271938528e-05,
+      "loss": 1.2924,
+      "step": 450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25287548040848e-05,
+      "loss": 1.2952,
+      "step": 451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249069799674199e-05,
+      "loss": 1.2679,
+      "step": 452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245255237690182e-05,
+      "loss": 1.3093,
+      "step": 453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241431802429485e-05,
+      "loss": 1.2619,
+      "step": 454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237599501883711e-05,
+      "loss": 1.2533,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233758344062995e-05,
+      "loss": 1.2276,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229908336995987e-05,
+      "loss": 1.2507,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226049488729824e-05,
+      "loss": 1.1504,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222181807330134e-05,
+      "loss": 1.3116,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.218305300881002e-05,
+      "loss": 1.2425,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214419977484961e-05,
+      "loss": 1.2977,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210525845262966e-05,
+      "loss": 1.2141,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206622912354393e-05,
+      "loss": 1.2555,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20271118691701e-05,
+      "loss": 1.2811,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198790677126959e-05,
+      "loss": 1.2316,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194861391178749e-05,
+      "loss": 1.2074,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190923337285224e-05,
+      "loss": 1.3024,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186976523677566e-05,
+      "loss": 1.2598,
+      "step": 468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183020958605257e-05,
+      "loss": 1.2171,
+      "step": 469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179056650336073e-05,
+      "loss": 1.2823,
+      "step": 470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175083607156068e-05,
+      "loss": 1.3018,
+      "step": 471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171101837369549e-05,
+      "loss": 1.1846,
+      "step": 472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167111349299065e-05,
+      "loss": 1.3144,
+      "step": 473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163112151285389e-05,
+      "loss": 1.2472,
+      "step": 474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159104251687499e-05,
+      "loss": 1.2875,
+      "step": 475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155087658882555e-05,
+      "loss": 1.2509,
+      "step": 476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151062381265897e-05,
+      "loss": 1.2896,
+      "step": 477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14702842725101e-05,
+      "loss": 1.2363,
+      "step": 478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142985805269516e-05,
+      "loss": 1.3002,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138934523771156e-05,
+      "loss": 1.2118,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134874591223772e-05,
+      "loss": 1.2982,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.130806016113283e-05,
+      "loss": 1.2536,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126728806943676e-05,
+      "loss": 1.2386,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122642972236983e-05,
+      "loss": 1.192,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118548520533269e-05,
+      "loss": 1.248,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114445460390604e-05,
+      "loss": 1.3077,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110333800385055e-05,
+      "loss": 1.2046,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10621354911066e-05,
+      "loss": 1.3074,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102084715179422e-05,
+      "loss": 1.2731,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097947307221273e-05,
+      "loss": 1.2872,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093801333884075e-05,
+      "loss": 1.2345,
+      "step": 491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089646803833589e-05,
+      "loss": 1.2885,
+      "step": 492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085483725753459e-05,
+      "loss": 1.3096,
+      "step": 493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0813121083452e-05,
+      "loss": 1.2385,
+      "step": 494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07713196032817e-05,
+      "loss": 1.3076,
+      "step": 495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072943290439566e-05,
+      "loss": 1.2612,
+      "step": 496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068746107434389e-05,
+      "loss": 1.1861,
+      "step": 497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064540420085438e-05,
+      "loss": 1.2492,
+      "step": 498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060326237183285e-05,
+      "loss": 1.251,
+      "step": 499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056103567536262e-05,
+      "loss": 1.2457,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051872419970437e-05,
+      "loss": 1.3248,
+      "step": 501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047632803329601e-05,
+      "loss": 1.2717,
+      "step": 502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043384726475244e-05,
+      "loss": 1.2049,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039128198286538e-05,
+      "loss": 1.2738,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034863227660326e-05,
+      "loss": 1.2107,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030589823511091e-05,
+      "loss": 1.3407,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026307994770946e-05,
+      "loss": 1.2796,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02201775038961e-05,
+      "loss": 1.2465,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0177190993344e-05,
+      "loss": 1.2166,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013412050590193e-05,
+      "loss": 1.3067,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009096613159425e-05,
+      "loss": 1.2141,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004772796062068e-05,
+      "loss": 1.3135,
+      "step": 512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000440608335604e-05,
+      "loss": 1.1298,
+      "step": 513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996100059035012e-05,
+      "loss": 1.204,
+      "step": 514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99175115723275e-05,
+      "loss": 1.3311,
+      "step": 515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987393912018731e-05,
+      "loss": 1.251,
+      "step": 516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983028332500313e-05,
+      "loss": 1.2619,
+      "step": 517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978654427802266e-05,
+      "loss": 1.2364,
+      "step": 518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974272207066767e-05,
+      "loss": 1.2789,
+      "step": 519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969881679453372e-05,
+      "loss": 1.2328,
+      "step": 520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965482854139003e-05,
+      "loss": 1.2353,
+      "step": 521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961075740317919e-05,
+      "loss": 1.2671,
+      "step": 522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956660347201711e-05,
+      "loss": 1.2558,
+      "step": 523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95223668401927e-05,
+      "loss": 1.3361,
+      "step": 524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947804760016778e-05,
+      "loss": 1.2562,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943364584457673e-05,
+      "loss": 1.2799,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.938916166622654e-05,
+      "loss": 1.2263,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934459515809637e-05,
+      "loss": 1.249,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929994641333748e-05,
+      "loss": 1.0888,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92552155252731e-05,
+      "loss": 1.2764,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921040258739804e-05,
+      "loss": 1.3002,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916550769337866e-05,
+      "loss": 1.2654,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912053093705264e-05,
+      "loss": 1.2735,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907547241242873e-05,
+      "loss": 1.2656,
+      "step": 534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903033221368662e-05,
+      "loss": 1.349,
+      "step": 535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898511043517667e-05,
+      "loss": 1.2146,
+      "step": 536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893980717141983e-05,
+      "loss": 1.3122,
+      "step": 537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889442251710727e-05,
+      "loss": 1.2375,
+      "step": 538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884895656710034e-05,
+      "loss": 1.2698,
+      "step": 539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88034094164303e-05,
+      "loss": 1.2717,
+      "step": 540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875778116029815e-05,
+      "loss": 1.228,
+      "step": 541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87120718940744e-05,
+      "loss": 1.2276,
+      "step": 542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866628171329886e-05,
+      "loss": 1.2554,
+      "step": 543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862041071368047e-05,
+      "loss": 1.2699,
+      "step": 544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857445899109715e-05,
+      "loss": 1.2124,
+      "step": 545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852842664159548e-05,
+      "loss": 1.2951,
+      "step": 546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84823137613906e-05,
+      "loss": 1.2474,
+      "step": 547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843612044686593e-05,
+      "loss": 1.2467,
+      "step": 548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838984679457308e-05,
+      "loss": 1.1304,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834349290123151e-05,
+      "loss": 1.2475,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829705886372845e-05,
+      "loss": 1.2459,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82505447791186e-05,
+      "loss": 1.2968,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820395074462402e-05,
+      "loss": 1.2512,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815727685763381e-05,
+      "loss": 1.2205,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811052321570406e-05,
+      "loss": 1.3053,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806368991655746e-05,
+      "loss": 1.2053,
+      "step": 556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801677705808332e-05,
+      "loss": 1.311,
+      "step": 557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796978473833712e-05,
+      "loss": 1.2384,
+      "step": 558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.792271305554053e-05,
+      "loss": 1.2864,
+      "step": 559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.787556210808101e-05,
+      "loss": 1.2324,
+      "step": 560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782833199451176e-05,
+      "loss": 1.1957,
+      "step": 561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778102281355143e-05,
+      "loss": 1.2027,
+      "step": 562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773363466408392e-05,
+      "loss": 1.2683,
+      "step": 563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768616764515823e-05,
+      "loss": 1.3102,
+      "step": 564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763862185598813e-05,
+      "loss": 1.211,
+      "step": 565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759099739595215e-05,
+      "loss": 1.3311,
+      "step": 566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754329436459314e-05,
+      "loss": 1.2128,
+      "step": 567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749551286161823e-05,
+      "loss": 1.2582,
+      "step": 568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74476529868986e-05,
+      "loss": 1.2314,
+      "step": 569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739971484046913e-05,
+      "loss": 1.2122,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735169852252847e-05,
+      "loss": 1.273,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.73036041334385e-05,
+      "loss": 1.242,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725543177372435e-05,
+      "loss": 1.2529,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720718154407413e-05,
+      "loss": 1.2735,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71588535453387e-05,
+      "loss": 1.3101,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711044787853149e-05,
+      "loss": 1.2088,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706196464482821e-05,
+      "loss": 1.3562,
+      "step": 577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701340394556676e-05,
+      "loss": 1.1172,
+      "step": 578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696476588224691e-05,
+      "loss": 1.1608,
+      "step": 579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691605055653019e-05,
+      "loss": 1.2738,
+      "step": 580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686725807023956e-05,
+      "loss": 1.2698,
+      "step": 581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681838852535927e-05,
+      "loss": 1.2416,
+      "step": 582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676944202403468e-05,
+      "loss": 1.2698,
+      "step": 583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672041866857196e-05,
+      "loss": 1.2064,
+      "step": 584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667131856143793e-05,
+      "loss": 1.2015,
+      "step": 585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662214180525982e-05,
+      "loss": 1.3064,
+      "step": 586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657288850282508e-05,
+      "loss": 1.2618,
+      "step": 587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652355875708118e-05,
+      "loss": 1.2489,
+      "step": 588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647415267113532e-05,
+      "loss": 1.2412,
+      "step": 589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64246703482543e-05,
+      "loss": 1.1962,
+      "step": 590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637511189186425e-05,
+      "loss": 1.2517,
+      "step": 591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632547740555042e-05,
+      "loss": 1.2861,
+      "step": 592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627576699305702e-05,
+      "loss": 1.1816,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.622598075828692e-05,
+      "loss": 1.284,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617611880530149e-05,
+      "loss": 1.2267,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612618123832033e-05,
+      "loss": 1.2231,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607616816172111e-05,
+      "loss": 1.1587,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602607968003935e-05,
+      "loss": 1.2505,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597591589796815e-05,
+      "loss": 1.2513,
+      "step": 599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5925676920358e-05,
+      "loss": 1.2445,
+      "step": 600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587536285221656e-05,
+      "loss": 1.2366,
+      "step": 601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582497379870847e-05,
+      "loss": 1.2658,
+      "step": 602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577450986515506e-05,
+      "loss": 1.2666,
+      "step": 603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572397115703419e-05,
+      "loss": 1.2267,
+      "step": 604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567335777998002e-05,
+      "loss": 1.2559,
+      "step": 605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562266983978278e-05,
+      "loss": 1.3137,
+      "step": 606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557190744238855e-05,
+      "loss": 1.1975,
+      "step": 607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5521070693899e-05,
+      "loss": 1.3178,
+      "step": 608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547015970057126e-05,
+      "loss": 1.2007,
+      "step": 609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54191745688176e-05,
+      "loss": 1.2633,
+      "step": 610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536811540520529e-05,
+      "loss": 1.2038,
+      "step": 611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531698231645631e-05,
+      "loss": 1.2707,
+      "step": 612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526577540944717e-05,
+      "loss": 1.2392,
+      "step": 613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521449479120867e-05,
+      "loss": 1.236,
+      "step": 614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516314056892566e-05,
+      "loss": 1.317,
+      "step": 615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511171284993687e-05,
+      "loss": 1.2118,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506021174173462e-05,
+      "loss": 1.2989,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500863735196463e-05,
+      "loss": 1.255,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495698978842581e-05,
+      "loss": 1.1959,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490526915907e-05,
+      "loss": 1.1328,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485347557200176e-05,
+      "loss": 1.0961,
+      "step": 621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480160913547815e-05,
+      "loss": 1.3059,
+      "step": 622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474966995790847e-05,
+      "loss": 1.2827,
+      "step": 623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469765814785409e-05,
+      "loss": 1.1747,
+      "step": 624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46455738140282e-05,
+      "loss": 1.2757,
+      "step": 625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.459341706529557e-05,
+      "loss": 1.2411,
+      "step": 626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454118801067229e-05,
+      "loss": 1.1979,
+      "step": 627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448888675932563e-05,
+      "loss": 1.3501,
+      "step": 628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443651342057377e-05,
+      "loss": 1.2158,
+      "step": 629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438406810388548e-05,
+      "loss": 1.3089,
+      "step": 630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43315509188801e-05,
+      "loss": 1.204,
+      "step": 631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42789619753271e-05,
+      "loss": 1.218,
+      "step": 632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422630138314594e-05,
+      "loss": 1.2227,
+      "step": 633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417356925240587e-05,
+      "loss": 1.2578,
+      "step": 634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412076569332568e-05,
+      "loss": 1.3398,
+      "step": 635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406789081627337e-05,
+      "loss": 1.1444,
+      "step": 636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401494473176614e-05,
+      "loss": 1.3123,
+      "step": 637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39619275504699e-05,
+      "loss": 1.199,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.390883938319923e-05,
+      "loss": 1.3017,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385568034091708e-05,
+      "loss": 1.2154,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380245053473451e-05,
+      "loss": 1.2363,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374915007591053e-05,
+      "loss": 1.2602,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36957790758518e-05,
+      "loss": 1.2966,
+      "step": 643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364233764611239e-05,
+      "loss": 1.127,
+      "step": 644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358882589839365e-05,
+      "loss": 1.2064,
+      "step": 645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353524394454388e-05,
+      "loss": 1.3105,
+      "step": 646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348159189655808e-05,
+      "loss": 1.2129,
+      "step": 647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34278698665778e-05,
+      "loss": 1.2847,
+      "step": 648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337407796689088e-05,
+      "loss": 1.1962,
+      "step": 649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332021630993114e-05,
+      "loss": 1.2205,
+      "step": 650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326628500827826e-05,
+      "loss": 1.3116,
+      "step": 651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321228417465747e-05,
+      "loss": 1.1353,
+      "step": 652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315821392193932e-05,
+      "loss": 1.2675,
+      "step": 653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310407436313948e-05,
+      "loss": 1.24,
+      "step": 654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304986561141844e-05,
+      "loss": 1.2727,
+      "step": 655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299558778008138e-05,
+      "loss": 1.1865,
+      "step": 656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294124098257781e-05,
+      "loss": 1.2647,
+      "step": 657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288682533250139e-05,
+      "loss": 1.2592,
+      "step": 658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283234094358976e-05,
+      "loss": 1.2636,
+      "step": 659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277778792972418e-05,
+      "loss": 1.2634,
+      "step": 660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272316640492932e-05,
+      "loss": 1.152,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266847648337313e-05,
+      "loss": 1.3153,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261371827936645e-05,
+      "loss": 1.2627,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255889190736288e-05,
+      "loss": 1.2172,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250399748195849e-05,
+      "loss": 1.3006,
+      "step": 665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244903511789157e-05,
+      "loss": 1.2247,
+      "step": 666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239400493004248e-05,
+      "loss": 1.1545,
+      "step": 667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233890703343328e-05,
+      "loss": 1.2781,
+      "step": 668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228374154322755e-05,
+      "loss": 1.1834,
+      "step": 669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222850857473021e-05,
+      "loss": 1.2191,
+      "step": 670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217320824338713e-05,
+      "loss": 1.309,
+      "step": 671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21178406647851e-05,
+      "loss": 1.2126,
+      "step": 672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206240595465135e-05,
+      "loss": 1.3061,
+      "step": 673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20069042288535e-05,
+      "loss": 1.1184,
+      "step": 674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19513356033992e-05,
+      "loss": 1.2043,
+      "step": 675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189570019443598e-05,
+      "loss": 1.196,
+      "step": 676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18399981182509e-05,
+      "loss": 1.226,
+      "step": 677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.178422949127042e-05,
+      "loss": 1.2434,
+      "step": 678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172839443006007e-05,
+      "loss": 1.2916,
+      "step": 679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167249305132423e-05,
+      "loss": 1.2439,
+      "step": 680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161652547190592e-05,
+      "loss": 1.1896,
+      "step": 681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156049180878654e-05,
+      "loss": 1.2727,
+      "step": 682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150439217908556e-05,
+      "loss": 1.2181,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14482267000604e-05,
+      "loss": 1.226,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139199548910606e-05,
+      "loss": 1.2451,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133569866375495e-05,
+      "loss": 1.2958,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127933634167666e-05,
+      "loss": 1.2166,
+      "step": 687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122290864067761e-05,
+      "loss": 1.3195,
+      "step": 688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116641567870093e-05,
+      "loss": 1.2209,
+      "step": 689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110985757382615e-05,
+      "loss": 1.2722,
+      "step": 690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105323444426891e-05,
+      "loss": 1.2134,
+      "step": 691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099654640838082e-05,
+      "loss": 1.1875,
+      "step": 692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093979358464912e-05,
+      "loss": 1.1502,
+      "step": 693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088297609169647e-05,
+      "loss": 1.2242,
+      "step": 694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082609404828075e-05,
+      "loss": 1.2377,
+      "step": 695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076914757329467e-05,
+      "loss": 1.279,
+      "step": 696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07121367857657e-05,
+      "loss": 1.1578,
+      "step": 697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065506180485566e-05,
+      "loss": 1.2582,
+      "step": 698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059792274986061e-05,
+      "loss": 1.2166,
+      "step": 699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05407197402105e-05,
+      "loss": 1.2325,
+      "step": 700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048345289546895e-05,
+      "loss": 1.2505,
+      "step": 701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042612233533302e-05,
+      "loss": 1.217,
+      "step": 702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036872817963295e-05,
+      "loss": 1.2576,
+      "step": 703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03112705483319e-05,
+      "loss": 1.2586,
+      "step": 704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.025374956152569e-05,
+      "loss": 1.2725,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.019616533944259e-05,
+      "loss": 1.2198,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013851800244301e-05,
+      "loss": 1.2852,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00808076710193e-05,
+      "loss": 1.2499,
+      "step": 708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002303446579548e-05,
+      "loss": 1.1603,
+      "step": 709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996519850752702e-05,
+      "loss": 1.301,
+      "step": 710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990729991710045e-05,
+      "loss": 1.2273,
+      "step": 711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984933881553332e-05,
+      "loss": 1.182,
+      "step": 712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97913153239738e-05,
+      "loss": 1.3058,
+      "step": 713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973322956370042e-05,
+      "loss": 1.1075,
+      "step": 714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.967508165612197e-05,
+      "loss": 1.2414,
+      "step": 715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9616871722777e-05,
+      "loss": 1.2588,
+      "step": 716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955859988533386e-05,
+      "loss": 1.22,
+      "step": 717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950026626559012e-05,
+      "loss": 1.1852,
+      "step": 718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944187098547262e-05,
+      "loss": 1.2821,
+      "step": 719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938341416703703e-05,
+      "loss": 1.1673,
+      "step": 720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932489593246763e-05,
+      "loss": 1.3113,
+      "step": 721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92663164040771e-05,
+      "loss": 1.1992,
+      "step": 722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.920767570430622e-05,
+      "loss": 1.2053,
+      "step": 723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91489739557236e-05,
+      "loss": 1.2138,
+      "step": 724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909021128102552e-05,
+      "loss": 1.2293,
+      "step": 725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.903138780303556e-05,
+      "loss": 1.2342,
+      "step": 726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.897250364470435e-05,
+      "loss": 1.2233,
+      "step": 727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891355892910945e-05,
+      "loss": 1.2795,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885455377945492e-05,
+      "loss": 1.2142,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879548831907115e-05,
+      "loss": 1.2619,
+      "step": 730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873636267141461e-05,
+      "loss": 1.2549,
+      "step": 731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867717696006755e-05,
+      "loss": 1.2279,
+      "step": 732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861793130873779e-05,
+      "loss": 1.258,
+      "step": 733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855862584125839e-05,
+      "loss": 1.179,
+      "step": 734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849926068158748e-05,
+      "loss": 1.2558,
+      "step": 735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843983595380792e-05,
+      "loss": 1.2675,
+      "step": 736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838035178212713e-05,
+      "loss": 1.2432,
+      "step": 737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832080829087669e-05,
+      "loss": 1.1088,
+      "step": 738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826120560451227e-05,
+      "loss": 1.2787,
+      "step": 739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820154384761318e-05,
+      "loss": 1.1917,
+      "step": 740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814182314488225e-05,
+      "loss": 1.2146,
+      "step": 741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80820436211455e-05,
+      "loss": 1.221,
+      "step": 742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80222054013519e-05,
+      "loss": 1.2201,
+      "step": 743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79623086105731e-05,
+      "loss": 1.1985,
+      "step": 744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790235337400318e-05,
+      "loss": 1.2548,
+      "step": 745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784233981695835e-05,
+      "loss": 1.164,
+      "step": 746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778226806487677e-05,
+      "loss": 1.2339,
+      "step": 747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77221382433182e-05,
+      "loss": 1.327,
+      "step": 748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766195047796379e-05,
+      "loss": 1.1228,
+      "step": 749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76017048946158e-05,
+      "loss": 1.2556,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.754140161919731e-05,
+      "loss": 1.2773,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748104077775206e-05,
+      "loss": 1.2174,
+      "step": 752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742062249644404e-05,
+      "loss": 1.241,
+      "step": 753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.736014690155732e-05,
+      "loss": 1.2383,
+      "step": 754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729961411949578e-05,
+      "loss": 1.207,
+      "step": 755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72390242767828e-05,
+      "loss": 1.2305,
+      "step": 756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717837750006106e-05,
+      "loss": 1.2738,
+      "step": 757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711767391609224e-05,
+      "loss": 1.2144,
+      "step": 758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705691365175671e-05,
+      "loss": 1.2509,
+      "step": 759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699609683405337e-05,
+      "loss": 1.1921,
+      "step": 760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69352235900993e-05,
+      "loss": 1.2894,
+      "step": 761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68742940471295e-05,
+      "loss": 1.2195,
+      "step": 762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681330833249669e-05,
+      "loss": 1.2513,
+      "step": 763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675226657367095e-05,
+      "loss": 1.128,
+      "step": 764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669116889823955e-05,
+      "loss": 1.1583,
+      "step": 765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663001543390657e-05,
+      "loss": 1.2302,
+      "step": 766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656880630849277e-05,
+      "loss": 1.2358,
+      "step": 767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650754164993521e-05,
+      "loss": 1.1965,
+      "step": 768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644622158628702e-05,
+      "loss": 1.2676,
+      "step": 769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638484624571714e-05,
+      "loss": 1.2861,
+      "step": 770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632341575651009e-05,
+      "loss": 1.2075,
+      "step": 771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626193024706561e-05,
+      "loss": 1.2353,
+      "step": 772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.620038984589841e-05,
+      "loss": 1.238,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.613879468163804e-05,
+      "loss": 1.19,
+      "step": 774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607714488302842e-05,
+      "loss": 1.1818,
+      "step": 775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60154405789277e-05,
+      "loss": 1.2151,
+      "step": 776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595368189830794e-05,
+      "loss": 1.2382,
+      "step": 777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.58918689702549e-05,
+      "loss": 1.2173,
+      "step": 778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583000192396767e-05,
+      "loss": 1.235,
+      "step": 779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576808088875848e-05,
+      "loss": 1.202,
+      "step": 780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570610599405241e-05,
+      "loss": 1.2455,
+      "step": 781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564407736938712e-05,
+      "loss": 1.2159,
+      "step": 782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558199514441253e-05,
+      "loss": 1.2005,
+      "step": 783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551985944889068e-05,
+      "loss": 1.275,
+      "step": 784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545767041269527e-05,
+      "loss": 1.2504,
+      "step": 785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539542816581157e-05,
+      "loss": 1.2273,
+      "step": 786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.533313283833603e-05,
+      "loss": 1.2063,
+      "step": 787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527078456047605e-05,
+      "loss": 1.2189,
+      "step": 788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.520838346254975e-05,
+      "loss": 1.136,
+      "step": 789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514592967498555e-05,
+      "loss": 1.231,
+      "step": 790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508342332832212e-05,
+      "loss": 1.1887,
+      "step": 791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502086455320791e-05,
+      "loss": 1.275,
+      "step": 792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.495825348040097e-05,
+      "loss": 1.1705,
+      "step": 793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489559024076869e-05,
+      "loss": 1.1937,
+      "step": 794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483287496528744e-05,
+      "loss": 1.2332,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.477010778504241e-05,
+      "loss": 1.2341,
+      "step": 796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470728883122725e-05,
+      "loss": 1.2136,
+      "step": 797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464441823514384e-05,
+      "loss": 1.199,
+      "step": 798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.458149612820198e-05,
+      "loss": 1.2372,
+      "step": 799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.451852264191914e-05,
+      "loss": 1.1502,
+      "step": 800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445549790792021e-05,
+      "loss": 1.2739,
+      "step": 801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439242205793717e-05,
+      "loss": 1.2158,
+      "step": 802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.432929522380885e-05,
+      "loss": 1.2614,
+      "step": 803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426611753748063e-05,
+      "loss": 1.2153,
+      "step": 804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420288913100419e-05,
+      "loss": 1.2208,
+      "step": 805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413961013653726e-05,
+      "loss": 1.2335,
+      "step": 806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40762806863432e-05,
+      "loss": 1.2172,
+      "step": 807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401290091279098e-05,
+      "loss": 1.2246,
+      "step": 808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394947094835463e-05,
+      "loss": 1.1157,
+      "step": 809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388599092561315e-05,
+      "loss": 1.2885,
+      "step": 810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382246097725015e-05,
+      "loss": 1.1935,
+      "step": 811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375888123605359e-05,
+      "loss": 1.2304,
+      "step": 812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369525183491553e-05,
+      "loss": 1.2979,
+      "step": 813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363157290683177e-05,
+      "loss": 1.2339,
+      "step": 814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356784458490171e-05,
+      "loss": 1.256,
+      "step": 815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.350406700232794e-05,
+      "loss": 1.1333,
+      "step": 816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344024029241601e-05,
+      "loss": 1.2615,
+      "step": 817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.337636458857416e-05,
+      "loss": 1.1877,
+      "step": 818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331244002431308e-05,
+      "loss": 1.263,
+      "step": 819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.32484667332455e-05,
+      "loss": 1.2057,
+      "step": 820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318444484908607e-05,
+      "loss": 1.246,
+      "step": 821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312037450565098e-05,
+      "loss": 1.1588,
+      "step": 822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305625583685771e-05,
+      "loss": 1.2098,
+      "step": 823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.299208897672473e-05,
+      "loss": 1.2908,
+      "step": 824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29278740593713e-05,
+      "loss": 1.1541,
+      "step": 825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.286361121901706e-05,
+      "loss": 1.2548,
+      "step": 826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27993005899818e-05,
+      "loss": 1.2188,
+      "step": 827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27349423066853e-05,
+      "loss": 1.2523,
+      "step": 828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267053650364685e-05,
+      "loss": 1.1152,
+      "step": 829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260608331548512e-05,
+      "loss": 1.3001,
+      "step": 830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254158287691775e-05,
+      "loss": 1.1783,
+      "step": 831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247703532276123e-05,
+      "loss": 1.2074,
+      "step": 832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241244078793047e-05,
+      "loss": 1.1984,
+      "step": 833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23477994074386e-05,
+      "loss": 1.2442,
+      "step": 834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228311131639666e-05,
+      "loss": 1.2877,
+      "step": 835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221837665001335e-05,
+      "loss": 1.1665,
+      "step": 836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215359554359465e-05,
+      "loss": 1.3412,
+      "step": 837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208876813254366e-05,
+      "loss": 1.1857,
+      "step": 838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20238945523603e-05,
+      "loss": 1.2212,
+      "step": 839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195897493864088e-05,
+      "loss": 1.1962,
+      "step": 840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189400942707804e-05,
+      "loss": 1.143,
+      "step": 841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18289981534603e-05,
+      "loss": 1.1683,
+      "step": 842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176394125367182e-05,
+      "loss": 1.2408,
+      "step": 843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169883886369215e-05,
+      "loss": 1.2334,
+      "step": 844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163369111959593e-05,
+      "loss": 1.2049,
+      "step": 845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156849815755258e-05,
+      "loss": 1.2922,
+      "step": 846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.150326011382604e-05,
+      "loss": 1.135,
+      "step": 847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143797712477444e-05,
+      "loss": 1.2314,
+      "step": 848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137264932684993e-05,
+      "loss": 1.2223,
+      "step": 849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.130727685659826e-05,
+      "loss": 1.2481,
+      "step": 850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124185985065856e-05,
+      "loss": 1.2128,
+      "step": 851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117639844576308e-05,
+      "loss": 1.2061,
+      "step": 852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11108927787368e-05,
+      "loss": 1.2043,
+      "step": 853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104534298649733e-05,
+      "loss": 1.2446,
+      "step": 854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097974920605435e-05,
+      "loss": 1.2256,
+      "step": 855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091411157450965e-05,
+      "loss": 1.2108,
+      "step": 856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084843022905656e-05,
+      "loss": 1.244,
+      "step": 857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078270530697983e-05,
+      "loss": 1.1022,
+      "step": 858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071693694565526e-05,
+      "loss": 1.1966,
+      "step": 859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06511252825495e-05,
+      "loss": 1.1934,
+      "step": 860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058527045521965e-05,
+      "loss": 1.2495,
+      "step": 861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.051937260131305e-05,
+      "loss": 1.2265,
+      "step": 862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.045343185856701e-05,
+      "loss": 1.2138,
+      "step": 863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038744836480839e-05,
+      "loss": 1.2123,
+      "step": 864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03214222579535e-05,
+      "loss": 1.2269,
+      "step": 865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025535367600771e-05,
+      "loss": 1.2645,
+      "step": 866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018924275706512e-05,
+      "loss": 1.2215,
+      "step": 867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01230896393083e-05,
+      "loss": 1.2435,
+      "step": 868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005689446100816e-05,
+      "loss": 1.183,
+      "step": 869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999065736052337e-05,
+      "loss": 1.2681,
+      "step": 870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99243784763003e-05,
+      "loss": 1.2232,
+      "step": 871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985805794687265e-05,
+      "loss": 1.2828,
+      "step": 872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979169591086115e-05,
+      "loss": 1.1117,
+      "step": 873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972529250697329e-05,
+      "loss": 1.2278,
+      "step": 874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965884787400301e-05,
+      "loss": 1.2766,
+      "step": 875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959236215083049e-05,
+      "loss": 1.1825,
+      "step": 876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.952583547642171e-05,
+      "loss": 1.1626,
+      "step": 877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945926798982829e-05,
+      "loss": 1.238,
+      "step": 878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939265983018717e-05,
+      "loss": 1.1961,
+      "step": 879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932601113672024e-05,
+      "loss": 1.1985,
+      "step": 880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925932204873419e-05,
+      "loss": 1.1844,
+      "step": 881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919259270562008e-05,
+      "loss": 1.2078,
+      "step": 882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.912582324685315e-05,
+      "loss": 1.2148,
+      "step": 883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905901381199245e-05,
+      "loss": 1.2474,
+      "step": 884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899216454068062e-05,
+      "loss": 1.2096,
+      "step": 885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892527557264358e-05,
+      "loss": 1.2789,
+      "step": 886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885834704769017e-05,
+      "loss": 1.1669,
+      "step": 887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879137910571191e-05,
+      "loss": 1.2017,
+      "step": 888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872437188668279e-05,
+      "loss": 1.1522,
+      "step": 889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.865732553065879e-05,
+      "loss": 1.2318,
+      "step": 890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859024017777779e-05,
+      "loss": 1.2178,
+      "step": 891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852311596825908e-05,
+      "loss": 1.2803,
+      "step": 892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845595304240326e-05,
+      "loss": 1.1329,
+      "step": 893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83887515405918e-05,
+      "loss": 1.2257,
+      "step": 894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832151160328682e-05,
+      "loss": 1.2738,
+      "step": 895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825423337103074e-05,
+      "loss": 1.1487,
+      "step": 896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818691698444609e-05,
+      "loss": 1.2823,
+      "step": 897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.811956258423507e-05,
+      "loss": 1.2046,
+      "step": 898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80521703111794e-05,
+      "loss": 1.1911,
+      "step": 899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798474030613995e-05,
+      "loss": 1.2112,
+      "step": 900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791727271005642e-05,
+      "loss": 1.0871,
+      "step": 901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78497676639471e-05,
+      "loss": 1.2353,
+      "step": 902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778222530890857e-05,
+      "loss": 1.277,
+      "step": 903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.771464578611539e-05,
+      "loss": 1.0917,
+      "step": 904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764702923681976e-05,
+      "loss": 1.2775,
+      "step": 905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757937580235138e-05,
+      "loss": 1.2354,
+      "step": 906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.75116856241169e-05,
+      "loss": 1.1864,
+      "step": 907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.744395884359988e-05,
+      "loss": 1.2912,
+      "step": 908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737619560236035e-05,
+      "loss": 1.2235,
+      "step": 909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730839604203454e-05,
+      "loss": 1.2414,
+      "step": 910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.724056030433463e-05,
+      "loss": 1.1955,
+      "step": 911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717268853104834e-05,
+      "loss": 1.2375,
+      "step": 912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710478086403883e-05,
+      "loss": 1.1582,
+      "step": 913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703683744524415e-05,
+      "loss": 1.1782,
+      "step": 914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696885841667718e-05,
+      "loss": 1.2761,
+      "step": 915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690084392042513e-05,
+      "loss": 1.1824,
+      "step": 916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683279409864948e-05,
+      "loss": 1.2593,
+      "step": 917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676470909358544e-05,
+      "loss": 1.2289,
+      "step": 918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669658904754177e-05,
+      "loss": 1.2089,
+      "step": 919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662843410290052e-05,
+      "loss": 1.244,
+      "step": 920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656024440211661e-05,
+      "loss": 1.1368,
+      "step": 921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64920200877177e-05,
+      "loss": 1.2755,
+      "step": 922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642376130230373e-05,
+      "loss": 1.2585,
+      "step": 923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63554681885467e-05,
+      "loss": 1.065,
+      "step": 924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628714088919037e-05,
+      "loss": 1.2056,
+      "step": 925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621877954704995e-05,
+      "loss": 1.2755,
+      "step": 926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615038430501183e-05,
+      "loss": 1.2018,
+      "step": 927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608195530603322e-05,
+      "loss": 1.3003,
+      "step": 928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601349269314188e-05,
+      "loss": 1.095,
+      "step": 929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.594499660943589e-05,
+      "loss": 1.2,
+      "step": 930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587646719808323e-05,
+      "loss": 1.2405,
+      "step": 931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.580790460232157e-05,
+      "loss": 1.183,
+      "step": 932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.573930896545792e-05,
+      "loss": 1.2592,
+      "step": 933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567068043086836e-05,
+      "loss": 1.2509,
+      "step": 934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560201914199773e-05,
+      "loss": 1.2556,
+      "step": 935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.553332524235937e-05,
+      "loss": 1.17,
+      "step": 936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.546459887553472e-05,
+      "loss": 1.3079,
+      "step": 937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.53958401851731e-05,
+      "loss": 1.1576,
+      "step": 938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.532704931499141e-05,
+      "loss": 1.154,
+      "step": 939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525822640877383e-05,
+      "loss": 1.2082,
+      "step": 940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518937161037144e-05,
+      "loss": 1.2194,
+      "step": 941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512048506370201e-05,
+      "loss": 1.2361,
+      "step": 942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505156691274969e-05,
+      "loss": 1.3108,
+      "step": 943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.498261730156466e-05,
+      "loss": 1.1649,
+      "step": 944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491363637426289e-05,
+      "loss": 1.2171,
+      "step": 945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484462427502572e-05,
+      "loss": 1.1907,
+      "step": 946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477558114809978e-05,
+      "loss": 1.1869,
+      "step": 947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.470650713779644e-05,
+      "loss": 1.2777,
+      "step": 948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463740238849165e-05,
+      "loss": 1.2272,
+      "step": 949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456826704462567e-05,
+      "loss": 1.1714,
+      "step": 950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44991012507026e-05,
+      "loss": 1.2449,
+      "step": 951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44299051512903e-05,
+      "loss": 1.2063,
+      "step": 952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436067889101986e-05,
+      "loss": 1.2228,
+      "step": 953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429142261458551e-05,
+      "loss": 1.0999,
+      "step": 954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.422213646674415e-05,
+      "loss": 1.1779,
+      "step": 955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.415282059231517e-05,
+      "loss": 1.2108,
+      "step": 956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408347513618005e-05,
+      "loss": 1.2503,
+      "step": 957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401410024328214e-05,
+      "loss": 1.1869,
+      "step": 958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394469605862625e-05,
+      "loss": 1.236,
+      "step": 959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.38752627272785e-05,
+      "loss": 1.1964,
+      "step": 960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.380580039436586e-05,
+      "loss": 1.1323,
+      "step": 961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373630920507597e-05,
+      "loss": 1.2792,
+      "step": 962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366678930465676e-05,
+      "loss": 1.266,
+      "step": 963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359724083841617e-05,
+      "loss": 1.1774,
+      "step": 964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.352766395172187e-05,
+      "loss": 1.2347,
+      "step": 965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345805879000088e-05,
+      "loss": 1.1991,
+      "step": 966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.338842549873936e-05,
+      "loss": 1.1706,
+      "step": 967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331876422348229e-05,
+      "loss": 1.2513,
+      "step": 968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.32490751098331e-05,
+      "loss": 1.228,
+      "step": 969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317935830345338e-05,
+      "loss": 1.1585,
+      "step": 970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31096139500627e-05,
+      "loss": 1.1743,
+      "step": 971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.303984219543811e-05,
+      "loss": 1.1424,
+      "step": 972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.297004318541396e-05,
+      "loss": 1.1377,
+      "step": 973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.290021706588161e-05,
+      "loss": 1.183,
+      "step": 974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.283036398278902e-05,
+      "loss": 1.2108,
+      "step": 975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.27604840821406e-05,
+      "loss": 1.2931,
+      "step": 976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.269057750999669e-05,
+      "loss": 1.1502,
+      "step": 977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.262064441247346e-05,
+      "loss": 1.1601,
+      "step": 978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.255068493574253e-05,
+      "loss": 1.2599,
+      "step": 979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248069922603058e-05,
+      "loss": 1.1128,
+      "step": 980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24106874296192e-05,
+      "loss": 1.2917,
+      "step": 981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.23406496928445e-05,
+      "loss": 1.1802,
+      "step": 982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227058616209675e-05,
+      "loss": 1.2115,
+      "step": 983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220049698382018e-05,
+      "loss": 1.1953,
+      "step": 984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213038230451261e-05,
+      "loss": 1.2582,
+      "step": 985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20602422707252e-05,
+      "loss": 1.1294,
+      "step": 986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.199007702906205e-05,
+      "loss": 1.2243,
+      "step": 987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.191988672617998e-05,
+      "loss": 1.2321,
+      "step": 988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184967150878819e-05,
+      "loss": 1.1785,
+      "step": 989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177943152364793e-05,
+      "loss": 1.2143,
+      "step": 990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.170916691757225e-05,
+      "loss": 1.1847,
+      "step": 991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163887783742566e-05,
+      "loss": 1.126,
+      "step": 992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.156856443012382e-05,
+      "loss": 1.2874,
+      "step": 993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14982268426332e-05,
+      "loss": 1.0989,
+      "step": 994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.142786522197089e-05,
+      "loss": 1.1971,
+      "step": 995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.135747971520411e-05,
+      "loss": 1.1959,
+      "step": 996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.12870704694501e-05,
+      "loss": 1.2087,
+      "step": 997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121663763187569e-05,
+      "loss": 1.203,
+      "step": 998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.114618134969698e-05,
+      "loss": 1.2776,
+      "step": 999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107570177017915e-05,
+      "loss": 1.157,
+      "step": 1000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.100519904063597e-05,
+      "loss": 1.2804,
+      "step": 1001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0934673308429725e-05,
+      "loss": 1.0957,
+      "step": 1002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.086412472097066e-05,
+      "loss": 1.156,
+      "step": 1003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.079355342571685e-05,
+      "loss": 1.272,
+      "step": 1004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0722959570173854e-05,
+      "loss": 1.2304,
+      "step": 1005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0652343301894346e-05,
+      "loss": 1.177,
+      "step": 1006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.058170476847782e-05,
+      "loss": 1.1873,
+      "step": 1007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051104411757041e-05,
+      "loss": 1.2484,
+      "step": 1008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044036149686436e-05,
+      "loss": 1.1397,
+      "step": 1009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.036965705409793e-05,
+      "loss": 1.3051,
+      "step": 1010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029893093705492e-05,
+      "loss": 1.1561,
+      "step": 1011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.022818329356449e-05,
+      "loss": 1.2185,
+      "step": 1012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.015741427150076e-05,
+      "loss": 1.2409,
+      "step": 1013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.008662401878251e-05,
+      "loss": 1.1891,
+      "step": 1014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001581268337298e-05,
+      "loss": 1.192,
+      "step": 1015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99449804132794e-05,
+      "loss": 1.2205,
+      "step": 1016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.987412735655277e-05,
+      "loss": 1.1923,
+      "step": 1017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.980325366128755e-05,
+      "loss": 1.182,
+      "step": 1018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.973235947562137e-05,
+      "loss": 1.1199,
+      "step": 1019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9661444947734626e-05,
+      "loss": 1.1747,
+      "step": 1020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.959051022585025e-05,
+      "loss": 1.1888,
+      "step": 1021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.951955545823342e-05,
+      "loss": 1.2213,
+      "step": 1022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944858079319118e-05,
+      "loss": 1.1583,
+      "step": 1023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937758637907216e-05,
+      "loss": 1.2138,
+      "step": 1024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.930657236426629e-05,
+      "loss": 1.2002,
+      "step": 1025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.923553889720447e-05,
+      "loss": 1.1629,
+      "step": 1026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.916448612635821e-05,
+      "loss": 1.165,
+      "step": 1027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.909341420023942e-05,
+      "loss": 1.2885,
+      "step": 1028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9022323267400036e-05,
+      "loss": 1.1339,
+      "step": 1029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.895121347643172e-05,
+      "loss": 1.247,
+      "step": 1030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.888008497596552e-05,
+      "loss": 1.202,
+      "step": 1031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.880893791467167e-05,
+      "loss": 1.222,
+      "step": 1032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.87377724412591e-05,
+      "loss": 1.2406,
+      "step": 1033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8666588704475275e-05,
+      "loss": 1.2307,
+      "step": 1034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.859538685310585e-05,
+      "loss": 1.2027,
+      "step": 1035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.852416703597431e-05,
+      "loss": 1.2057,
+      "step": 1036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84529294019417e-05,
+      "loss": 1.1709,
+      "step": 1037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.83816740999063e-05,
+      "loss": 1.1851,
+      "step": 1038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.831040127880336e-05,
+      "loss": 1.2719,
+      "step": 1039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.823911108760468e-05,
+      "loss": 1.1841,
+      "step": 1040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8167803675318414e-05,
+      "loss": 1.2695,
+      "step": 1041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8096479190988695e-05,
+      "loss": 1.2046,
+      "step": 1042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802513778369535e-05,
+      "loss": 1.171,
+      "step": 1043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795377960255356e-05,
+      "loss": 1.0482,
+      "step": 1044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7882404796713584e-05,
+      "loss": 1.1943,
+      "step": 1045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7811013515360413e-05,
+      "loss": 1.2039,
+      "step": 1046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.773960590771348e-05,
+      "loss": 1.2089,
+      "step": 1047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766818212302636e-05,
+      "loss": 1.2128,
+      "step": 1048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.759674231058643e-05,
+      "loss": 1.2319,
+      "step": 1049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7525286619714544e-05,
+      "loss": 1.226,
+      "step": 1050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.745381519976477e-05,
+      "loss": 1.1516,
+      "step": 1051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738232820012407e-05,
+      "loss": 1.1554,
+      "step": 1052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7310825770211906e-05,
+      "loss": 1.2002,
+      "step": 1053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723930805948008e-05,
+      "loss": 1.2194,
+      "step": 1054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7167775217412226e-05,
+      "loss": 1.236,
+      "step": 1055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.709622739352371e-05,
+      "loss": 1.1789,
+      "step": 1056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7024664737361166e-05,
+      "loss": 1.2028,
+      "step": 1057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6953087398502215e-05,
+      "loss": 1.2034,
+      "step": 1058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.688149552655519e-05,
+      "loss": 1.2585,
+      "step": 1059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.680988927115879e-05,
+      "loss": 1.15,
+      "step": 1060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6738268781981804e-05,
+      "loss": 1.2307,
+      "step": 1061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6666634208722705e-05,
+      "loss": 1.1864,
+      "step": 1062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6594985701109494e-05,
+      "loss": 1.1933,
+      "step": 1063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.652332340889923e-05,
+      "loss": 1.2181,
+      "step": 1064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.64516474818778e-05,
+      "loss": 1.2022,
+      "step": 1065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.637995806985961e-05,
+      "loss": 1.1943,
+      "step": 1066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6308255322687245e-05,
+      "loss": 1.2131,
+      "step": 1067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.623653939023111e-05,
+      "loss": 1.1536,
+      "step": 1068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.616481042238928e-05,
+      "loss": 1.0679,
+      "step": 1069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.609306856908697e-05,
+      "loss": 1.2399,
+      "step": 1070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.602131398027637e-05,
+      "loss": 1.1378,
+      "step": 1071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.59495468059363e-05,
+      "loss": 1.2511,
+      "step": 1072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.587776719607187e-05,
+      "loss": 1.1217,
+      "step": 1073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5805975300714185e-05,
+      "loss": 1.1573,
+      "step": 1074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.573417126992003e-05,
+      "loss": 1.24,
+      "step": 1075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5662355253771546e-05,
+      "loss": 1.1526,
+      "step": 1076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5590527402375944e-05,
+      "loss": 1.2239,
+      "step": 1077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551868786586517e-05,
+      "loss": 1.1151,
+      "step": 1078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.544683679439556e-05,
+      "loss": 1.2045,
+      "step": 1079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.537497433814762e-05,
+      "loss": 1.212,
+      "step": 1080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.530310064732559e-05,
+      "loss": 1.2442,
+      "step": 1081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.523121587215724e-05,
+      "loss": 1.2162,
+      "step": 1082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.515932016289347e-05,
+      "loss": 1.1891,
+      "step": 1083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.508741366980809e-05,
+      "loss": 1.1703,
+      "step": 1084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.501549654319734e-05,
+      "loss": 1.2043,
+      "step": 1085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494356893337985e-05,
+      "loss": 1.2063,
+      "step": 1086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4871630990696e-05,
+      "loss": 1.1751,
+      "step": 1087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.479968286550786e-05,
+      "loss": 1.1798,
+      "step": 1088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4727724708198765e-05,
+      "loss": 1.1584,
+      "step": 1089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.465575666917302e-05,
+      "loss": 1.2298,
+      "step": 1090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.458377889885557e-05,
+      "loss": 1.1438,
+      "step": 1091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.451179154769169e-05,
+      "loss": 1.2737,
+      "step": 1092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4439794766146746e-05,
+      "loss": 1.2296,
+      "step": 1093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.436778870470572e-05,
+      "loss": 1.1618,
+      "step": 1094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4295773513873085e-05,
+      "loss": 1.2002,
+      "step": 1095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.422374934417228e-05,
+      "loss": 1.1802,
+      "step": 1096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.415171634614567e-05,
+      "loss": 1.2215,
+      "step": 1097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.407967467035391e-05,
+      "loss": 1.208,
+      "step": 1098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4007624467375894e-05,
+      "loss": 1.223,
+      "step": 1099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.393556588780832e-05,
+      "loss": 1.1587,
+      "step": 1100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.386349908226538e-05,
+      "loss": 1.1087,
+      "step": 1101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379142420137845e-05,
+      "loss": 1.1702,
+      "step": 1102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.371934139579583e-05,
+      "loss": 1.2702,
+      "step": 1103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3647250816182334e-05,
+      "loss": 1.2021,
+      "step": 1104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357515261321906e-05,
+      "loss": 1.2007,
+      "step": 1105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350304693760301e-05,
+      "loss": 1.2471,
+      "step": 1106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343093394004682e-05,
+      "loss": 1.1734,
+      "step": 1107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.335881377127845e-05,
+      "loss": 1.2192,
+      "step": 1108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.32866865820408e-05,
+      "loss": 1.0343,
+      "step": 1109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.321455252309151e-05,
+      "loss": 1.2839,
+      "step": 1110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3142411745202516e-05,
+      "loss": 1.1759,
+      "step": 1111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.307026439915982e-05,
+      "loss": 1.2758,
+      "step": 1112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2998110635763175e-05,
+      "loss": 1.1824,
+      "step": 1113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2925950605825724e-05,
+      "loss": 1.2128,
+      "step": 1114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.285378446017373e-05,
+      "loss": 1.2096,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.278161234964617e-05,
+      "loss": 1.1645,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27094344250946e-05,
+      "loss": 1.23,
+      "step": 1117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.263725083738262e-05,
+      "loss": 1.2247,
+      "step": 1118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256506173738575e-05,
+      "loss": 1.1231,
+      "step": 1119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.249286727599098e-05,
+      "loss": 1.1809,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2420667604096526e-05,
+      "loss": 1.1474,
+      "step": 1121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2348462872611504e-05,
+      "loss": 1.1812,
+      "step": 1122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227625323245557e-05,
+      "loss": 1.2388,
+      "step": 1123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220403883455869e-05,
+      "loss": 1.1697,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2131819829860716e-05,
+      "loss": 1.1783,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2059596369311204e-05,
+      "loss": 1.236,
+      "step": 1126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.198736860386893e-05,
+      "loss": 1.179,
+      "step": 1127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.191513668450178e-05,
+      "loss": 1.2044,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.18429007621862e-05,
+      "loss": 1.2052,
+      "step": 1129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1770660987907096e-05,
+      "loss": 1.2311,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.169841751265738e-05,
+      "loss": 1.1528,
+      "step": 1131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.162617048743773e-05,
+      "loss": 1.2421,
+      "step": 1132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1553920063256186e-05,
+      "loss": 1.1776,
+      "step": 1133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.148166639112799e-05,
+      "loss": 1.1859,
+      "step": 1134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140940962207508e-05,
+      "loss": 1.1563,
+      "step": 1135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1337149907125904e-05,
+      "loss": 1.1681,
+      "step": 1136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.126488739731508e-05,
+      "loss": 1.2731,
+      "step": 1137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.119262224368303e-05,
+      "loss": 1.0836,
+      "step": 1138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1120354597275756e-05,
+      "loss": 1.197,
+      "step": 1139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10480846091444e-05,
+      "loss": 1.1893,
+      "step": 1140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.097581243034509e-05,
+      "loss": 1.1909,
+      "step": 1141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090353821193844e-05,
+      "loss": 1.1912,
+      "step": 1142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0831262104989387e-05,
+      "loss": 1.2243,
+      "step": 1143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.075898426056679e-05,
+      "loss": 1.206,
+      "step": 1144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0686704829743156e-05,
+      "loss": 1.1853,
+      "step": 1145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0614423963594294e-05,
+      "loss": 1.2859,
+      "step": 1146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0542141813199006e-05,
+      "loss": 1.217,
+      "step": 1147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.046985852963879e-05,
+      "loss": 1.1578,
+      "step": 1148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039757426399754e-05,
+      "loss": 1.1914,
+      "step": 1149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032528916736115e-05,
+      "loss": 1.1357,
+      "step": 1150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0253003390817264e-05,
+      "loss": 1.1694,
+      "step": 1151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.018071708545501e-05,
+      "loss": 1.2733,
+      "step": 1152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.01084304023645e-05,
+      "loss": 1.1972,
+      "step": 1153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.003614349263678e-05,
+      "loss": 1.167,
+      "step": 1154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.996385650736323e-05,
+      "loss": 1.2606,
+      "step": 1155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9891569597635504e-05,
+      "loss": 1.1255,
+      "step": 1156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9819282914545014e-05,
+      "loss": 1.0719,
+      "step": 1157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974699660918273e-05,
+      "loss": 1.2046,
+      "step": 1158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967471083263886e-05,
+      "loss": 1.1702,
+      "step": 1159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.960242573600248e-05,
+      "loss": 1.1845,
+      "step": 1160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.953014147036121e-05,
+      "loss": 1.2474,
+      "step": 1161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945785818680101e-05,
+      "loss": 1.1932,
+      "step": 1162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.938557603640573e-05,
+      "loss": 1.246,
+      "step": 1163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.931329517025686e-05,
+      "loss": 1.16,
+      "step": 1164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.924101573943321e-05,
+      "loss": 1.1763,
+      "step": 1165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.916873789501061e-05,
+      "loss": 1.2274,
+      "step": 1166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909646178806157e-05,
+      "loss": 1.1606,
+      "step": 1167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.902418756965492e-05,
+      "loss": 1.2481,
+      "step": 1168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8951915390855604e-05,
+      "loss": 1.1799,
+      "step": 1169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.887964540272427e-05,
+      "loss": 1.2095,
+      "step": 1170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.880737775631698e-05,
+      "loss": 1.1548,
+      "step": 1171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.873511260268493e-05,
+      "loss": 1.2461,
+      "step": 1172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866285009287411e-05,
+      "loss": 1.1871,
+      "step": 1173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8590590377924927e-05,
+      "loss": 1.1548,
+      "step": 1174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.851833360887201e-05,
+      "loss": 1.2472,
+      "step": 1175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844607993674382e-05,
+      "loss": 1.0791,
+      "step": 1176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.837382951256229e-05,
+      "loss": 1.2697,
+      "step": 1177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8301582487342645e-05,
+      "loss": 1.1878,
+      "step": 1178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82293390120929e-05,
+      "loss": 1.1557,
+      "step": 1179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.815709923781381e-05,
+      "loss": 1.1954,
+      "step": 1180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8084863315498234e-05,
+      "loss": 1.0732,
+      "step": 1181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.801263139613107e-05,
+      "loss": 1.1352,
+      "step": 1182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7940403630688814e-05,
+      "loss": 1.2508,
+      "step": 1183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.786818017013929e-05,
+      "loss": 1.1363,
+      "step": 1184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779596116544134e-05,
+      "loss": 1.2088,
+      "step": 1185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772374676754444e-05,
+      "loss": 1.2189,
+      "step": 1186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.765153712738851e-05,
+      "loss": 1.1654,
+      "step": 1187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757933239590348e-05,
+      "loss": 1.2619,
+      "step": 1188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.750713272400904e-05,
+      "loss": 1.2518,
+      "step": 1189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7434938262614264e-05,
+      "loss": 1.196,
+      "step": 1190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7362749162617404e-05,
+      "loss": 1.1252,
+      "step": 1191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729056557490541e-05,
+      "loss": 1.179,
+      "step": 1192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7218387650353834e-05,
+      "loss": 1.1044,
+      "step": 1193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.714621553982629e-05,
+      "loss": 1.2054,
+      "step": 1194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707404939417429e-05,
+      "loss": 1.2961,
+      "step": 1195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.700188936423683e-05,
+      "loss": 1.1228,
+      "step": 1196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6929735600840187e-05,
+      "loss": 1.2655,
+      "step": 1197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6857588254797516e-05,
+      "loss": 1.1653,
+      "step": 1198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.67854474769085e-05,
+      "loss": 1.1834,
+      "step": 1199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.671331341795921e-05,
+      "loss": 1.1967,
+      "step": 1200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664118622872156e-05,
+      "loss": 1.1241,
+      "step": 1201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.656906605995319e-05,
+      "loss": 1.2272,
+      "step": 1202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649695306239701e-05,
+      "loss": 1.2068,
+      "step": 1203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6424847386780965e-05,
+      "loss": 1.166,
+      "step": 1204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635274918381768e-05,
+      "loss": 1.0583,
+      "step": 1205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.628065860420418e-05,
+      "loss": 1.2392,
+      "step": 1206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6208575798621547e-05,
+      "loss": 1.0905,
+      "step": 1207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.613650091773464e-05,
+      "loss": 1.2539,
+      "step": 1208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606443411219169e-05,
+      "loss": 1.1519,
+      "step": 1209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.599237553262412e-05,
+      "loss": 1.1905,
+      "step": 1210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.592032532964611e-05,
+      "loss": 1.1915,
+      "step": 1211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.584828365385436e-05,
+      "loss": 1.1727,
+      "step": 1212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5776250655827715e-05,
+      "loss": 1.2072,
+      "step": 1213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.570422648612693e-05,
+      "loss": 1.222,
+      "step": 1214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.563221129529428e-05,
+      "loss": 1.1619,
+      "step": 1215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5560205233853266e-05,
+      "loss": 1.1458,
+      "step": 1216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5488208452308314e-05,
+      "loss": 1.2184,
+      "step": 1217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5416221101144446e-05,
+      "loss": 1.1111,
+      "step": 1218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.534424333082699e-05,
+      "loss": 1.1684,
+      "step": 1219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5272275291801226e-05,
+      "loss": 1.1901,
+      "step": 1220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.520031713449215e-05,
+      "loss": 1.1888,
+      "step": 1221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.512836900930401e-05,
+      "loss": 1.1492,
+      "step": 1222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505643106662017e-05,
+      "loss": 1.1892,
+      "step": 1223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4984503456802664e-05,
+      "loss": 1.1742,
+      "step": 1224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.491258633019194e-05,
+      "loss": 1.2484,
+      "step": 1225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484067983710652e-05,
+      "loss": 1.188,
+      "step": 1226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.476878412784277e-05,
+      "loss": 1.1296,
+      "step": 1227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.469689935267441e-05,
+      "loss": 1.2276,
+      "step": 1228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462502566185239e-05,
+      "loss": 1.1921,
+      "step": 1229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455316320560445e-05,
+      "loss": 1.1621,
+      "step": 1230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4481312134134853e-05,
+      "loss": 1.1957,
+      "step": 1231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.440947259762406e-05,
+      "loss": 1.1733,
+      "step": 1232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.433764474622845e-05,
+      "loss": 1.1975,
+      "step": 1233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4265828730079987e-05,
+      "loss": 1.0503,
+      "step": 1234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.419402469928582e-05,
+      "loss": 1.1776,
+      "step": 1235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.412223280392814e-05,
+      "loss": 1.1713,
+      "step": 1236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.405045319406371e-05,
+      "loss": 1.2766,
+      "step": 1237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397868601972364e-05,
+      "loss": 1.1385,
+      "step": 1238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.390693143091305e-05,
+      "loss": 1.2392,
+      "step": 1239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.383518957761073e-05,
+      "loss": 1.1148,
+      "step": 1240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.376346060976888e-05,
+      "loss": 1.1467,
+      "step": 1241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.369174467731277e-05,
+      "loss": 1.1776,
+      "step": 1242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3620041930140396e-05,
+      "loss": 1.2291,
+      "step": 1243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.354835251812221e-05,
+      "loss": 1.1671,
+      "step": 1244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3476676591100795e-05,
+      "loss": 1.2351,
+      "step": 1245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3405014298890524e-05,
+      "loss": 1.2011,
+      "step": 1246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.33333657912773e-05,
+      "loss": 1.1187,
+      "step": 1247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326173121801821e-05,
+      "loss": 1.1828,
+      "step": 1248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3190110728841215e-05,
+      "loss": 1.1815,
+      "step": 1249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.311850447344482e-05,
+      "loss": 1.2211,
+      "step": 1250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30469126014978e-05,
+      "loss": 1.1605,
+      "step": 1251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297533526263885e-05,
+      "loss": 1.1558,
+      "step": 1252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.290377260647629e-05,
+      "loss": 1.0767,
+      "step": 1253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283222478258778e-05,
+      "loss": 1.2218,
+      "step": 1254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2760691940519934e-05,
+      "loss": 1.1403,
+      "step": 1255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2689174229788106e-05,
+      "loss": 1.2884,
+      "step": 1256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2617671799875944e-05,
+      "loss": 1.124,
+      "step": 1257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2546184800235236e-05,
+      "loss": 1.0789,
+      "step": 1258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.247471338028547e-05,
+      "loss": 1.2632,
+      "step": 1259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240325768941357e-05,
+      "loss": 1.1492,
+      "step": 1260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233181787697363e-05,
+      "loss": 1.2358,
+      "step": 1261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2260394092286526e-05,
+      "loss": 1.1844,
+      "step": 1262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2188986484639605e-05,
+      "loss": 1.1905,
+      "step": 1263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.211759520328643e-05,
+      "loss": 1.1466,
+      "step": 1264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2046220397446465e-05,
+      "loss": 1.2546,
+      "step": 1265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.197486221630467e-05,
+      "loss": 1.1594,
+      "step": 1266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.190352080901131e-05,
+      "loss": 1.1862,
+      "step": 1267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.183219632468159e-05,
+      "loss": 1.242,
+      "step": 1268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.176088891239534e-05,
+      "loss": 1.1817,
+      "step": 1269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.168959872119665e-05,
+      "loss": 1.2054,
+      "step": 1270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.161832590009371e-05,
+      "loss": 1.1575,
+      "step": 1271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.154707059805832e-05,
+      "loss": 1.1793,
+      "step": 1272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.147583296402571e-05,
+      "loss": 1.1958,
+      "step": 1273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140461314689415e-05,
+      "loss": 1.1127,
+      "step": 1274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.133341129552473e-05,
+      "loss": 1.1604,
+      "step": 1275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.126222755874091e-05,
+      "loss": 1.2256,
+      "step": 1276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.119106208532834e-05,
+      "loss": 1.2348,
+      "step": 1277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1119915024034485e-05,
+      "loss": 1.1236,
+      "step": 1278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.10487865235683e-05,
+      "loss": 1.2674,
+      "step": 1279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.097767673259999e-05,
+      "loss": 1.1847,
+      "step": 1280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0906585799760585e-05,
+      "loss": 1.1512,
+      "step": 1281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0835513873641804e-05,
+      "loss": 1.1121,
+      "step": 1282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0764461102795545e-05,
+      "loss": 1.1531,
+      "step": 1283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0693427635733714e-05,
+      "loss": 1.1997,
+      "step": 1284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0622413620927844e-05,
+      "loss": 1.2496,
+      "step": 1285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055141920680883e-05,
+      "loss": 1.1677,
+      "step": 1286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0480444541766576e-05,
+      "loss": 1.2346,
+      "step": 1287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0409489774149757e-05,
+      "loss": 1.1912,
+      "step": 1288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033855505226539e-05,
+      "loss": 1.1053,
+      "step": 1289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026764052437863e-05,
+      "loss": 1.2746,
+      "step": 1290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0196746338712457e-05,
+      "loss": 1.1965,
+      "step": 1291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.012587264344725e-05,
+      "loss": 1.1461,
+      "step": 1292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0055019586720634e-05,
+      "loss": 1.2059,
+      "step": 1293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998418731662702e-05,
+      "loss": 1.2291,
+      "step": 1294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9913375981217496e-05,
+      "loss": 1.1308,
+      "step": 1295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.984258572849926e-05,
+      "loss": 1.2485,
+      "step": 1296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.977181670643553e-05,
+      "loss": 1.1953,
+      "step": 1297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.970106906294509e-05,
+      "loss": 1.1363,
+      "step": 1298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.963034294590208e-05,
+      "loss": 1.0788,
+      "step": 1299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955963850313566e-05,
+      "loss": 1.1445,
+      "step": 1300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94889558824296e-05,
+      "loss": 1.2416,
+      "step": 1301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.941829523152218e-05,
+      "loss": 1.189,
+      "step": 1302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934765669810567e-05,
+      "loss": 1.109,
+      "step": 1303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9277040429826164e-05,
+      "loss": 1.217,
+      "step": 1304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.920644657428315e-05,
+      "loss": 1.1437,
+      "step": 1305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.913587527902937e-05,
+      "loss": 1.1073,
+      "step": 1306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.90653266915703e-05,
+      "loss": 1.1702,
+      "step": 1307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899480095936402e-05,
+      "loss": 1.2432,
+      "step": 1308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892429822982086e-05,
+      "loss": 1.1472,
+      "step": 1309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.885381865030302e-05,
+      "loss": 1.2056,
+      "step": 1310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.878336236812432e-05,
+      "loss": 1.1442,
+      "step": 1311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871292953054991e-05,
+      "loss": 1.2011,
+      "step": 1312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.864252028479591e-05,
+      "loss": 1.1756,
+      "step": 1313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8572134778029147e-05,
+      "loss": 1.1599,
+      "step": 1314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8501773157366806e-05,
+      "loss": 1.1675,
+      "step": 1315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.843143556987619e-05,
+      "loss": 1.1778,
+      "step": 1316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.836112216257435e-05,
+      "loss": 1.2018,
+      "step": 1317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.829083308242776e-05,
+      "loss": 1.1887,
+      "step": 1318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8220568476352096e-05,
+      "loss": 1.2169,
+      "step": 1319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815032849121183e-05,
+      "loss": 1.1388,
+      "step": 1320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8080113273820026e-05,
+      "loss": 1.1804,
+      "step": 1321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.800992297093795e-05,
+      "loss": 1.2173,
+      "step": 1322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.793975772927481e-05,
+      "loss": 1.175,
+      "step": 1323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.786961769548739e-05,
+      "loss": 1.1574,
+      "step": 1324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.779950301617984e-05,
+      "loss": 1.0553,
+      "step": 1325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.772941383790328e-05,
+      "loss": 1.2031,
+      "step": 1326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.765935030715553e-05,
+      "loss": 1.1975,
+      "step": 1327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.75893125703808e-05,
+      "loss": 1.1521,
+      "step": 1328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.751930077396943e-05,
+      "loss": 1.2021,
+      "step": 1329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.744931506425749e-05,
+      "loss": 1.1563,
+      "step": 1330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.737935558752654e-05,
+      "loss": 1.1089,
+      "step": 1331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7309422490003335e-05,
+      "loss": 1.2317,
+      "step": 1332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.723951591785941e-05,
+      "loss": 1.1618,
+      "step": 1333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7169636017210973e-05,
+      "loss": 1.2474,
+      "step": 1334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.709978293411839e-05,
+      "loss": 1.1704,
+      "step": 1335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7029956814586045e-05,
+      "loss": 1.1847,
+      "step": 1336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6960157804561903e-05,
+      "loss": 1.2146,
+      "step": 1337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.689038604993731e-05,
+      "loss": 1.1601,
+      "step": 1338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.682064169654663e-05,
+      "loss": 1.2052,
+      "step": 1339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.675092489016693e-05,
+      "loss": 1.0568,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.668123577651773e-05,
+      "loss": 1.25,
+      "step": 1341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6611574501260636e-05,
+      "loss": 1.1718,
+      "step": 1342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6541941209999134e-05,
+      "loss": 1.1651,
+      "step": 1343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6472336048278145e-05,
+      "loss": 1.1862,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6402759161583845e-05,
+      "loss": 1.1993,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.633321069534325e-05,
+      "loss": 1.0916,
+      "step": 1346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.626369079492404e-05,
+      "loss": 1.1998,
+      "step": 1347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.619419960563416e-05,
+      "loss": 1.1705,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612473727272151e-05,
+      "loss": 1.0517,
+      "step": 1349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.605530394137376e-05,
+      "loss": 1.2751,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.598589975671787e-05,
+      "loss": 1.1427,
+      "step": 1351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5916524863819954e-05,
+      "loss": 1.2056,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.584717940768484e-05,
+      "loss": 1.1174,
+      "step": 1353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.577786353325586e-05,
+      "loss": 1.183,
+      "step": 1354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5708577385414495e-05,
+      "loss": 1.151,
+      "step": 1355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5639321108980154e-05,
+      "loss": 1.1759,
+      "step": 1356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5570094848709724e-05,
+      "loss": 1.1754,
+      "step": 1357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.550089874929741e-05,
+      "loss": 1.1702,
+      "step": 1358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.543173295537435e-05,
+      "loss": 1.2621,
+      "step": 1359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.536259761150835e-05,
+      "loss": 1.1031,
+      "step": 1360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5293492862203594e-05,
+      "loss": 1.2223,
+      "step": 1361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.522441885190023e-05,
+      "loss": 1.1431,
+      "step": 1362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5155375724974284e-05,
+      "loss": 1.2042,
+      "step": 1363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508636362573713e-05,
+      "loss": 1.1831,
+      "step": 1364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5017382698435354e-05,
+      "loss": 1.1764,
+      "step": 1365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494843308725032e-05,
+      "loss": 1.1494,
+      "step": 1366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487951493629801e-05,
+      "loss": 1.2029,
+      "step": 1367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.481062838962859e-05,
+      "loss": 1.1963,
+      "step": 1368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.474177359122618e-05,
+      "loss": 1.1734,
+      "step": 1369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4672950685008583e-05,
+      "loss": 1.112,
+      "step": 1370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.460415981482691e-05,
+      "loss": 1.0837,
+      "step": 1371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4535401124465296e-05,
+      "loss": 1.2803,
+      "step": 1372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4466674757640636e-05,
+      "loss": 1.2008,
+      "step": 1373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.439798085800228e-05,
+      "loss": 1.1986,
+      "step": 1374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.432931956913166e-05,
+      "loss": 1.1657,
+      "step": 1375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.426069103454209e-05,
+      "loss": 1.1778,
+      "step": 1376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.419209539767844e-05,
+      "loss": 1.1489,
+      "step": 1377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.412353280191678e-05,
+      "loss": 1.1726,
+      "step": 1378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4055003390564114e-05,
+      "loss": 1.2938,
+      "step": 1379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3986507306858125e-05,
+      "loss": 1.1176,
+      "step": 1380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.39180446939668e-05,
+      "loss": 1.1458,
+      "step": 1381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.384961569498817e-05,
+      "loss": 1.2087,
+      "step": 1382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3781220452950044e-05,
+      "loss": 1.1709,
+      "step": 1383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3712859110809645e-05,
+      "loss": 1.1725,
+      "step": 1384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.364453181145331e-05,
+      "loss": 1.1891,
+      "step": 1385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.357623869769628e-05,
+      "loss": 1.2181,
+      "step": 1386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350797991228231e-05,
+      "loss": 1.1332,
+      "step": 1387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343975559788339e-05,
+      "loss": 1.2054,
+      "step": 1388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.337156589709949e-05,
+      "loss": 1.087,
+      "step": 1389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3303410952458237e-05,
+      "loss": 1.2317,
+      "step": 1390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3235290906414574e-05,
+      "loss": 1.1166,
+      "step": 1391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.316720590135052e-05,
+      "loss": 1.2221,
+      "step": 1392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309915607957487e-05,
+      "loss": 1.1736,
+      "step": 1393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303114158332285e-05,
+      "loss": 1.2119,
+      "step": 1394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.296316255475586e-05,
+      "loss": 1.1623,
+      "step": 1395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289521913596117e-05,
+      "loss": 1.1074,
+      "step": 1396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2827311468951646e-05,
+      "loss": 1.1539,
+      "step": 1397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.275943969566538e-05,
+      "loss": 1.1889,
+      "step": 1398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.269160395796547e-05,
+      "loss": 1.1632,
+      "step": 1399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2623804397639666e-05,
+      "loss": 1.1695,
+      "step": 1400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.255604115640014e-05,
+      "loss": 1.161,
+      "step": 1401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2488314375883125e-05,
+      "loss": 1.168,
+      "step": 1402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.242062419764864e-05,
+      "loss": 1.1018,
+      "step": 1403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2352970763180234e-05,
+      "loss": 1.1711,
+      "step": 1404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2285354213884617e-05,
+      "loss": 1.148,
+      "step": 1405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2217774691091435e-05,
+      "loss": 1.2279,
+      "step": 1406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.215023233605291e-05,
+      "loss": 1.1578,
+      "step": 1407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2082727289943604e-05,
+      "loss": 1.167,
+      "step": 1408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201525969386007e-05,
+      "loss": 1.1707,
+      "step": 1409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1947829688820595e-05,
+      "loss": 1.2278,
+      "step": 1410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.188043741576494e-05,
+      "loss": 1.129,
+      "step": 1411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.181308301555393e-05,
+      "loss": 1.1629,
+      "step": 1412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.174576662896927e-05,
+      "loss": 1.1394,
+      "step": 1413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.16784883967132e-05,
+      "loss": 1.1756,
+      "step": 1414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.161124845940822e-05,
+      "loss": 1.1957,
+      "step": 1415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.154404695759675e-05,
+      "loss": 1.1621,
+      "step": 1416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.147688403174092e-05,
+      "loss": 1.1631,
+      "step": 1417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140975982222222e-05,
+      "loss": 1.1085,
+      "step": 1418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1342674469341216e-05,
+      "loss": 1.139,
+      "step": 1419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.127562811331723e-05,
+      "loss": 1.1975,
+      "step": 1420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.12086208942881e-05,
+      "loss": 1.1913,
+      "step": 1421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114165295230986e-05,
+      "loss": 1.1398,
+      "step": 1422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.107472442735643e-05,
+      "loss": 1.2529,
+      "step": 1423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1007835459319364e-05,
+      "loss": 1.1047,
+      "step": 1424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.094098618800756e-05,
+      "loss": 1.1916,
+      "step": 1425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0874176753146864e-05,
+      "loss": 1.1604,
+      "step": 1426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.080740729437993e-05,
+      "loss": 1.2266,
+      "step": 1427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0740677951265826e-05,
+      "loss": 1.0982,
+      "step": 1428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.067398886327977e-05,
+      "loss": 1.1909,
+      "step": 1429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.060734016981284e-05,
+      "loss": 1.1931,
+      "step": 1430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0540732010171705e-05,
+      "loss": 1.1432,
+      "step": 1431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0474164523578297e-05,
+      "loss": 1.2363,
+      "step": 1432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0407637849169518e-05,
+      "loss": 1.1786,
+      "step": 1433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0341152125996997e-05,
+      "loss": 1.1803,
+      "step": 1434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0274707493026734e-05,
+      "loss": 1.1958,
+      "step": 1435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0208304089138872e-05,
+      "loss": 1.18,
+      "step": 1436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.014194205312736e-05,
+      "loss": 1.0305,
+      "step": 1437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0075621523699714e-05,
+      "loss": 1.1102,
+      "step": 1438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0009342639476647e-05,
+      "loss": 1.1581,
+      "step": 1439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9943105538991857e-05,
+      "loss": 1.1597,
+      "step": 1440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9876910360691707e-05,
+      "loss": 1.2466,
+      "step": 1441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9810757242934916e-05,
+      "loss": 1.129,
+      "step": 1442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.974464632399232e-05,
+      "loss": 1.2301,
+      "step": 1443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9678577742046488e-05,
+      "loss": 1.0866,
+      "step": 1444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.961255163519162e-05,
+      "loss": 1.1437,
+      "step": 1445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9546568141433006e-05,
+      "loss": 1.2401,
+      "step": 1446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9480627398686954e-05,
+      "loss": 1.109,
+      "step": 1447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.941472954478036e-05,
+      "loss": 1.2141,
+      "step": 1448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9348874717450505e-05,
+      "loss": 1.1349,
+      "step": 1449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.928306305434473e-05,
+      "loss": 1.2086,
+      "step": 1450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921729469302018e-05,
+      "loss": 1.1334,
+      "step": 1451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9151569770943445e-05,
+      "loss": 1.2264,
+      "step": 1452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908588842549037e-05,
+      "loss": 1.1343,
+      "step": 1453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.902025079394566e-05,
+      "loss": 1.1016,
+      "step": 1454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.89546570135027e-05,
+      "loss": 1.2483,
+      "step": 1455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.88891072212632e-05,
+      "loss": 1.1546,
+      "step": 1456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8823601554236933e-05,
+      "loss": 1.2206,
+      "step": 1457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.875814014934144e-05,
+      "loss": 1.1516,
+      "step": 1458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8692723143401746e-05,
+      "loss": 1.1767,
+      "step": 1459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8627350673150073e-05,
+      "loss": 1.1239,
+      "step": 1460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8562022875225557e-05,
+      "loss": 1.0129,
+      "step": 1461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8496739886173995e-05,
+      "loss": 1.1824,
+      "step": 1462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.843150184244744e-05,
+      "loss": 1.2057,
+      "step": 1463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.836630888040406e-05,
+      "loss": 1.1662,
+      "step": 1464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830116113630786e-05,
+      "loss": 1.1626,
+      "step": 1465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8236058746328205e-05,
+      "loss": 1.1709,
+      "step": 1466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8171001846539725e-05,
+      "loss": 1.1232,
+      "step": 1467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8105990572921977e-05,
+      "loss": 1.239,
+      "step": 1468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8041025061359127e-05,
+      "loss": 1.1474,
+      "step": 1469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7976105447639723e-05,
+      "loss": 1.1598,
+      "step": 1470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7911231867456344e-05,
+      "loss": 1.181,
+      "step": 1471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7846404456405363e-05,
+      "loss": 1.1328,
+      "step": 1472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7781623349986663e-05,
+      "loss": 1.1123,
+      "step": 1473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.771688868360336e-05,
+      "loss": 1.1772,
+      "step": 1474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7652200592561417e-05,
+      "loss": 1.2047,
+      "step": 1475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7587559212069553e-05,
+      "loss": 1.1583,
+      "step": 1476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7522964677238795e-05,
+      "loss": 1.2248,
+      "step": 1477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.745841712308227e-05,
+      "loss": 1.1744,
+      "step": 1478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.739391668451491e-05,
+      "loss": 1.0945,
+      "step": 1479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7329463496353154e-05,
+      "loss": 1.2041,
+      "step": 1480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.72650576933147e-05,
+      "loss": 1.1759,
+      "step": 1481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7200699410018193e-05,
+      "loss": 1.152,
+      "step": 1482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.713638878098298e-05,
+      "loss": 1.2161,
+      "step": 1483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7072125940628702e-05,
+      "loss": 1.1293,
+      "step": 1484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7007911023275255e-05,
+      "loss": 0.9974,
+      "step": 1485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6943744163142294e-05,
+      "loss": 1.2568,
+      "step": 1486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6879625494349042e-05,
+      "loss": 1.1126,
+      "step": 1487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.681555515091395e-05,
+      "loss": 1.2039,
+      "step": 1488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6751533266754524e-05,
+      "loss": 1.1628,
+      "step": 1489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6687559975686948e-05,
+      "loss": 1.1411,
+      "step": 1490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.662363541142585e-05,
+      "loss": 1.1511,
+      "step": 1491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.655975970758401e-05,
+      "loss": 1.2503,
+      "step": 1492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6495932997672074e-05,
+      "loss": 1.1511,
+      "step": 1493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6432155415098292e-05,
+      "loss": 1.1849,
+      "step": 1494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6368427093168234e-05,
+      "loss": 1.1413,
+      "step": 1495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6304748165084503e-05,
+      "loss": 1.1112,
+      "step": 1496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6241118763946433e-05,
+      "loss": 1.2475,
+      "step": 1497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6177539022749853e-05,
+      "loss": 1.1012,
+      "step": 1498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6114009074386846e-05,
+      "loss": 1.1892,
+      "step": 1499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6050529051645384e-05,
+      "loss": 1.2214,
+      "step": 1500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5987099087209033e-05,
+      "loss": 1.1075,
+      "step": 1501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.59237193136568e-05,
+      "loss": 1.1217,
+      "step": 1502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5860389863462765e-05,
+      "loss": 1.2593,
+      "step": 1503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.579711086899581e-05,
+      "loss": 1.1022,
+      "step": 1504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5733882462519378e-05,
+      "loss": 1.2399,
+      "step": 1505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5670704776191156e-05,
+      "loss": 1.2062,
+      "step": 1506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5607577942062833e-05,
+      "loss": 1.124,
+      "step": 1507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.554450209207979e-05,
+      "loss": 1.1724,
+      "step": 1508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5481477358080878e-05,
+      "loss": 1.1917,
+      "step": 1509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5418503871798048e-05,
+      "loss": 1.0676,
+      "step": 1510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5355581764856185e-05,
+      "loss": 1.1546,
+      "step": 1511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5292711168772744e-05,
+      "loss": 1.1752,
+      "step": 1512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5229892214957608e-05,
+      "loss": 1.1979,
+      "step": 1513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5167125034712576e-05,
+      "loss": 1.1087,
+      "step": 1514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.510440975923133e-05,
+      "loss": 1.1075,
+      "step": 1515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5041746519599036e-05,
+      "loss": 1.1448,
+      "step": 1516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.49791354467921e-05,
+      "loss": 1.2728,
+      "step": 1517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4916576671677888e-05,
+      "loss": 1.1231,
+      "step": 1518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.485407032501445e-05,
+      "loss": 1.2584,
+      "step": 1519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4791616537450268e-05,
+      "loss": 1.0747,
+      "step": 1520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4729215439523944e-05,
+      "loss": 1.2052,
+      "step": 1521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4666867161663992e-05,
+      "loss": 1.1531,
+      "step": 1522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.460457183418845e-05,
+      "loss": 1.2618,
+      "step": 1523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4542329587304748e-05,
+      "loss": 1.1297,
+      "step": 1524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4480140551109325e-05,
+      "loss": 1.2215,
+      "step": 1525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.441800485558748e-05,
+      "loss": 1.1245,
+      "step": 1526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4355922630612897e-05,
+      "loss": 1.1461,
+      "step": 1527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4293894005947592e-05,
+      "loss": 1.2072,
+      "step": 1528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4231919111241524e-05,
+      "loss": 1.1744,
+      "step": 1529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4169998076032336e-05,
+      "loss": 1.1807,
+      "step": 1530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.410813102974512e-05,
+      "loss": 1.1621,
+      "step": 1531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4046318101692055e-05,
+      "loss": 1.1459,
+      "step": 1532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3984559421072304e-05,
+      "loss": 1.1923,
+      "step": 1533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.392285511697159e-05,
+      "loss": 1.1002,
+      "step": 1534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.386120531836198e-05,
+      "loss": 1.085,
+      "step": 1535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3799610154101604e-05,
+      "loss": 1.2494,
+      "step": 1536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.373806975293442e-05,
+      "loss": 1.1077,
+      "step": 1537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3676584243489923e-05,
+      "loss": 1.098,
+      "step": 1538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3615153754282865e-05,
+      "loss": 1.1987,
+      "step": 1539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3553778413712997e-05,
+      "loss": 1.1295,
+      "step": 1540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3492458350064806e-05,
+      "loss": 1.1626,
+      "step": 1541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.343119369150724e-05,
+      "loss": 1.1583,
+      "step": 1542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3369984566093438e-05,
+      "loss": 1.1762,
+      "step": 1543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3308831101760486e-05,
+      "loss": 1.1765,
+      "step": 1544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3247733426329048e-05,
+      "loss": 1.2014,
+      "step": 1545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3186691667503313e-05,
+      "loss": 1.1756,
+      "step": 1546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3125705952870498e-05,
+      "loss": 1.1015,
+      "step": 1547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3064776409900725e-05,
+      "loss": 1.2115,
+      "step": 1548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3003903165946643e-05,
+      "loss": 1.2263,
+      "step": 1549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2943086348243304e-05,
+      "loss": 1.1807,
+      "step": 1550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.288232608390778e-05,
+      "loss": 1.0791,
+      "step": 1551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.282162249993895e-05,
+      "loss": 1.1581,
+      "step": 1552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.276097572321721e-05,
+      "loss": 1.2027,
+      "step": 1553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.270038588050424e-05,
+      "loss": 1.1037,
+      "step": 1554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2639853098442685e-05,
+      "loss": 1.1195,
+      "step": 1555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2579377503555964e-05,
+      "loss": 1.2676,
+      "step": 1556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2518959222247954e-05,
+      "loss": 1.1208,
+      "step": 1557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.24585983808027e-05,
+      "loss": 1.1234,
+      "step": 1558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2398295105384206e-05,
+      "loss": 1.1907,
+      "step": 1559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.233804952203621e-05,
+      "loss": 1.1683,
+      "step": 1560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2277861756681812e-05,
+      "loss": 1.0777,
+      "step": 1561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2217731935123243e-05,
+      "loss": 1.1747,
+      "step": 1562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2157660183041655e-05,
+      "loss": 1.1012,
+      "step": 1563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2097646625996832e-05,
+      "loss": 1.2068,
+      "step": 1564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2037691389426907e-05,
+      "loss": 1.248,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1977794598648106e-05,
+      "loss": 1.2054,
+      "step": 1566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1917956378854505e-05,
+      "loss": 1.1026,
+      "step": 1567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1858176855117756e-05,
+      "loss": 1.1543,
+      "step": 1568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1798456152386832e-05,
+      "loss": 1.1131,
+      "step": 1569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.173879439548776e-05,
+      "loss": 1.2137,
+      "step": 1570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.167919170912333e-05,
+      "loss": 1.1801,
+      "step": 1571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1619648217872897e-05,
+      "loss": 1.1795,
+      "step": 1572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1560164046192073e-05,
+      "loss": 1.1442,
+      "step": 1573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.150073931841253e-05,
+      "loss": 1.223,
+      "step": 1574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1441374158741618e-05,
+      "loss": 1.1166,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.138206869126222e-05,
+      "loss": 1.1907,
+      "step": 1576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1322823039932448e-05,
+      "loss": 1.1839,
+      "step": 1577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1263637328585394e-05,
+      "loss": 1.1425,
+      "step": 1578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.120451168092885e-05,
+      "loss": 1.1037,
+      "step": 1579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1145446220545084e-05,
+      "loss": 1.1487,
+      "step": 1580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108644107089055e-05,
+      "loss": 1.1268,
+      "step": 1581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1027496355295644e-05,
+      "loss": 1.1821,
+      "step": 1582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.096861219696447e-05,
+      "loss": 1.0997,
+      "step": 1583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0909788718974493e-05,
+      "loss": 1.2348,
+      "step": 1584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0851026044276406e-05,
+      "loss": 1.1527,
+      "step": 1585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0792324295693784e-05,
+      "loss": 1.1276,
+      "step": 1586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.073368359592291e-05,
+      "loss": 1.1477,
+      "step": 1587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0675104067532374e-05,
+      "loss": 1.1396,
+      "step": 1588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0616585832962975e-05,
+      "loss": 1.1541,
+      "step": 1589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0558129014527382e-05,
+      "loss": 1.1162,
+      "step": 1590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.049973373440988e-05,
+      "loss": 1.1948,
+      "step": 1591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0441400114666174e-05,
+      "loss": 1.1351,
+      "step": 1592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.038312827722299e-05,
+      "loss": 1.1414,
+      "step": 1593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.032491834387804e-05,
+      "loss": 1.2408,
+      "step": 1594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0266770436299575e-05,
+      "loss": 1.0964,
+      "step": 1595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.020868467602623e-05,
+      "loss": 1.2,
+      "step": 1596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0150661184466696e-05,
+      "loss": 1.1868,
+      "step": 1597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0092700082899567e-05,
+      "loss": 1.1586,
+      "step": 1598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0034801492473005e-05,
+      "loss": 1.167,
+      "step": 1599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9976965534204516e-05,
+      "loss": 1.1629,
+      "step": 1600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.99191923289807e-05,
+      "loss": 1.1355,
+      "step": 1601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9861481997556998e-05,
+      "loss": 1.1517,
+      "step": 1602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.980383466055742e-05,
+      "loss": 1.173,
+      "step": 1603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9746250438474307e-05,
+      "loss": 1.1379,
+      "step": 1604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9688729451668114e-05,
+      "loss": 1.0565,
+      "step": 1605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.963127182036706e-05,
+      "loss": 1.1789,
+      "step": 1606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9573877664666973e-05,
+      "loss": 1.2395,
+      "step": 1607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9516547104531046e-05,
+      "loss": 1.0909,
+      "step": 1608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9459280259789515e-05,
+      "loss": 1.1481,
+      "step": 1609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.94020772501394e-05,
+      "loss": 1.1971,
+      "step": 1610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9344938195144352e-05,
+      "loss": 1.1313,
+      "step": 1611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.928786321423432e-05,
+      "loss": 1.1835,
+      "step": 1612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9230852426705344e-05,
+      "loss": 1.1615,
+      "step": 1613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9173905951719267e-05,
+      "loss": 1.2361,
+      "step": 1614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9117023908303533e-05,
+      "loss": 1.1468,
+      "step": 1615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9060206415350894e-05,
+      "loss": 1.2065,
+      "step": 1616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9003453591619198e-05,
+      "loss": 1.0803,
+      "step": 1617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8946765555731117e-05,
+      "loss": 1.137,
+      "step": 1618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8890142426173872e-05,
+      "loss": 1.2468,
+      "step": 1619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8833584321299063e-05,
+      "loss": 1.0952,
+      "step": 1620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8777091359322387e-05,
+      "loss": 1.2109,
+      "step": 1621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8720663658323355e-05,
+      "loss": 1.1747,
+      "step": 1622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8664301336245056e-05,
+      "loss": 1.1958,
+      "step": 1623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8608004510893956e-05,
+      "loss": 1.1089,
+      "step": 1624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.855177329993962e-05,
+      "loss": 1.1654,
+      "step": 1625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.849560782091445e-05,
+      "loss": 1.1605,
+      "step": 1626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8439508191213477e-05,
+      "loss": 1.2011,
+      "step": 1627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8383474528094085e-05,
+      "loss": 1.0999,
+      "step": 1628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8327506948675776e-05,
+      "loss": 1.055,
+      "step": 1629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8271605569939942e-05,
+      "loss": 1.2117,
+      "step": 1630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.82157705087296e-05,
+      "loss": 1.1694,
+      "step": 1631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8160001881749116e-05,
+      "loss": 1.2383,
+      "step": 1632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.810429980556404e-05,
+      "loss": 1.0225,
+      "step": 1633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8048664396600796e-05,
+      "loss": 1.1781,
+      "step": 1634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.799309577114652e-05,
+      "loss": 1.1511,
+      "step": 1635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7937594045348666e-05,
+      "loss": 1.1768,
+      "step": 1636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7882159335214916e-05,
+      "loss": 1.1637,
+      "step": 1637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.782679175661287e-05,
+      "loss": 1.1034,
+      "step": 1638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.777149142526981e-05,
+      "loss": 1.2094,
+      "step": 1639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771625845677247e-05,
+      "loss": 1.1264,
+      "step": 1640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.766109296656673e-05,
+      "loss": 1.1438,
+      "step": 1641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7605995069957522e-05,
+      "loss": 1.1935,
+      "step": 1642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7550964882108422e-05,
+      "loss": 1.1432,
+      "step": 1643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7496002518041538e-05,
+      "loss": 1.196,
+      "step": 1644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7441108092637144e-05,
+      "loss": 1.1339,
+      "step": 1645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.738628172063357e-05,
+      "loss": 1.1346,
+      "step": 1646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7331523516626873e-05,
+      "loss": 1.1884,
+      "step": 1647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7276833595070692e-05,
+      "loss": 1.1297,
+      "step": 1648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7222212070275846e-05,
+      "loss": 1.1085,
+      "step": 1649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7167659056410246e-05,
+      "loss": 1.071,
+      "step": 1650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.711317466749861e-05,
+      "loss": 1.1036,
+      "step": 1651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.705875901742221e-05,
+      "loss": 1.2283,
+      "step": 1652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7004412219918646e-05,
+      "loss": 1.1626,
+      "step": 1653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6950134388581557e-05,
+      "loss": 1.1876,
+      "step": 1654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6895925636860526e-05,
+      "loss": 1.1631,
+      "step": 1655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6841786078060673e-05,
+      "loss": 1.2177,
+      "step": 1656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6787715825342537e-05,
+      "loss": 1.1032,
+      "step": 1657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.673371499172174e-05,
+      "loss": 1.1014,
+      "step": 1658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.667978369006886e-05,
+      "loss": 1.2525,
+      "step": 1659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6625922033109132e-05,
+      "loss": 1.097,
+      "step": 1660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6572130133422203e-05,
+      "loss": 1.1362,
+      "step": 1661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.651840810344193e-05,
+      "loss": 1.2001,
+      "step": 1662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6464756055456133e-05,
+      "loss": 1.1273,
+      "step": 1663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6411174101606347e-05,
+      "loss": 1.1729,
+      "step": 1664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.635766235388761e-05,
+      "loss": 1.1219,
+      "step": 1665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6304220924148226e-05,
+      "loss": 1.1661,
+      "step": 1666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6250849924089484e-05,
+      "loss": 1.1547,
+      "step": 1667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6197549465265484e-05,
+      "loss": 1.1743,
+      "step": 1668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6144319659082925e-05,
+      "loss": 1.1509,
+      "step": 1669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6091160616800783e-05,
+      "loss": 1.1076,
+      "step": 1670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6038072449530117e-05,
+      "loss": 1.1014,
+      "step": 1671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5985055268233878e-05,
+      "loss": 1.1973,
+      "step": 1672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5932109183726627e-05,
+      "loss": 1.1539,
+      "step": 1673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5879234306674338e-05,
+      "loss": 1.116,
+      "step": 1674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.582643074759413e-05,
+      "loss": 1.1409,
+      "step": 1675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5773698616854067e-05,
+      "loss": 1.1584,
+      "step": 1676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5721038024672906e-05,
+      "loss": 1.1689,
+      "step": 1677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5668449081119896e-05,
+      "loss": 1.1471,
+      "step": 1678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5615931896114528e-05,
+      "loss": 1.1446,
+      "step": 1679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5563486579426256e-05,
+      "loss": 1.1526,
+      "step": 1680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5511113240674363e-05,
+      "loss": 1.0551,
+      "step": 1681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5458811989327708e-05,
+      "loss": 1.1459,
+      "step": 1682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.540658293470445e-05,
+      "loss": 1.1865,
+      "step": 1683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5354426185971804e-05,
+      "loss": 1.15,
+      "step": 1684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.530234185214592e-05,
+      "loss": 1.1832,
+      "step": 1685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5250330042091543e-05,
+      "loss": 1.1423,
+      "step": 1686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5198390864521866e-05,
+      "loss": 1.1331,
+      "step": 1687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5146524427998244e-05,
+      "loss": 1.1744,
+      "step": 1688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5094730840929999e-05,
+      "loss": 1.1607,
+      "step": 1689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5043010211574187e-05,
+      "loss": 1.1772,
+      "step": 1690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4991362648035368e-05,
+      "loss": 1.1081,
+      "step": 1691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4939788258265402e-05,
+      "loss": 1.2327,
+      "step": 1692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4888287150063151e-05,
+      "loss": 1.1537,
+      "step": 1693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4836859431074357e-05,
+      "loss": 1.1378,
+      "step": 1694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4785505208791334e-05,
+      "loss": 1.1816,
+      "step": 1695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4734224590552842e-05,
+      "loss": 1.2007,
+      "step": 1696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.46830176835437e-05,
+      "loss": 1.178,
+      "step": 1697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.463188459479472e-05,
+      "loss": 1.1865,
+      "step": 1698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.458082543118241e-05,
+      "loss": 1.008,
+      "step": 1699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4529840299428754e-05,
+      "loss": 1.0769,
+      "step": 1700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4478929306101025e-05,
+      "loss": 1.2435,
+      "step": 1701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.442809255761146e-05,
+      "loss": 1.1026,
+      "step": 1702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4377330160217218e-05,
+      "loss": 1.26,
+      "step": 1703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4326642220019975e-05,
+      "loss": 1.1474,
+      "step": 1704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4276028842965827e-05,
+      "loss": 1.1062,
+      "step": 1705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4225490134844966e-05,
+      "loss": 1.1153,
+      "step": 1706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4175026201291552e-05,
+      "loss": 1.2212,
+      "step": 1707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4124637147783432e-05,
+      "loss": 1.1076,
+      "step": 1708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.407432307964201e-05,
+      "loss": 1.1122,
+      "step": 1709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4024084102031859e-05,
+      "loss": 1.1728,
+      "step": 1710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3973920319960655e-05,
+      "loss": 1.1265,
+      "step": 1711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3923831838278889e-05,
+      "loss": 1.2159,
+      "step": 1712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.387381876167968e-05,
+      "loss": 1.1688,
+      "step": 1713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3823881194698535e-05,
+      "loss": 1.162,
+      "step": 1714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3774019241713077e-05,
+      "loss": 1.1214,
+      "step": 1715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3724233006942972e-05,
+      "loss": 1.1493,
+      "step": 1716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3674522594449573e-05,
+      "loss": 1.0406,
+      "step": 1717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.362488810813577e-05,
+      "loss": 1.1503,
+      "step": 1718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3575329651745716e-05,
+      "loss": 1.1575,
+      "step": 1719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3525847328864688e-05,
+      "loss": 1.1141,
+      "step": 1720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.347644124291883e-05,
+      "loss": 1.2517,
+      "step": 1721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3427111497174922e-05,
+      "loss": 1.0428,
+      "step": 1722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3377858194740189e-05,
+      "loss": 1.1923,
+      "step": 1723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3328681438562079e-05,
+      "loss": 1.1286,
+      "step": 1724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3279581331428043e-05,
+      "loss": 1.1791,
+      "step": 1725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3230557975965318e-05,
+      "loss": 1.168,
+      "step": 1726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3181611474640742e-05,
+      "loss": 1.1582,
+      "step": 1727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3132741929760467e-05,
+      "loss": 1.1894,
+      "step": 1728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3083949443469811e-05,
+      "loss": 1.1551,
+      "step": 1729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3035234117753082e-05,
+      "loss": 1.1674,
+      "step": 1730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2986596054433254e-05,
+      "loss": 1.0633,
+      "step": 1731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2938035355171802e-05,
+      "loss": 1.1982,
+      "step": 1732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2889552121468524e-05,
+      "loss": 1.1861,
+      "step": 1733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2841146454661301e-05,
+      "loss": 1.1412,
+      "step": 1734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2792818455925875e-05,
+      "loss": 1.1428,
+      "step": 1735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.274456822627566e-05,
+      "loss": 1.1576,
+      "step": 1736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2696395866561516e-05,
+      "loss": 1.1256,
+      "step": 1737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.264830147747154e-05,
+      "loss": 1.23,
+      "step": 1738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.260028515953086e-05,
+      "loss": 1.0944,
+      "step": 1739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2552347013101428e-05,
+      "loss": 1.1893,
+      "step": 1740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2504487138381781e-05,
+      "loss": 0.9785,
+      "step": 1741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2456705635406862e-05,
+      "loss": 1.0995,
+      "step": 1742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2409002604047853e-05,
+      "loss": 1.248,
+      "step": 1743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.236137814401187e-05,
+      "loss": 1.1362,
+      "step": 1744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2313832354841792e-05,
+      "loss": 1.1694,
+      "step": 1745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2266365335916085e-05,
+      "loss": 1.1217,
+      "step": 1746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2218977186448582e-05,
+      "loss": 1.1175,
+      "step": 1747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2171668005488246e-05,
+      "loss": 1.2332,
+      "step": 1748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2124437891918993e-05,
+      "loss": 1.2316,
+      "step": 1749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2077286944459477e-05,
+      "loss": 1.1243,
+      "step": 1750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.203021526166287e-05,
+      "loss": 1.1396,
+      "step": 1751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1983222941916678e-05,
+      "loss": 1.1307,
+      "step": 1752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1936310083442542e-05,
+      "loss": 1.1239,
+      "step": 1753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.188947678429596e-05,
+      "loss": 1.1953,
+      "step": 1754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1842723142366196e-05,
+      "loss": 1.1588,
+      "step": 1755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1796049255375979e-05,
+      "loss": 1.1006,
+      "step": 1756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.17494552208814e-05,
+      "loss": 1.1747,
+      "step": 1757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1702941136271557e-05,
+      "loss": 1.1488,
+      "step": 1758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1656507098768487e-05,
+      "loss": 1.1244,
+      "step": 1759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1610153205426927e-05,
+      "loss": 1.1494,
+      "step": 1760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1563879553134072e-05,
+      "loss": 1.2046,
+      "step": 1761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1517686238609427e-05,
+      "loss": 1.1197,
+      "step": 1762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1471573358404526e-05,
+      "loss": 1.1675,
+      "step": 1763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1425541008902851e-05,
+      "loss": 1.0905,
+      "step": 1764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1379589286319525e-05,
+      "loss": 1.0284,
+      "step": 1765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1333718286701161e-05,
+      "loss": 1.2312,
+      "step": 1766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1287928105925611e-05,
+      "loss": 1.1104,
+      "step": 1767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1242218839701851e-05,
+      "loss": 1.242,
+      "step": 1768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.11965905835697e-05,
+      "loss": 1.1218,
+      "step": 1769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1151043432899672e-05,
+      "loss": 1.0979,
+      "step": 1770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1105577482892748e-05,
+      "loss": 1.139,
+      "step": 1771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1060192828580185e-05,
+      "loss": 1.1278,
+      "step": 1772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1014889564823328e-05,
+      "loss": 1.1292,
+      "step": 1773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0969667786313386e-05,
+      "loss": 1.1931,
+      "step": 1774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0924527587571288e-05,
+      "loss": 1.1658,
+      "step": 1775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0879469062947367e-05,
+      "loss": 1.1376,
+      "step": 1776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0834492306621342e-05,
+      "loss": 1.2373,
+      "step": 1777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0789597412601982e-05,
+      "loss": 1.1258,
+      "step": 1778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0744784474726915e-05,
+      "loss": 1.0894,
+      "step": 1779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0700053586662518e-05,
+      "loss": 1.1781,
+      "step": 1780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0655404841903648e-05,
+      "loss": 1.1094,
+      "step": 1781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.061083833377347e-05,
+      "loss": 1.1722,
+      "step": 1782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0566354155423263e-05,
+      "loss": 1.2162,
+      "step": 1783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0521952399832235e-05,
+      "loss": 1.0999,
+      "step": 1784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0477633159807294e-05,
+      "loss": 1.1994,
+      "step": 1785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0433396527982892e-05,
+      "loss": 1.1973,
+      "step": 1786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0389242596820824e-05,
+      "loss": 1.1125,
+      "step": 1787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0345171458609992e-05,
+      "loss": 1.1301,
+      "step": 1788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0301183205466292e-05,
+      "loss": 1.2053,
+      "step": 1789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0257277929332332e-05,
+      "loss": 1.1162,
+      "step": 1790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0213455721977355e-05,
+      "loss": 1.1303,
+      "step": 1791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0169716674996887e-05,
+      "loss": 1.2038,
+      "step": 1792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0126060879812694e-05,
+      "loss": 1.1372,
+      "step": 1793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0082488427672515e-05,
+      "loss": 1.0548,
+      "step": 1794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0038999409649891e-05,
+      "loss": 1.1288,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.99559391664398e-06,
+      "loss": 1.137,
+      "step": 1796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.952272039379324e-06,
+      "loss": 1.1914,
+      "step": 1797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.909033868405748e-06,
+      "loss": 1.1034,
+      "step": 1798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.86587949409808e-06,
+      "loss": 1.1923,
+      "step": 1799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.82280900665602e-06,
+      "loss": 1.1237,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.779822496103902e-06,
+      "loss": 1.1981,
+      "step": 1801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.736920052290555e-06,
+      "loss": 1.0934,
+      "step": 1802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.694101764889102e-06,
+      "loss": 1.2368,
+      "step": 1803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.651367723396749e-06,
+      "loss": 1.0312,
+      "step": 1804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.608718017134622e-06,
+      "loss": 1.2061,
+      "step": 1805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.566152735247575e-06,
+      "loss": 1.1467,
+      "step": 1806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.523671966703995e-06,
+      "loss": 1.1088,
+      "step": 1807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.481275800295631e-06,
+      "loss": 1.1624,
+      "step": 1808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.438964324637395e-06,
+      "loss": 1.1784,
+      "step": 1809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.39673762816715e-06,
+      "loss": 1.1511,
+      "step": 1810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.354595799145626e-06,
+      "loss": 1.2105,
+      "step": 1811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31253892565611e-06,
+      "loss": 1.1136,
+      "step": 1812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.270567095604355e-06,
+      "loss": 1.1281,
+      "step": 1813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.228680396718308e-06,
+      "loss": 1.0892,
+      "step": 1814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.186878916548025e-06,
+      "loss": 1.0758,
+      "step": 1815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.145162742465423e-06,
+      "loss": 1.2377,
+      "step": 1816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.103531961664118e-06,
+      "loss": 1.1425,
+      "step": 1817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.06198666115925e-06,
+      "loss": 1.1256,
+      "step": 1818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.020526927787265e-06,
+      "loss": 1.1429,
+      "step": 1819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.979152848205785e-06,
+      "loss": 1.1594,
+      "step": 1820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.937864508893396e-06,
+      "loss": 1.0882,
+      "step": 1821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.896661996149475e-06,
+      "loss": 1.1517,
+      "step": 1822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.855545396093973e-06,
+      "loss": 1.2237,
+      "step": 1823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.81451479466731e-06,
+      "loss": 1.1067,
+      "step": 1824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.773570277630166e-06,
+      "loss": 1.2091,
+      "step": 1825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.73271193056326e-06,
+      "loss": 1.1234,
+      "step": 1826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.691939838867186e-06,
+      "loss": 1.1067,
+      "step": 1827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.651254087762295e-06,
+      "loss": 1.1922,
+      "step": 1828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.610654762288439e-06,
+      "loss": 1.105,
+      "step": 1829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.57014194730485e-06,
+      "loss": 1.1776,
+      "step": 1830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.529715727489912e-06,
+      "loss": 1.1478,
+      "step": 1831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.489376187341036e-06,
+      "loss": 1.1535,
+      "step": 1832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.449123411174453e-06,
+      "loss": 1.1521,
+      "step": 1833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.408957483125024e-06,
+      "loss": 1.1024,
+      "step": 1834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.36887848714612e-06,
+      "loss": 1.0701,
+      "step": 1835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.328886507009359e-06,
+      "loss": 1.1988,
+      "step": 1836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.288981626304516e-06,
+      "loss": 1.114,
+      "step": 1837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.249163928439325e-06,
+      "loss": 1.0417,
+      "step": 1838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.209433496639274e-06,
+      "loss": 1.2415,
+      "step": 1839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.169790413947443e-06,
+      "loss": 1.1178,
+      "step": 1840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.130234763224343e-06,
+      "loss": 1.1037,
+      "step": 1841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.090766627147755e-06,
+      "loss": 1.1521,
+      "step": 1842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.051386088212526e-06,
+      "loss": 1.109,
+      "step": 1843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.012093228730415e-06,
+      "loss": 1.1623,
+      "step": 1844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.972888130829909e-06,
+      "loss": 1.1801,
+      "step": 1845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.93377087645607e-06,
+      "loss": 1.1731,
+      "step": 1846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.894741547370344e-06,
+      "loss": 1.1337,
+      "step": 1847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.855800225150417e-06,
+      "loss": 1.1714,
+      "step": 1848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.816946991189983e-06,
+      "loss": 1.1231,
+      "step": 1849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.778181926698658e-06,
+      "loss": 1.1546,
+      "step": 1850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.73950511270175e-06,
+      "loss": 1.1442,
+      "step": 1851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.70091663004015e-06,
+      "loss": 1.1492,
+      "step": 1852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.66241655937005e-06,
+      "loss": 1.1525,
+      "step": 1853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.624004981162903e-06,
+      "loss": 1.1702,
+      "step": 1854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.585681975705166e-06,
+      "loss": 1.0839,
+      "step": 1855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.547447623098192e-06,
+      "loss": 1.2305,
+      "step": 1856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.509302003258028e-06,
+      "loss": 1.1339,
+      "step": 1857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.471245195915211e-06,
+      "loss": 1.1212,
+      "step": 1858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.433277280614726e-06,
+      "loss": 1.0773,
+      "step": 1859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.395398336715697e-06,
+      "loss": 1.22,
+      "step": 1860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.357608443391311e-06,
+      "loss": 1.0823,
+      "step": 1861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.319907679628601e-06,
+      "loss": 1.1403,
+      "step": 1862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.282296124228333e-06,
+      "loss": 1.1158,
+      "step": 1863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.244773855804787e-06,
+      "loss": 1.1606,
+      "step": 1864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.207340952785641e-06,
+      "loss": 1.1792,
+      "step": 1865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.169997493411773e-06,
+      "loss": 1.0909,
+      "step": 1866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.132743555737114e-06,
+      "loss": 1.1317,
+      "step": 1867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.095579217628468e-06,
+      "loss": 1.1736,
+      "step": 1868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.058504556765383e-06,
+      "loss": 1.1304,
+      "step": 1869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0215196506399515e-06,
+      "loss": 1.0775,
+      "step": 1870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984624576556647e-06,
+      "loss": 1.126,
+      "step": 1871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.947819411632223e-06,
+      "loss": 1.1565,
+      "step": 1872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.911104232795479e-06,
+      "loss": 1.1262,
+      "step": 1873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.874479116787141e-06,
+      "loss": 1.2291,
+      "step": 1874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.837944140159658e-06,
+      "loss": 1.1048,
+      "step": 1875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.801499379277116e-06,
+      "loss": 1.2024,
+      "step": 1876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7651449103150136e-06,
+      "loss": 1.1488,
+      "step": 1877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.728880809260124e-06,
+      "loss": 1.1071,
+      "step": 1878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.692707151910343e-06,
+      "loss": 1.0882,
+      "step": 1879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.656624013874524e-06,
+      "loss": 1.156,
+      "step": 1880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.620631470572314e-06,
+      "loss": 1.1799,
+      "step": 1881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.584729597234008e-06,
+      "loss": 1.1766,
+      "step": 1882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.548918468900389e-06,
+      "loss": 1.144,
+      "step": 1883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.513198160422546e-06,
+      "loss": 1.1614,
+      "step": 1884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4775687464617595e-06,
+      "loss": 1.0339,
+      "step": 1885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.442030301489338e-06,
+      "loss": 1.0812,
+      "step": 1886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.406582899786445e-06,
+      "loss": 1.1909,
+      "step": 1887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3712266154439035e-06,
+      "loss": 1.1311,
+      "step": 1888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.335961522362144e-06,
+      "loss": 1.1834,
+      "step": 1889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.300787694250959e-06,
+      "loss": 1.1027,
+      "step": 1890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.265705204629402e-06,
+      "loss": 1.1766,
+      "step": 1891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.230714126825604e-06,
+      "loss": 1.1489,
+      "step": 1892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.195814533976624e-06,
+      "loss": 1.1004,
+      "step": 1893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.161006499028315e-06,
+      "loss": 1.226,
+      "step": 1894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.12629009473516e-06,
+      "loss": 1.0968,
+      "step": 1895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.091665393660112e-06,
+      "loss": 1.1492,
+      "step": 1896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.057132468174437e-06,
+      "loss": 1.1118,
+      "step": 1897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.022691390457596e-06,
+      "loss": 1.1726,
+      "step": 1898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.988342232497057e-06,
+      "loss": 1.1646,
+      "step": 1899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.954085066088205e-06,
+      "loss": 1.1459,
+      "step": 1900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.919919962834069e-06,
+      "loss": 1.162,
+      "step": 1901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.885846994145311e-06,
+      "loss": 1.115,
+      "step": 1902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.85186623124e-06,
+      "loss": 1.1304,
+      "step": 1903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.817977745143477e-06,
+      "loss": 1.1522,
+      "step": 1904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.78418160668821e-06,
+      "loss": 1.1839,
+      "step": 1905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.750477886513656e-06,
+      "loss": 1.1447,
+      "step": 1906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.716866655066083e-06,
+      "loss": 1.1635,
+      "step": 1907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6833479825984636e-06,
+      "loss": 1.1762,
+      "step": 1908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.649921939170294e-06,
+      "loss": 1.035,
+      "step": 1909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.61658859464746e-06,
+      "loss": 1.1305,
+      "step": 1910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.583348018702089e-06,
+      "loss": 1.0776,
+      "step": 1911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.550200280812412e-06,
+      "loss": 1.2045,
+      "step": 1912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.51714545026264e-06,
+      "loss": 1.0795,
+      "step": 1913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.484183596142739e-06,
+      "loss": 1.1271,
+      "step": 1914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.451314787348377e-06,
+      "loss": 1.1117,
+      "step": 1915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4185390925807264e-06,
+      "loss": 1.2348,
+      "step": 1916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.385856580346349e-06,
+      "loss": 1.135,
+      "step": 1917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.353267318957039e-06,
+      "loss": 1.1113,
+      "step": 1918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.320771376529643e-06,
+      "loss": 1.1667,
+      "step": 1919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2883688209860175e-06,
+      "loss": 1.0699,
+      "step": 1920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.256059720052786e-06,
+      "loss": 1.1724,
+      "step": 1921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.223844141261259e-06,
+      "loss": 1.1629,
+      "step": 1922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191722151947226e-06,
+      "loss": 1.168,
+      "step": 1923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.15969381925091e-06,
+      "loss": 1.1631,
+      "step": 1924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.12775921011675e-06,
+      "loss": 1.1276,
+      "step": 1925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.095918391293303e-06,
+      "loss": 1.155,
+      "step": 1926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.06417142933307e-06,
+      "loss": 1.1191,
+      "step": 1927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.032518390592389e-06,
+      "loss": 1.0917,
+      "step": 1928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.000959341231287e-06,
+      "loss": 1.1245,
+      "step": 1929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.969494347213316e-06,
+      "loss": 1.0793,
+      "step": 1930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938123474305473e-06,
+      "loss": 1.1414,
+      "step": 1931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.906846788077985e-06,
+      "loss": 1.1886,
+      "step": 1932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.875664353904225e-06,
+      "loss": 1.1289,
+      "step": 1933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.844576236960596e-06,
+      "loss": 1.1712,
+      "step": 1934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8135825022263494e-06,
+      "loss": 1.1582,
+      "step": 1935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.782683214483413e-06,
+      "loss": 1.1254,
+      "step": 1936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.75187843831637e-06,
+      "loss": 1.0932,
+      "step": 1937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.721168238112228e-06,
+      "loss": 1.151,
+      "step": 1938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.690552678060317e-06,
+      "loss": 1.1996,
+      "step": 1939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.660031822152161e-06,
+      "loss": 1.1609,
+      "step": 1940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.629605734181331e-06,
+      "loss": 1.112,
+      "step": 1941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.599274477743315e-06,
+      "loss": 1.1376,
+      "step": 1942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569038116235403e-06,
+      "loss": 1.0823,
+      "step": 1943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.538896712856527e-06,
+      "loss": 1.1616,
+      "step": 1944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.508850330607128e-06,
+      "loss": 1.1841,
+      "step": 1945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.478899032289047e-06,
+      "loss": 1.1308,
+      "step": 1946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.449042880505411e-06,
+      "loss": 1.1497,
+      "step": 1947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.419281937660458e-06,
+      "loss": 1.1437,
+      "step": 1948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.38961626595939e-06,
+      "loss": 1.139,
+      "step": 1949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.360045927408313e-06,
+      "loss": 1.0682,
+      "step": 1950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.330570983814075e-06,
+      "loss": 1.1328,
+      "step": 1951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.301191496784113e-06,
+      "loss": 1.1903,
+      "step": 1952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.271907527726354e-06,
+      "loss": 1.1216,
+      "step": 1953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.242719137849077e-06,
+      "loss": 1.1676,
+      "step": 1954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.21362638816078e-06,
+      "loss": 1.1242,
+      "step": 1955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.184629339470064e-06,
+      "loss": 1.1714,
+      "step": 1956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1557280523854995e-06,
+      "loss": 1.1593,
+      "step": 1957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.126922587315485e-06,
+      "loss": 1.183,
+      "step": 1958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0982130044681485e-06,
+      "loss": 1.1676,
+      "step": 1959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.069599363851207e-06,
+      "loss": 1.1301,
+      "step": 1960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.041081725271856e-06,
+      "loss": 1.0732,
+      "step": 1961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0126601483366e-06,
+      "loss": 1.1186,
+      "step": 1962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.984334692451181e-06,
+      "loss": 1.1444,
+      "step": 1963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.956105416820427e-06,
+      "loss": 1.1769,
+      "step": 1964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.927972380448136e-06,
+      "loss": 1.1853,
+      "step": 1965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.899935642136942e-06,
+      "loss": 1.1348,
+      "step": 1966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.871995260488215e-06,
+      "loss": 1.1362,
+      "step": 1967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8441512939019095e-06,
+      "loss": 1.0502,
+      "step": 1968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8164038005764676e-06,
+      "loss": 1.152,
+      "step": 1969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7887528385086855e-06,
+      "loss": 1.2026,
+      "step": 1970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7611984654935707e-06,
+      "loss": 1.1125,
+      "step": 1971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7337407391242695e-06,
+      "loss": 1.2139,
+      "step": 1972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7063797167919134e-06,
+      "loss": 1.1876,
+      "step": 1973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.679115455685528e-06,
+      "loss": 1.1056,
+      "step": 1974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6519480127918447e-06,
+      "loss": 1.1114,
+      "step": 1975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6248774448952695e-06,
+      "loss": 1.2073,
+      "step": 1976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5979038085777016e-06,
+      "loss": 1.0945,
+      "step": 1977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5710271602184486e-06,
+      "loss": 1.1633,
+      "step": 1978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5442475559941013e-06,
+      "loss": 1.0116,
+      "step": 1979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.517565051878391e-06,
+      "loss": 1.1569,
+      "step": 1980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4909797036421165e-06,
+      "loss": 1.2498,
+      "step": 1981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.464491566853001e-06,
+      "loss": 1.0741,
+      "step": 1982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.438100696875579e-06,
+      "loss": 1.2488,
+      "step": 1983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4118071488710656e-06,
+      "loss": 1.1197,
+      "step": 1984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.385610977797271e-06,
+      "loss": 1.0847,
+      "step": 1985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.35951223840848e-06,
+      "loss": 1.1395,
+      "step": 1986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3335109852553113e-06,
+      "loss": 1.202,
+      "step": 1987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3076072726846318e-06,
+      "loss": 1.0994,
+      "step": 1988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2818011548394366e-06,
+      "loss": 1.1381,
+      "step": 1989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.256092685658718e-06,
+      "loss": 1.1752,
+      "step": 1990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.230481918877382e-06,
+      "loss": 1.1023,
+      "step": 1991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2049689080261124e-06,
+      "loss": 1.1752,
+      "step": 1992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1795537064312596e-06,
+      "loss": 1.1131,
+      "step": 1993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.154236367214741e-06,
+      "loss": 1.1901,
+      "step": 1994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1290169432939553e-06,
+      "loss": 1.1567,
+      "step": 1995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.103895487381603e-06,
+      "loss": 1.1026,
+      "step": 1996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.078872051985615e-06,
+      "loss": 1.1075,
+      "step": 1997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.053946689409076e-06,
+      "loss": 1.0119,
+      "step": 1998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0291194517500564e-06,
+      "loss": 1.1997,
+      "step": 1999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0043903909015404e-06,
+      "loss": 1.0858,
+      "step": 2000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9797595585512985e-06,
+      "loss": 1.2405,
+      "step": 2001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.955227006181793e-06,
+      "loss": 1.0193,
+      "step": 2002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.930792785070069e-06,
+      "loss": 1.2048,
+      "step": 2003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.906456946287639e-06,
+      "loss": 1.1341,
+      "step": 2004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8822195407003817e-06,
+      "loss": 1.1012,
+      "step": 2005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.858080618968423e-06,
+      "loss": 1.1412,
+      "step": 2006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8340402315460426e-06,
+      "loss": 1.181,
+      "step": 2007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8100984286816013e-06,
+      "loss": 1.1324,
+      "step": 2008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7862552604173753e-06,
+      "loss": 1.0932,
+      "step": 2009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7625107765894608e-06,
+      "loss": 1.1657,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.738865026827714e-06,
+      "loss": 1.1016,
+      "step": 2011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7153180605556104e-06,
+      "loss": 1.1994,
+      "step": 2012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.691869926990165e-06,
+      "loss": 1.1663,
+      "step": 2013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6685206751417947e-06,
+      "loss": 1.0956,
+      "step": 2014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6452703538142556e-06,
+      "loss": 1.1452,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6221190116045125e-06,
+      "loss": 1.1996,
+      "step": 2016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5990666969026577e-06,
+      "loss": 1.1398,
+      "step": 2017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.576113457891788e-06,
+      "loss": 1.1901,
+      "step": 2018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5532593425479146e-06,
+      "loss": 1.0943,
+      "step": 2019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5305043986398824e-06,
+      "loss": 1.1076,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.507848673729224e-06,
+      "loss": 1.0601,
+      "step": 2021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4852922151701265e-06,
+      "loss": 1.1253,
+      "step": 2022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4628350701092595e-06,
+      "loss": 1.2353,
+      "step": 2023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.440477285485715e-06,
+      "loss": 1.1387,
+      "step": 2024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.418218908030928e-06,
+      "loss": 1.0961,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3960599842685394e-06,
+      "loss": 1.0981,
+      "step": 2026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3740005605143167e-06,
+      "loss": 1.1435,
+      "step": 2027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3520406828760443e-06,
+      "loss": 1.1928,
+      "step": 2028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.330180397253473e-06,
+      "loss": 1.2515,
+      "step": 2029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3084197493381586e-06,
+      "loss": 1.1159,
+      "step": 2030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2867587846134186e-06,
+      "loss": 1.1229,
+      "step": 2031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2651975483541923e-06,
+      "loss": 1.1578,
+      "step": 2032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.243736085626991e-06,
+      "loss": 1.1281,
+      "step": 2033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.222374441289771e-06,
+      "loss": 1.1156,
+      "step": 2034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2011126599918775e-06,
+      "loss": 1.1492,
+      "step": 2035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.179950786173879e-06,
+      "loss": 1.1531,
+      "step": 2036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1588888640675653e-06,
+      "loss": 1.1499,
+      "step": 2037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.137926937695778e-06,
+      "loss": 1.2096,
+      "step": 2038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.11706505087238e-06,
+      "loss": 1.0884,
+      "step": 2039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0963032472021247e-06,
+      "loss": 1.1612,
+      "step": 2040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0756415700805498e-06,
+      "loss": 1.1997,
+      "step": 2041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0550800626939537e-06,
+      "loss": 1.0692,
+      "step": 2042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0346187680192364e-06,
+      "loss": 1.1745,
+      "step": 2043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0142577288238538e-06,
+      "loss": 1.1355,
+      "step": 2044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9939969876656972e-06,
+      "loss": 1.0346,
+      "step": 2045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.973836586893019e-06,
+      "loss": 1.2013,
+      "step": 2046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9537765686443498e-06,
+      "loss": 1.1362,
+      "step": 2047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9338169748484123e-06,
+      "loss": 1.1581,
+      "step": 2048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.913957847224007e-06,
+      "loss": 1.1603,
+      "step": 2049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8941992272799635e-06,
+      "loss": 1.0249,
+      "step": 2050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8745411563150128e-06,
+      "loss": 1.1549,
+      "step": 2051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8549836754177429e-06,
+      "loss": 1.2065,
+      "step": 2052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8355268254664825e-06,
+      "loss": 1.1021,
+      "step": 2053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8161706471292172e-06,
+      "loss": 1.2449,
+      "step": 2054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.796915180863523e-06,
+      "loss": 1.0778,
+      "step": 2055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7777604669164726e-06,
+      "loss": 1.1383,
+      "step": 2056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7587065453245455e-06,
+      "loss": 1.1434,
+      "step": 2057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7397534559135452e-06,
+      "loss": 1.1903,
+      "step": 2058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7209012382985278e-06,
+      "loss": 1.0522,
+      "step": 2059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7021499318837064e-06,
+      "loss": 1.1416,
+      "step": 2060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6834995758623684e-06,
+      "loss": 1.216,
+      "step": 2061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6649502092168034e-06,
+      "loss": 1.0928,
+      "step": 2062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6465018707182255e-06,
+      "loss": 1.1473,
+      "step": 2063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6281545989266678e-06,
+      "loss": 1.1458,
+      "step": 2064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6099084321909264e-06,
+      "loss": 1.1557,
+      "step": 2065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5917634086484778e-06,
+      "loss": 1.1302,
+      "step": 2066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5737195662253623e-06,
+      "loss": 1.0872,
+      "step": 2067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5557769426361724e-06,
+      "loss": 1.2067,
+      "step": 2068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5379355753839197e-06,
+      "loss": 1.1553,
+      "step": 2069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5201955017599801e-06,
+      "loss": 1.1164,
+      "step": 2070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5025567588439926e-06,
+      "loss": 1.107,
+      "step": 2071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4850193835038218e-06,
+      "loss": 1.2461,
+      "step": 2072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4675834123954347e-06,
+      "loss": 1.0767,
+      "step": 2073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.450248881962868e-06,
+      "loss": 1.0658,
+      "step": 2074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4330158284381167e-06,
+      "loss": 1.1019,
+      "step": 2075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4158842878410738e-06,
+      "loss": 1.1203,
+      "step": 2076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3988542959794627e-06,
+      "loss": 1.241,
+      "step": 2077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.381925888448743e-06,
+      "loss": 1.1408,
+      "step": 2078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3650991006320558e-06,
+      "loss": 1.1875,
+      "step": 2079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3483739677001284e-06,
+      "loss": 1.1366,
+      "step": 2080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.331750524611225e-06,
+      "loss": 1.1552,
+      "step": 2081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3152288061110518e-06,
+      "loss": 1.0605,
+      "step": 2082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.298808846732702e-06,
+      "loss": 1.2116,
+      "step": 2083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2824906807965664e-06,
+      "loss": 1.1102,
+      "step": 2084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2662743424102896e-06,
+      "loss": 1.18,
+      "step": 2085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.250159865468653e-06,
+      "loss": 1.195,
+      "step": 2086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.234147283653553e-06,
+      "loss": 1.106,
+      "step": 2087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2182366304339e-06,
+      "loss": 1.1854,
+      "step": 2088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.202427939065548e-06,
+      "loss": 1.1741,
+      "step": 2089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1867212425912489e-06,
+      "loss": 1.1012,
+      "step": 2090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.171116573840575e-06,
+      "loss": 1.1548,
+      "step": 2091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1556139654298192e-06,
+      "loss": 1.0886,
+      "step": 2092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1402134497619621e-06,
+      "loss": 1.1701,
+      "step": 2093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1249150590266044e-06,
+      "loss": 1.0498,
+      "step": 2094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1097188251998792e-06,
+      "loss": 1.0697,
+      "step": 2095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0946247800443898e-06,
+      "loss": 1.2065,
+      "step": 2096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0796329551091667e-06,
+      "loss": 1.1193,
+      "step": 2097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.064743381729566e-06,
+      "loss": 1.0673,
+      "step": 2098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0499560910272266e-06,
+      "loss": 1.188,
+      "step": 2099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0352711139100024e-06,
+      "loss": 1.1375,
+      "step": 2100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0206884810718964e-06,
+      "loss": 1.1122,
+      "step": 2101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0062082229929826e-06,
+      "loss": 1.2055,
+      "step": 2102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.918303699393728e-07,
+      "loss": 1.2033,
+      "step": 2103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.775549519631333e-07,
+      "loss": 1.1058,
+      "step": 2104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.633819989022075e-07,
+      "loss": 1.207,
+      "step": 2105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.493115403803876e-07,
+      "loss": 1.1364,
+      "step": 2106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.353436058072263e-07,
+      "loss": 1.07,
+      "step": 2107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.214782243779918e-07,
+      "loss": 1.2079,
+      "step": 2108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.077154250735909e-07,
+      "loss": 1.166,
+      "step": 2109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.940552366605237e-07,
+      "loss": 1.1154,
+      "step": 2110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.804976876908178e-07,
+      "loss": 1.1762,
+      "step": 2111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.670428065019609e-07,
+      "loss": 1.1147,
+      "step": 2112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.536906212168572e-07,
+      "loss": 1.1309,
+      "step": 2113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.404411597437434e-07,
+      "loss": 1.2587,
+      "step": 2114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.27294449776167e-07,
+      "loss": 1.0057,
+      "step": 2115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.142505187928862e-07,
+      "loss": 1.1515,
+      "step": 2116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.013093940578698e-07,
+      "loss": 1.129,
+      "step": 2117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.884711026201585e-07,
+      "loss": 1.0914,
+      "step": 2118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.75735671313893e-07,
+      "loss": 1.2334,
+      "step": 2119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.631031267581967e-07,
+      "loss": 1.138,
+      "step": 2120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.505734953571486e-07,
+      "loss": 1.1115,
+      "step": 2121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.381468032997385e-07,
+      "loss": 1.1032,
+      "step": 2122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.258230765597618e-07,
+      "loss": 1.0905,
+      "step": 2123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.136023408958304e-07,
+      "loss": 1.125,
+      "step": 2124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.014846218512783e-07,
+      "loss": 1.2086,
+      "step": 2125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.894699447541231e-07,
+      "loss": 1.1645,
+      "step": 2126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.775583347169878e-07,
+      "loss": 1.1364,
+      "step": 2127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.657498166370901e-07,
+      "loss": 1.1792,
+      "step": 2128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.540444151961533e-07,
+      "loss": 1.1055,
+      "step": 2129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.424421548603954e-07,
+      "loss": 1.1261,
+      "step": 2130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.309430598804234e-07,
+      "loss": 1.1959,
+      "step": 2131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.195471542912168e-07,
+      "loss": 1.0766,
+      "step": 2132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.082544619120944e-07,
+      "loss": 1.1316,
+      "step": 2133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.970650063466143e-07,
+      "loss": 1.2018,
+      "step": 2134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.859788109825793e-07,
+      "loss": 1.1154,
+      "step": 2135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.749958989919313e-07,
+      "loss": 1.1903,
+      "step": 2136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.641162933307575e-07,
+      "loss": 1.1304,
+      "step": 2137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.533400167392178e-07,
+      "loss": 1.1099,
+      "step": 2138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.426670917414834e-07,
+      "loss": 1.0326,
+      "step": 2139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.320975406457096e-07,
+      "loss": 1.2079,
+      "step": 2140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.216313855439859e-07,
+      "loss": 1.0286,
+      "step": 2141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.112686483122797e-07,
+      "loss": 1.1691,
+      "step": 2142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.010093506104152e-07,
+      "loss": 1.1774,
+      "step": 2143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.908535138819892e-07,
+      "loss": 1.1082,
+      "step": 2144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.80801159354366e-07,
+      "loss": 1.1433,
+      "step": 2145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7085230803860513e-07,
+      "loss": 1.1123,
+      "step": 2146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.610069807294337e-07,
+      "loss": 1.1589,
+      "step": 2147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.512651980051907e-07,
+      "loss": 1.1328,
+      "step": 2148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4162698022778836e-07,
+      "loss": 1.1349,
+      "step": 2149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.320923475426786e-07,
+      "loss": 1.1428,
+      "step": 2150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.226613198788088e-07,
+      "loss": 1.1524,
+      "step": 2151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.133339169485606e-07,
+      "loss": 1.1486,
+      "step": 2152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.041101582477225e-07,
+      "loss": 1.1484,
+      "step": 2153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9499006305546703e-07,
+      "loss": 1.2149,
+      "step": 2154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8597365043427926e-07,
+      "loss": 1.09,
+      "step": 2155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7706093922992845e-07,
+      "loss": 1.1983,
+      "step": 2156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6825194807144083e-07,
+      "loss": 1.1348,
+      "step": 2157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.595466953710491e-07,
+      "loss": 1.1231,
+      "step": 2158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5094519932415417e-07,
+      "loss": 1.1235,
+      "step": 2159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.424474779092746e-07,
+      "loss": 1.1834,
+      "step": 2160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3405354888805254e-07,
+      "loss": 1.1156,
+      "step": 2161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2576342980515926e-07,
+      "loss": 1.1492,
+      "step": 2162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.175771379883008e-07,
+      "loss": 1.1326,
+      "step": 2163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0949469054816216e-07,
+      "loss": 1.1295,
+      "step": 2164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.015161043783854e-07,
+      "loss": 1.0206,
+      "step": 2165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9364139615550843e-07,
+      "loss": 1.1339,
+      "step": 2166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8587058233895956e-07,
+      "loss": 1.1832,
+      "step": 2167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.78203679171013e-07,
+      "loss": 1.143,
+      "step": 2168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7064070267674433e-07,
+      "loss": 1.1708,
+      "step": 2169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.631816686640143e-07,
+      "loss": 1.1167,
+      "step": 2170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5582659272341824e-07,
+      "loss": 1.1214,
+      "step": 2171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4857549022826444e-07,
+      "loss": 1.1547,
+      "step": 2172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.41428376334546e-07,
+      "loss": 1.0015,
+      "step": 2173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.343852659808965e-07,
+      "loss": 1.218,
+      "step": 2174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.274461738885625e-07,
+      "loss": 1.1199,
+      "step": 2175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.206111145613865e-07,
+      "loss": 1.1699,
+      "step": 2176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1388010228575173e-07,
+      "loss": 1.1025,
+      "step": 2177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0725315113057643e-07,
+      "loss": 1.1844,
+      "step": 2178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.007302749472695e-07,
+      "loss": 1.1361,
+      "step": 2179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9431148736971383e-07,
+      "loss": 1.1732,
+      "step": 2180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8799680181421088e-07,
+      "loss": 1.1561,
+      "step": 2181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8178623147949715e-07,
+      "loss": 1.0959,
+      "step": 2182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7567978934667219e-07,
+      "loss": 1.1357,
+      "step": 2183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6967748817919293e-07,
+      "loss": 1.152,
+      "step": 2184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.637793405228627e-07,
+      "loss": 1.1997,
+      "step": 2185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.579853587057589e-07,
+      "loss": 1.1064,
+      "step": 2186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5229555483826097e-07,
+      "loss": 1.1615,
+      "step": 2187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4670994081297795e-07,
+      "loss": 1.1511,
+      "step": 2188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4122852830475984e-07,
+      "loss": 0.9507,
+      "step": 2189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3585132877065865e-07,
+      "loss": 1.1326,
+      "step": 2190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.30578353449895e-07,
+      "loss": 1.1414,
+      "step": 2191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2540961336384717e-07,
+      "loss": 1.1978,
+      "step": 2192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2034511931602322e-07,
+      "loss": 1.1588,
+      "step": 2193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1538488189204998e-07,
+      "loss": 1.0975,
+      "step": 2194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1052891145963417e-07,
+      "loss": 1.102,
+      "step": 2195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0577721816854013e-07,
+      "loss": 1.1881,
+      "step": 2196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0112981195058435e-07,
+      "loss": 1.1432,
+      "step": 2197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.658670251960766e-08,
+      "loss": 1.0324,
+      "step": 2198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.214789937144752e-08,
+      "loss": 1.1699,
+      "step": 2199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.781341178393244e-08,
+      "loss": 1.1127,
+      "step": 2200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.358324881684309e-08,
+      "loss": 1.1301,
+      "step": 2201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.945741931191242e-08,
+      "loss": 1.1451,
+      "step": 2202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.543593189280329e-08,
+      "loss": 1.1563,
+      "step": 2203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.15187949650753e-08,
+      "loss": 1.1648,
+      "step": 2204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.77060167161847e-08,
+      "loss": 1.1063,
+      "step": 2205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.399760511546226e-08,
+      "loss": 1.1493,
+      "step": 2206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.0393567914091e-08,
+      "loss": 1.1735,
+      "step": 2207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.689391264510069e-08,
+      "loss": 1.1205,
+      "step": 2208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.349864662334558e-08,
+      "loss": 1.1448,
+      "step": 2209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0207776945476734e-08,
+      "loss": 1.0309,
+      "step": 2210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.702131048995306e-08,
+      "loss": 1.1321,
+      "step": 2211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.393925391701359e-08,
+      "loss": 1.1298,
+      "step": 2212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0961613668660804e-08,
+      "loss": 1.1822,
+      "step": 2213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8088395968638443e-08,
+      "loss": 1.1466,
+      "step": 2214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.531960682245372e-08,
+      "loss": 1.1211,
+      "step": 2215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.265525201731623e-08,
+      "loss": 1.2312,
+      "step": 2216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.009533712217127e-08,
+      "loss": 1.0992,
+      "step": 2217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7639867487655458e-08,
+      "loss": 1.1596,
+      "step": 2218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5288848246102227e-08,
+      "loss": 1.148,
+      "step": 2219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.304228431152522e-08,
+      "loss": 1.1348,
+      "step": 2220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0900180379607172e-08,
+      "loss": 1.1467,
+      "step": 2221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.886254092770545e-08,
+      "loss": 1.0464,
+      "step": 2222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.692937021480212e-08,
+      "loss": 1.1282,
+      "step": 2223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5100672281553873e-08,
+      "loss": 1.1449,
+      "step": 2224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3376450950236541e-08,
+      "loss": 1.2058,
+      "step": 2225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1756709824750633e-08,
+      "loss": 1.0942,
+      "step": 2226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0241452290615794e-08,
+      "loss": 1.2451,
+      "step": 2227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.830681514970795e-09,
+      "loss": 1.1541,
+      "step": 2228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.52440044656244e-09,
+      "loss": 1.1066,
+      "step": 2229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.322611815728907e-09,
+      "loss": 0.9908,
+      "step": 2230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2253181343997485e-09,
+      "loss": 1.133,
+      "step": 2231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2325216961014435e-09,
+      "loss": 1.1898,
+      "step": 2232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3442245759407464e-09,
+      "loss": 1.1293,
+      "step": 2233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5604286305991322e-09,
+      "loss": 1.1853,
+      "step": 2234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.881135498349451e-09,
+      "loss": 1.1126,
+      "step": 2235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.306346599011521e-09,
+      "loss": 1.1484,
+      "step": 2236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.360631340020853e-10,
+      "loss": 1.1115,
+      "step": 2237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.702860862793034e-10,
+      "loss": 1.2353,
+      "step": 2238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0901622037605707e-10,
+      "loss": 1.0539,
+      "step": 2239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2254082399949553e-11,
+      "loss": 1.1163,
+      "step": 2240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0716,
+      "step": 2241
+    },
+    {
+      "epoch": 1.0,
+      "step": 2241,
+      "total_flos": 0.0,
+      "train_loss": 0.6425575492957616,
+      "train_runtime": 13487.3058,
+      "train_samples_per_second": 340.934,
+      "train_steps_per_second": 0.166
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2241,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}