diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,69978 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9998284587014323,
+  "eval_steps": 500,
+  "global_step": 11658,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428572e-08,
+      "loss": 1.0623,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7142857142857143e-07,
+      "loss": 1.0099,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5714285714285716e-07,
+      "loss": 0.9594,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4285714285714286e-07,
+      "loss": 1.0325,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2857142857142857e-07,
+      "loss": 0.9535,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.142857142857143e-07,
+      "loss": 1.0188,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.0435,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.857142857142857e-07,
+      "loss": 0.9946,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.714285714285715e-07,
+      "loss": 0.9755,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428571e-07,
+      "loss": 0.9476,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.42857142857143e-07,
+      "loss": 0.8912,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0285714285714286e-06,
+      "loss": 0.9569,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1142857142857143e-06,
+      "loss": 0.9781,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.0127,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2857142857142858e-06,
+      "loss": 0.9024,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3714285714285715e-06,
+      "loss": 0.8209,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4571428571428571e-06,
+      "loss": 0.8849,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.542857142857143e-06,
+      "loss": 0.8498,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6285714285714286e-06,
+      "loss": 0.8311,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7142857142857143e-06,
+      "loss": 0.8452,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 0.7644,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.885714285714286e-06,
+      "loss": 0.8725,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9714285714285714e-06,
+      "loss": 0.7374,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0571428571428573e-06,
+      "loss": 0.754,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.7884,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2285714285714286e-06,
+      "loss": 0.7426,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.314285714285714e-06,
+      "loss": 0.8866,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.7481,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4857142857142858e-06,
+      "loss": 0.7179,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5714285714285716e-06,
+      "loss": 0.9086,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.657142857142857e-06,
+      "loss": 0.7711,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.742857142857143e-06,
+      "loss": 0.715,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8285714285714288e-06,
+      "loss": 0.7261,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9142857142857142e-06,
+      "loss": 0.6876,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 0.6682,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.085714285714286e-06,
+      "loss": 0.6282,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.171428571428572e-06,
+      "loss": 0.639,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2571428571428572e-06,
+      "loss": 0.6755,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.342857142857143e-06,
+      "loss": 0.6577,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4285714285714285e-06,
+      "loss": 0.6769,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5142857142857144e-06,
+      "loss": 0.693,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 0.6436,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6857142857142857e-06,
+      "loss": 0.6752,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.771428571428572e-06,
+      "loss": 0.611,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.857142857142857e-06,
+      "loss": 0.6888,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942857142857143e-06,
+      "loss": 0.6257,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.028571428571429e-06,
+      "loss": 0.6553,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.114285714285715e-06,
+      "loss": 0.6193,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.6623,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.5733,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.371428571428571e-06,
+      "loss": 0.6274,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.457142857142857e-06,
+      "loss": 0.5903,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.542857142857143e-06,
+      "loss": 0.6129,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.628571428571428e-06,
+      "loss": 0.635,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.714285714285714e-06,
+      "loss": 0.5916,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.6136,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8857142857142865e-06,
+      "loss": 0.6205,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9714285714285715e-06,
+      "loss": 0.6064,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.057142857142857e-06,
+      "loss": 0.6479,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.142857142857143e-06,
+      "loss": 0.6092,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.228571428571429e-06,
+      "loss": 0.5679,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.314285714285714e-06,
+      "loss": 0.5915,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-06,
+      "loss": 0.5215,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.485714285714286e-06,
+      "loss": 0.5994,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.571428571428572e-06,
+      "loss": 0.6006,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6571428571428576e-06,
+      "loss": 0.5598,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7428571428571426e-06,
+      "loss": 0.6075,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8285714285714284e-06,
+      "loss": 0.5594,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.914285714285714e-06,
+      "loss": 0.5196,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.5832,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.085714285714285e-06,
+      "loss": 0.5883,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.171428571428572e-06,
+      "loss": 0.6144,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.257142857142858e-06,
+      "loss": 0.6431,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.342857142857144e-06,
+      "loss": 0.5511,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.428571428571429e-06,
+      "loss": 0.6023,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5142857142857145e-06,
+      "loss": 0.5744,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6e-06,
+      "loss": 0.5936,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.685714285714286e-06,
+      "loss": 0.6012,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.771428571428571e-06,
+      "loss": 0.5789,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.857142857142857e-06,
+      "loss": 0.5878,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.942857142857143e-06,
+      "loss": 0.5738,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.028571428571429e-06,
+      "loss": 0.5947,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.114285714285715e-06,
+      "loss": 0.5847,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-06,
+      "loss": 0.5989,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2857142857142855e-06,
+      "loss": 0.5511,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.371428571428571e-06,
+      "loss": 0.5422,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.457142857142857e-06,
+      "loss": 0.6246,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.542857142857144e-06,
+      "loss": 0.6645,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.628571428571429e-06,
+      "loss": 0.6478,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.714285714285714e-06,
+      "loss": 0.5838,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.8e-06,
+      "loss": 0.5493,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.885714285714286e-06,
+      "loss": 0.5634,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.971428571428572e-06,
+      "loss": 0.5463,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.057142857142857e-06,
+      "loss": 0.5615,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.142857142857142e-06,
+      "loss": 0.5844,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.22857142857143e-06,
+      "loss": 0.5386,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.314285714285715e-06,
+      "loss": 0.4926,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.5427,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.485714285714287e-06,
+      "loss": 0.6117,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.5599,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.657142857142858e-06,
+      "loss": 0.5637,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.742857142857143e-06,
+      "loss": 0.5817,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.828571428571429e-06,
+      "loss": 0.5947,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.914285714285714e-06,
+      "loss": 0.4862,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9e-06,
+      "loss": 0.5664,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.085714285714286e-06,
+      "loss": 0.5299,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.171428571428572e-06,
+      "loss": 0.5597,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.257142857142856e-06,
+      "loss": 0.6039,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.342857142857144e-06,
+      "loss": 0.5693,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.428571428571428e-06,
+      "loss": 0.5651,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.514285714285714e-06,
+      "loss": 0.5301,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.5401,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.685714285714285e-06,
+      "loss": 0.5269,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.771428571428573e-06,
+      "loss": 0.5388,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.857142857142857e-06,
+      "loss": 0.5198,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.942857142857143e-06,
+      "loss": 0.5313,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0028571428571429e-05,
+      "loss": 0.5319,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0114285714285715e-05,
+      "loss": 0.5467,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.02e-05,
+      "loss": 0.5401,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0285714285714286e-05,
+      "loss": 0.5621,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.037142857142857e-05,
+      "loss": 0.5056,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0457142857142858e-05,
+      "loss": 0.5443,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0542857142857142e-05,
+      "loss": 0.5424,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0628571428571428e-05,
+      "loss": 0.4561,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.5775,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.08e-05,
+      "loss": 0.5227,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0885714285714286e-05,
+      "loss": 0.5268,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0971428571428572e-05,
+      "loss": 0.5716,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1057142857142858e-05,
+      "loss": 0.5559,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.5268,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.122857142857143e-05,
+      "loss": 0.5586,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1314285714285715e-05,
+      "loss": 0.4869,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 0.5741,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1485714285714285e-05,
+      "loss": 0.5364,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 0.5801,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1657142857142857e-05,
+      "loss": 0.5632,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1742857142857143e-05,
+      "loss": 0.5204,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1828571428571429e-05,
+      "loss": 0.5649,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1914285714285714e-05,
+      "loss": 0.5236,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5224,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2085714285714286e-05,
+      "loss": 0.5237,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.217142857142857e-05,
+      "loss": 0.5158,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2257142857142858e-05,
+      "loss": 0.5276,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2342857142857144e-05,
+      "loss": 0.5646,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 0.5462,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2514285714285715e-05,
+      "loss": 0.5166,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.26e-05,
+      "loss": 0.5065,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2685714285714287e-05,
+      "loss": 0.5113,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2771428571428571e-05,
+      "loss": 0.5333,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.5497,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2942857142857143e-05,
+      "loss": 0.5422,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3028571428571429e-05,
+      "loss": 0.5102,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3114285714285715e-05,
+      "loss": 0.5546,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.32e-05,
+      "loss": 0.5519,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3285714285714285e-05,
+      "loss": 0.5565,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3371428571428572e-05,
+      "loss": 0.5023,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3457142857142857e-05,
+      "loss": 0.5339,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3542857142857142e-05,
+      "loss": 0.5327,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3628571428571428e-05,
+      "loss": 0.5894,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3714285714285714e-05,
+      "loss": 0.5418,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 0.5275,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3885714285714286e-05,
+      "loss": 0.4696,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3971428571428572e-05,
+      "loss": 0.5028,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4057142857142858e-05,
+      "loss": 0.5207,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4142857142857143e-05,
+      "loss": 0.4501,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.422857142857143e-05,
+      "loss": 0.5589,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4314285714285715e-05,
+      "loss": 0.5182,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.44e-05,
+      "loss": 0.5501,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4485714285714287e-05,
+      "loss": 0.5076,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4571428571428571e-05,
+      "loss": 0.4686,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4657142857142857e-05,
+      "loss": 0.5308,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4742857142857143e-05,
+      "loss": 0.5223,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4828571428571429e-05,
+      "loss": 0.5501,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4914285714285715e-05,
+      "loss": 0.4994,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5e-05,
+      "loss": 0.5448,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5085714285714288e-05,
+      "loss": 0.5728,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.517142857142857e-05,
+      "loss": 0.5193,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5257142857142858e-05,
+      "loss": 0.4982,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5342857142857146e-05,
+      "loss": 0.488,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5428571428571428e-05,
+      "loss": 0.5192,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5514285714285714e-05,
+      "loss": 0.5675,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.56e-05,
+      "loss": 0.4924,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.568571428571429e-05,
+      "loss": 0.5836,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.577142857142857e-05,
+      "loss": 0.5136,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 0.5625,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5942857142857143e-05,
+      "loss": 0.493,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.602857142857143e-05,
+      "loss": 0.5002,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6114285714285715e-05,
+      "loss": 0.5434,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.62e-05,
+      "loss": 0.5276,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6285714285714283e-05,
+      "loss": 0.5148,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6371428571428572e-05,
+      "loss": 0.5517,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.645714285714286e-05,
+      "loss": 0.5478,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.654285714285714e-05,
+      "loss": 0.5448,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.662857142857143e-05,
+      "loss": 0.5644,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 0.6163,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.5334,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6885714285714284e-05,
+      "loss": 0.4924,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6971428571428574e-05,
+      "loss": 0.5164,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.705714285714286e-05,
+      "loss": 0.5009,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.4962,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7228571428571428e-05,
+      "loss": 0.5022,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7314285714285717e-05,
+      "loss": 0.4748,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.74e-05,
+      "loss": 0.5328,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7485714285714285e-05,
+      "loss": 0.5261,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 0.5161,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7657142857142857e-05,
+      "loss": 0.468,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7742857142857143e-05,
+      "loss": 0.5796,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.782857142857143e-05,
+      "loss": 0.5005,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7914285714285718e-05,
+      "loss": 0.4758,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5419,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8085714285714286e-05,
+      "loss": 0.5009,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8171428571428572e-05,
+      "loss": 0.4452,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8257142857142858e-05,
+      "loss": 0.4919,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8342857142857144e-05,
+      "loss": 0.5234,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 0.5574,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8514285714285712e-05,
+      "loss": 0.4718,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.86e-05,
+      "loss": 0.5384,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8685714285714287e-05,
+      "loss": 0.5012,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.877142857142857e-05,
+      "loss": 0.517,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8857142857142856e-05,
+      "loss": 0.5101,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8942857142857145e-05,
+      "loss": 0.4982,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9028571428571427e-05,
+      "loss": 0.5509,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9114285714285713e-05,
+      "loss": 0.5062,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.5515,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.5224,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.937142857142857e-05,
+      "loss": 0.5381,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9457142857142857e-05,
+      "loss": 0.4744,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9542857142857146e-05,
+      "loss": 0.4901,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.962857142857143e-05,
+      "loss": 0.5103,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9714285714285714e-05,
+      "loss": 0.4818,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.98e-05,
+      "loss": 0.4811,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9885714285714286e-05,
+      "loss": 0.5426,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9971428571428572e-05,
+      "loss": 0.5273,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0057142857142858e-05,
+      "loss": 0.5157,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.014285714285714e-05,
+      "loss": 0.484,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.022857142857143e-05,
+      "loss": 0.5372,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0314285714285715e-05,
+      "loss": 0.4763,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.04e-05,
+      "loss": 0.5354,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0485714285714287e-05,
+      "loss": 0.4889,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0571428571428573e-05,
+      "loss": 0.5476,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.065714285714286e-05,
+      "loss": 0.5429,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.074285714285714e-05,
+      "loss": 0.5002,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.082857142857143e-05,
+      "loss": 0.504,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0914285714285716e-05,
+      "loss": 0.4807,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1e-05,
+      "loss": 0.5607,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1085714285714285e-05,
+      "loss": 0.5391,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1171428571428574e-05,
+      "loss": 0.5335,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1257142857142856e-05,
+      "loss": 0.4899,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1342857142857142e-05,
+      "loss": 0.5588,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.4946,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1514285714285717e-05,
+      "loss": 0.5179,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.16e-05,
+      "loss": 0.5028,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1685714285714286e-05,
+      "loss": 0.5576,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.177142857142857e-05,
+      "loss": 0.4906,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1857142857142857e-05,
+      "loss": 0.5101,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1942857142857143e-05,
+      "loss": 0.5106,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.202857142857143e-05,
+      "loss": 0.5681,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2114285714285715e-05,
+      "loss": 0.4791,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.22e-05,
+      "loss": 0.5403,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2285714285714287e-05,
+      "loss": 0.4911,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.237142857142857e-05,
+      "loss": 0.5419,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.245714285714286e-05,
+      "loss": 0.5266,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2542857142857144e-05,
+      "loss": 0.5639,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.262857142857143e-05,
+      "loss": 0.5412,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2714285714285713e-05,
+      "loss": 0.5306,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.5209,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2885714285714288e-05,
+      "loss": 0.5076,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.297142857142857e-05,
+      "loss": 0.5049,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3057142857142856e-05,
+      "loss": 0.5135,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3142857142857145e-05,
+      "loss": 0.5186,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3228571428571428e-05,
+      "loss": 0.5448,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3314285714285714e-05,
+      "loss": 0.5228,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 0.4713,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3485714285714285e-05,
+      "loss": 0.5942,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 0.5018,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3657142857142857e-05,
+      "loss": 0.5554,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3742857142857146e-05,
+      "loss": 0.5054,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.382857142857143e-05,
+      "loss": 0.4917,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3914285714285715e-05,
+      "loss": 0.4917,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4e-05,
+      "loss": 0.5374,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4085714285714286e-05,
+      "loss": 0.5603,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4171428571428572e-05,
+      "loss": 0.4303,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4257142857142858e-05,
+      "loss": 0.4949,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.434285714285714e-05,
+      "loss": 0.4695,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.442857142857143e-05,
+      "loss": 0.5451,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4514285714285716e-05,
+      "loss": 0.4614,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 0.4901,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4685714285714288e-05,
+      "loss": 0.4969,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4771428571428573e-05,
+      "loss": 0.4819,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.485714285714286e-05,
+      "loss": 0.5566,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4942857142857142e-05,
+      "loss": 0.4867,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.502857142857143e-05,
+      "loss": 0.512,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5114285714285717e-05,
+      "loss": 0.5271,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.52e-05,
+      "loss": 0.5487,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5285714285714285e-05,
+      "loss": 0.4919,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5371428571428574e-05,
+      "loss": 0.5161,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5457142857142857e-05,
+      "loss": 0.4962,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5542857142857143e-05,
+      "loss": 0.4895,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.562857142857143e-05,
+      "loss": 0.4884,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 0.5072,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.58e-05,
+      "loss": 0.4842,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5885714285714286e-05,
+      "loss": 0.4598,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5971428571428575e-05,
+      "loss": 0.5021,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6057142857142858e-05,
+      "loss": 0.4854,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6142857142857144e-05,
+      "loss": 0.4889,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.622857142857143e-05,
+      "loss": 0.5065,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6314285714285715e-05,
+      "loss": 0.4799,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.64e-05,
+      "loss": 0.463,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6485714285714287e-05,
+      "loss": 0.468,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.657142857142857e-05,
+      "loss": 0.4916,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.665714285714286e-05,
+      "loss": 0.5743,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6742857142857145e-05,
+      "loss": 0.5421,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6828571428571427e-05,
+      "loss": 0.4967,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6914285714285713e-05,
+      "loss": 0.499,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.5155,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7085714285714285e-05,
+      "loss": 0.4872,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.717142857142857e-05,
+      "loss": 0.4869,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7257142857142857e-05,
+      "loss": 0.4804,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7342857142857146e-05,
+      "loss": 0.535,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7428571428571428e-05,
+      "loss": 0.5146,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7514285714285714e-05,
+      "loss": 0.4967,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 0.5384,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7685714285714286e-05,
+      "loss": 0.4947,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7771428571428572e-05,
+      "loss": 0.4343,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7857142857142858e-05,
+      "loss": 0.4472,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7942857142857143e-05,
+      "loss": 0.5484,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.802857142857143e-05,
+      "loss": 0.4978,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8114285714285715e-05,
+      "loss": 0.5227,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 0.5412,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8285714285714287e-05,
+      "loss": 0.5268,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8371428571428573e-05,
+      "loss": 0.4912,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.845714285714286e-05,
+      "loss": 0.5502,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.854285714285714e-05,
+      "loss": 0.5318,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.862857142857143e-05,
+      "loss": 0.5388,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8714285714285716e-05,
+      "loss": 0.4468,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.88e-05,
+      "loss": 0.5096,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8885714285714288e-05,
+      "loss": 0.5008,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8971428571428574e-05,
+      "loss": 0.5607,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9057142857142856e-05,
+      "loss": 0.5411,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9142857142857142e-05,
+      "loss": 0.5308,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.922857142857143e-05,
+      "loss": 0.5084,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9314285714285714e-05,
+      "loss": 0.5519,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.94e-05,
+      "loss": 0.5411,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9485714285714286e-05,
+      "loss": 0.5162,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9571428571428575e-05,
+      "loss": 0.517,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9657142857142857e-05,
+      "loss": 0.5125,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9742857142857143e-05,
+      "loss": 0.5003,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982857142857143e-05,
+      "loss": 0.4895,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9914285714285715e-05,
+      "loss": 0.4978,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3e-05,
+      "loss": 0.4769,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999999942111885e-05,
+      "loss": 0.5085,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999997684475455e-05,
+      "loss": 0.4796,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999999479006994e-05,
+      "loss": 0.5004,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999990737902527e-05,
+      "loss": 0.4783,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999985527973535e-05,
+      "loss": 0.5022,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999979160283367e-05,
+      "loss": 0.5079,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999971634832513e-05,
+      "loss": 0.5036,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999996295162155e-05,
+      "loss": 0.5021,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999953110651152e-05,
+      "loss": 0.5015,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999942111922077e-05,
+      "loss": 0.4936,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999929955435173e-05,
+      "loss": 0.4977,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999991664119138e-05,
+      "loss": 0.4646,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999902169191724e-05,
+      "loss": 0.5074,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999886539437322e-05,
+      "loss": 0.4936,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999869751929385e-05,
+      "loss": 0.5216,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999851806669208e-05,
+      "loss": 0.4661,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999983270365817e-05,
+      "loss": 0.5483,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999812442897746e-05,
+      "loss": 0.4871,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999791024389503e-05,
+      "loss": 0.4658,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.99997684481351e-05,
+      "loss": 0.4861,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999744714136274e-05,
+      "loss": 0.4914,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999719822394848e-05,
+      "loss": 0.5159,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999693772912758e-05,
+      "loss": 0.4926,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999666565692005e-05,
+      "loss": 0.5252,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999638200734693e-05,
+      "loss": 0.4666,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999960867804301e-05,
+      "loss": 0.4763,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999577997619234e-05,
+      "loss": 0.4588,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9999546159465735e-05,
+      "loss": 0.4626,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999951316358497e-05,
+      "loss": 0.5868,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999479009979487e-05,
+      "loss": 0.5541,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999443698651918e-05,
+      "loss": 0.4765,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999407229604993e-05,
+      "loss": 0.5656,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999936960284152e-05,
+      "loss": 0.5239,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999330818364414e-05,
+      "loss": 0.514,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999929087617666e-05,
+      "loss": 0.4655,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999249776281342e-05,
+      "loss": 0.5203,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999207518681633e-05,
+      "loss": 0.5267,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99991641033808e-05,
+      "loss": 0.4954,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999119530382186e-05,
+      "loss": 0.4891,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999073799689235e-05,
+      "loss": 0.4363,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9999026911305475e-05,
+      "loss": 0.444,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999897886523453e-05,
+      "loss": 0.4759,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998929661480098e-05,
+      "loss": 0.4819,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999887930004599e-05,
+      "loss": 0.4444,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998827780936085e-05,
+      "loss": 0.4923,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999877510415436e-05,
+      "loss": 0.4972,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999872126970488e-05,
+      "loss": 0.478,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998666277591804e-05,
+      "loss": 0.3863,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998610127819374e-05,
+      "loss": 0.485,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998552820391924e-05,
+      "loss": 0.4696,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999849435531388e-05,
+      "loss": 0.4688,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998434732589747e-05,
+      "loss": 0.449,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998373952224136e-05,
+      "loss": 0.472,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998312014221734e-05,
+      "loss": 0.5317,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998248918587324e-05,
+      "loss": 0.4778,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998184665325772e-05,
+      "loss": 0.5545,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9998119254442035e-05,
+      "loss": 0.5143,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999805268594117e-05,
+      "loss": 0.4496,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997984959828313e-05,
+      "loss": 0.5233,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999791607610869e-05,
+      "loss": 0.4987,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997846034787615e-05,
+      "loss": 0.4984,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997774835870495e-05,
+      "loss": 0.549,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997702479362828e-05,
+      "loss": 0.4588,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997628965270194e-05,
+      "loss": 0.485,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997554293598275e-05,
+      "loss": 0.5077,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999747846435283e-05,
+      "loss": 0.5014,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997401477539708e-05,
+      "loss": 0.503,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997323333164856e-05,
+      "loss": 0.4471,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997244031234308e-05,
+      "loss": 0.4872,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999716357175418e-05,
+      "loss": 0.4935,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9997081954730682e-05,
+      "loss": 0.466,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996999180170117e-05,
+      "loss": 0.4623,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999691524807887e-05,
+      "loss": 0.4662,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996830158463424e-05,
+      "loss": 0.5262,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996743911330342e-05,
+      "loss": 0.5437,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996656506686283e-05,
+      "loss": 0.5789,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996567944537993e-05,
+      "loss": 0.5079,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999647822489231e-05,
+      "loss": 0.5248,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999638734775615e-05,
+      "loss": 0.5127,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996295313136536e-05,
+      "loss": 0.4288,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996202121040572e-05,
+      "loss": 0.4765,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996107771475445e-05,
+      "loss": 0.526,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9996012264448445e-05,
+      "loss": 0.4493,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995915599966934e-05,
+      "loss": 0.5313,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999581777803838e-05,
+      "loss": 0.5292,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995718798670335e-05,
+      "loss": 0.5182,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9995618661870434e-05,
+      "loss": 0.4951,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995517367646405e-05,
+      "loss": 0.5204,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995414916006067e-05,
+      "loss": 0.4348,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995311306957334e-05,
+      "loss": 0.4514,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9995206540508197e-05,
+      "loss": 0.5017,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999510061666674e-05,
+      "loss": 0.5105,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994993535441145e-05,
+      "loss": 0.4962,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994885296839673e-05,
+      "loss": 0.5117,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994775900870678e-05,
+      "loss": 0.4812,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999466534754261e-05,
+      "loss": 0.4789,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999455363686399e-05,
+      "loss": 0.4953,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994440768843453e-05,
+      "loss": 0.4986,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994326743489698e-05,
+      "loss": 0.4801,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994211560811538e-05,
+      "loss": 0.5462,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9994095220817855e-05,
+      "loss": 0.4814,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993977723517635e-05,
+      "loss": 0.4745,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993859068919942e-05,
+      "loss": 0.5519,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993739257033937e-05,
+      "loss": 0.5327,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993618287868866e-05,
+      "loss": 0.4734,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993496161434064e-05,
+      "loss": 0.4713,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999337287773896e-05,
+      "loss": 0.5086,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999324843679307e-05,
+      "loss": 0.4562,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9993122838605995e-05,
+      "loss": 0.4565,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992996083187436e-05,
+      "loss": 0.5149,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992868170547175e-05,
+      "loss": 0.4651,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992739100695075e-05,
+      "loss": 0.4932,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992608873641112e-05,
+      "loss": 0.4594,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992477489395327e-05,
+      "loss": 0.4882,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992344947967868e-05,
+      "loss": 0.4491,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999221124936896e-05,
+      "loss": 0.4868,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9992076393608928e-05,
+      "loss": 0.5083,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999194038069817e-05,
+      "loss": 0.5196,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991803210647195e-05,
+      "loss": 0.4801,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999166488346659e-05,
+      "loss": 0.4807,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991525399167025e-05,
+      "loss": 0.5138,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991384757759268e-05,
+      "loss": 0.5719,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991242959254177e-05,
+      "loss": 0.4377,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9991100003662694e-05,
+      "loss": 0.5127,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990955890995853e-05,
+      "loss": 0.4615,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999081062126478e-05,
+      "loss": 0.4894,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990664194480685e-05,
+      "loss": 0.4382,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999051661065487e-05,
+      "loss": 0.4616,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9990367869798733e-05,
+      "loss": 0.5354,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999021797192374e-05,
+      "loss": 0.5312,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.999006691704147e-05,
+      "loss": 0.5155,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989914705163582e-05,
+      "loss": 0.4844,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989761336301823e-05,
+      "loss": 0.5242,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989606810468027e-05,
+      "loss": 0.4419,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998945112767413e-05,
+      "loss": 0.5109,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989294287932145e-05,
+      "loss": 0.4286,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989136291254172e-05,
+      "loss": 0.4485,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998897713765241e-05,
+      "loss": 0.4402,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988816827139142e-05,
+      "loss": 0.5017,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998865535972674e-05,
+      "loss": 0.4838,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988492735427673e-05,
+      "loss": 0.4861,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998832895425449e-05,
+      "loss": 0.518,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988164016219827e-05,
+      "loss": 0.4556,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987997921336422e-05,
+      "loss": 0.4751,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987830669617085e-05,
+      "loss": 0.5148,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987662261074735e-05,
+      "loss": 0.507,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9987492695722368e-05,
+      "loss": 0.4712,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998732197357307e-05,
+      "loss": 0.5032,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998715009464002e-05,
+      "loss": 0.4919,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986977058936488e-05,
+      "loss": 0.4837,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986802866475817e-05,
+      "loss": 0.4588,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998662751727146e-05,
+      "loss": 0.4419,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986451011336958e-05,
+      "loss": 0.4994,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998627334868592e-05,
+      "loss": 0.4793,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986094529332072e-05,
+      "loss": 0.4082,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985914553289208e-05,
+      "loss": 0.4726,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998573342057122e-05,
+      "loss": 0.4569,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985551131192094e-05,
+      "loss": 0.5302,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998536768516589e-05,
+      "loss": 0.4881,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985183082506778e-05,
+      "loss": 0.4039,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984997323228996e-05,
+      "loss": 0.4749,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984810407346895e-05,
+      "loss": 0.5237,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998462233487489e-05,
+      "loss": 0.5074,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99844331058275e-05,
+      "loss": 0.4848,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984242720219333e-05,
+      "loss": 0.4758,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984051178065085e-05,
+      "loss": 0.4603,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983858479379534e-05,
+      "loss": 0.494,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998366462417756e-05,
+      "loss": 0.506,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998346961247412e-05,
+      "loss": 0.491,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998327344428427e-05,
+      "loss": 0.4382,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983076119623148e-05,
+      "loss": 0.4404,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982877638505986e-05,
+      "loss": 0.4865,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99826780009481e-05,
+      "loss": 0.4804,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982477206964908e-05,
+      "loss": 0.5256,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982275256571898e-05,
+      "loss": 0.4963,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982072149784664e-05,
+      "loss": 0.5257,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9981867886618878e-05,
+      "loss": 0.4777,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998166246709031e-05,
+      "loss": 0.536,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998145589121481e-05,
+      "loss": 0.5123,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998124815900833e-05,
+      "loss": 0.4915,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9981039270486897e-05,
+      "loss": 0.5138,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980829225666635e-05,
+      "loss": 0.4959,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980618024563757e-05,
+      "loss": 0.4854,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980405667194566e-05,
+      "loss": 0.4828,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9980192153575455e-05,
+      "loss": 0.4974,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979977483722892e-05,
+      "loss": 0.4466,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997976165765346e-05,
+      "loss": 0.4899,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997954467538381e-05,
+      "loss": 0.5078,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9979326536930693e-05,
+      "loss": 0.4821,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997910724231094e-05,
+      "loss": 0.4769,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997888679154148e-05,
+      "loss": 0.4748,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9978665184639338e-05,
+      "loss": 0.5008,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9978442421621602e-05,
+      "loss": 0.4543,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997821850250548e-05,
+      "loss": 0.5073,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997799342730824e-05,
+      "loss": 0.4562,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977767196047272e-05,
+      "loss": 0.4654,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977539808740023e-05,
+      "loss": 0.4702,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977311265404047e-05,
+      "loss": 0.4847,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9977081566056993e-05,
+      "loss": 0.4638,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997685071071658e-05,
+      "loss": 0.4594,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.997661869940063e-05,
+      "loss": 0.4512,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9976385532127045e-05,
+      "loss": 0.4874,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9976151208913834e-05,
+      "loss": 0.4418,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997591572977907e-05,
+      "loss": 0.503,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975679094740938e-05,
+      "loss": 0.4463,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975441303817697e-05,
+      "loss": 0.4576,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9975202357027704e-05,
+      "loss": 0.4603,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.99749622543894e-05,
+      "loss": 0.4771,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997472099592132e-05,
+      "loss": 0.4763,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997447858164208e-05,
+      "loss": 0.4645,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9974235011570393e-05,
+      "loss": 0.4847,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973990285725058e-05,
+      "loss": 0.531,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973744404124967e-05,
+      "loss": 0.4472,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973497366789097e-05,
+      "loss": 0.5211,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973249173736514e-05,
+      "loss": 0.4718,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972999824986373e-05,
+      "loss": 0.4658,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972749320557924e-05,
+      "loss": 0.4617,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972497660470498e-05,
+      "loss": 0.543,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9972244844743523e-05,
+      "loss": 0.4555,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9971990873396512e-05,
+      "loss": 0.4513,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997173574644906e-05,
+      "loss": 0.5003,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997147946392087e-05,
+      "loss": 0.4535,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9971222025831715e-05,
+      "loss": 0.4955,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997096343220147e-05,
+      "loss": 0.4787,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997070368305009e-05,
+      "loss": 0.4912,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9970442778397625e-05,
+      "loss": 0.4306,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9970180718264215e-05,
+      "loss": 0.4887,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969917502670082e-05,
+      "loss": 0.4251,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969653131635554e-05,
+      "loss": 0.4605,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969387605181018e-05,
+      "loss": 0.4628,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9969120923326982e-05,
+      "loss": 0.4689,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968853086094026e-05,
+      "loss": 0.448,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996858409350282e-05,
+      "loss": 0.4834,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968313945574136e-05,
+      "loss": 0.4664,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968042642328808e-05,
+      "loss": 0.477,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967770183787792e-05,
+      "loss": 0.5361,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9967496569972105e-05,
+      "loss": 0.4402,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996722180090288e-05,
+      "loss": 0.5132,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996694587660131e-05,
+      "loss": 0.6299,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9966668797088705e-05,
+      "loss": 0.44,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996639056238644e-05,
+      "loss": 0.4878,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9966111172515998e-05,
+      "loss": 0.4894,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996583062749894e-05,
+      "loss": 0.4891,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996554892735692e-05,
+      "loss": 0.4945,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9965266072111682e-05,
+      "loss": 0.4715,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9964982061785052e-05,
+      "loss": 0.4838,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996469689639896e-05,
+      "loss": 0.4979,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9964410575975416e-05,
+      "loss": 0.4973,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996412310053651e-05,
+      "loss": 0.4806,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996383447010444e-05,
+      "loss": 0.48,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963544684701477e-05,
+      "loss": 0.4411,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9963253744349992e-05,
+      "loss": 0.4383,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996296164907244e-05,
+      "loss": 0.471,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9962668398891365e-05,
+      "loss": 0.5218,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9962373993829408e-05,
+      "loss": 0.458,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996207843390928e-05,
+      "loss": 0.4919,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9961781719153805e-05,
+      "loss": 0.4925,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9961483849585875e-05,
+      "loss": 0.4549,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996118482522849e-05,
+      "loss": 0.4975,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9960884646104724e-05,
+      "loss": 0.5191,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9960583312237746e-05,
+      "loss": 0.4582,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996028082365082e-05,
+      "loss": 0.4803,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959977180367285e-05,
+      "loss": 0.479,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959672382410582e-05,
+      "loss": 0.4848,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959366429804237e-05,
+      "loss": 0.4022,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9959059322571868e-05,
+      "loss": 0.4899,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995875106073717e-05,
+      "loss": 0.4391,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9958441644323943e-05,
+      "loss": 0.4914,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9958131073356067e-05,
+      "loss": 0.4625,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995781934785751e-05,
+      "loss": 0.445,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9957506467852335e-05,
+      "loss": 0.4899,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9957192433364695e-05,
+      "loss": 0.4538,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956877244418826e-05,
+      "loss": 0.4448,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956560901039047e-05,
+      "loss": 0.4917,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9956243403249788e-05,
+      "loss": 0.4804,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995592475107555e-05,
+      "loss": 0.5026,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9955604944540926e-05,
+      "loss": 0.504,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9955283983670598e-05,
+      "loss": 0.489,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9954961868489346e-05,
+      "loss": 0.4614,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995463859902203e-05,
+      "loss": 0.4689,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9954314175293594e-05,
+      "loss": 0.4303,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953988597329087e-05,
+      "loss": 0.4646,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953661865153636e-05,
+      "loss": 0.5109,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953333978792453e-05,
+      "loss": 0.5291,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9953004938270858e-05,
+      "loss": 0.4467,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995267474361424e-05,
+      "loss": 0.4942,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9952343394848087e-05,
+      "loss": 0.4461,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995201089199797e-05,
+      "loss": 0.4278,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.995167723508956e-05,
+      "loss": 0.4382,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.99513424241486e-05,
+      "loss": 0.4391,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9951006459200943e-05,
+      "loss": 0.4411,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9950669340272515e-05,
+      "loss": 0.5228,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9950331067389337e-05,
+      "loss": 0.4388,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994999164057751e-05,
+      "loss": 0.4395,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994965105986325e-05,
+      "loss": 0.4864,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9949309325272828e-05,
+      "loss": 0.4665,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9948966436832632e-05,
+      "loss": 0.4399,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994862239456912e-05,
+      "loss": 0.5047,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9948277198508845e-05,
+      "loss": 0.4721,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9947930848678465e-05,
+      "loss": 0.4907,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9947583345104694e-05,
+      "loss": 0.4733,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9947234687814365e-05,
+      "loss": 0.4344,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9946884876834388e-05,
+      "loss": 0.4694,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994653391219176e-05,
+      "loss": 0.453,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994618179391357e-05,
+      "loss": 0.5082,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9945828522027e-05,
+      "loss": 0.4292,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994547409655931e-05,
+      "loss": 0.5112,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9945118517537857e-05,
+      "loss": 0.462,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9944761784990096e-05,
+      "loss": 0.4405,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994440389894355e-05,
+      "loss": 0.4536,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9944044859425846e-05,
+      "loss": 0.4748,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.99436846664647e-05,
+      "loss": 0.4557,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994332332008791e-05,
+      "loss": 0.5135,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994296082032336e-05,
+      "loss": 0.532,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9942597167199037e-05,
+      "loss": 0.4664,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994223236074301e-05,
+      "loss": 0.4495,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994186640098343e-05,
+      "loss": 0.4821,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9941499287948554e-05,
+      "loss": 0.4884,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9941131021666705e-05,
+      "loss": 0.4994,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.994076160216631e-05,
+      "loss": 0.4083,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9940391029475887e-05,
+      "loss": 0.4352,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9940019303624035e-05,
+      "loss": 0.5186,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993964642463945e-05,
+      "loss": 0.537,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9939272392550907e-05,
+      "loss": 0.4243,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993889720738728e-05,
+      "loss": 0.5135,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993852086917752e-05,
+      "loss": 0.4422,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993814337795068e-05,
+      "loss": 0.4465,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9937764733735895e-05,
+      "loss": 0.4694,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9937384936562394e-05,
+      "loss": 0.4832,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9937003986459483e-05,
+      "loss": 0.5009,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993662188345657e-05,
+      "loss": 0.4892,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993623862758315e-05,
+      "loss": 0.4846,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99358542188688e-05,
+      "loss": 0.4424,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9935468657343196e-05,
+      "loss": 0.4751,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993508194303609e-05,
+      "loss": 0.4718,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9934694075977338e-05,
+      "loss": 0.4958,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9934305056196868e-05,
+      "loss": 0.5107,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9933914883724713e-05,
+      "loss": 0.4812,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993352355859098e-05,
+      "loss": 0.4646,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9933131080825887e-05,
+      "loss": 0.4585,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9932737450459718e-05,
+      "loss": 0.5021,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993234266752285e-05,
+      "loss": 0.4477,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9931946732045768e-05,
+      "loss": 0.4988,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9931549644059022e-05,
+      "loss": 0.4245,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993115140359326e-05,
+      "loss": 0.4562,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9930752010679222e-05,
+      "loss": 0.4482,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.993035146534774e-05,
+      "loss": 0.4765,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9929949767629724e-05,
+      "loss": 0.506,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992954691755618e-05,
+      "loss": 0.45,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9929142915158204e-05,
+      "loss": 0.4456,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9928737760466974e-05,
+      "loss": 0.4327,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992833145351376e-05,
+      "loss": 0.448,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9927923994329932e-05,
+      "loss": 0.4152,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9927515382946928e-05,
+      "loss": 0.5206,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99271056193963e-05,
+      "loss": 0.4474,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992669470370966e-05,
+      "loss": 0.4403,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992628263591874e-05,
+      "loss": 0.4659,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9925869416055326e-05,
+      "loss": 0.4702,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992545504415133e-05,
+      "loss": 0.4431,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9925039520238724e-05,
+      "loss": 0.4468,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9924622844349586e-05,
+      "loss": 0.4418,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992420501651607e-05,
+      "loss": 0.4862,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9923786036770437e-05,
+      "loss": 0.4551,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9923365905145008e-05,
+      "loss": 0.4366,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992294462167223e-05,
+      "loss": 0.4977,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9922522186384606e-05,
+      "loss": 0.452,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.992209859931475e-05,
+      "loss": 0.4609,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9921673860495348e-05,
+      "loss": 0.4611,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9921247969959186e-05,
+      "loss": 0.5109,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9920820927739135e-05,
+      "loss": 0.46,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9920392733868164e-05,
+      "loss": 0.4363,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9919963388379314e-05,
+      "loss": 0.4973,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9919532891305728e-05,
+      "loss": 0.46,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.991910124268063e-05,
+      "loss": 0.4722,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9918668442537338e-05,
+      "loss": 0.4446,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9918234490909258e-05,
+      "loss": 0.4793,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.991779938782989e-05,
+      "loss": 0.5084,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99173631333328e-05,
+      "loss": 0.4916,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9916925727451678e-05,
+      "loss": 0.4519,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9916487170220275e-05,
+      "loss": 0.4783,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9916047461672443e-05,
+      "loss": 0.4915,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9915606601842122e-05,
+      "loss": 0.47,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991516459076334e-05,
+      "loss": 0.4651,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9914721428470208e-05,
+      "loss": 0.4474,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9914277114996936e-05,
+      "loss": 0.482,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991383165037782e-05,
+      "loss": 0.4575,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9913385034647232e-05,
+      "loss": 0.4338,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.991293726783966e-05,
+      "loss": 0.4361,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9912488349989646e-05,
+      "loss": 0.4729,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9912038281131856e-05,
+      "loss": 0.4941,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9911587061301016e-05,
+      "loss": 0.4351,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9911134690531957e-05,
+      "loss": 0.4535,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9910681168859602e-05,
+      "loss": 0.46,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9910226496318943e-05,
+      "loss": 0.4805,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9909770672945085e-05,
+      "loss": 0.4641,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990931369877321e-05,
+      "loss": 0.5055,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9908855573838575e-05,
+      "loss": 0.4891,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9908396298176552e-05,
+      "loss": 0.429,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990793587182259e-05,
+      "loss": 0.4427,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9907474294812223e-05,
+      "loss": 0.4688,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9907011567181078e-05,
+      "loss": 0.4691,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990654768896487e-05,
+      "loss": 0.4703,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9906082660199406e-05,
+      "loss": 0.5034,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990561648092058e-05,
+      "loss": 0.4878,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990514915116436e-05,
+      "loss": 0.43,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9904680670966836e-05,
+      "loss": 0.4895,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9904211040364153e-05,
+      "loss": 0.4367,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9903740259392563e-05,
+      "loss": 0.5352,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9903268328088408e-05,
+      "loss": 0.4725,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9902795246488106e-05,
+      "loss": 0.4503,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990232101462818e-05,
+      "loss": 0.5305,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990184563254522e-05,
+      "loss": 0.4536,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9901369100275928e-05,
+      "loss": 0.4701,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9900891417857086e-05,
+      "loss": 0.424,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.990041258532556e-05,
+      "loss": 0.4494,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989993260271831e-05,
+      "loss": 0.4418,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989945147007238e-05,
+      "loss": 0.4665,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9898969187424903e-05,
+      "loss": 0.49,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9898485754813113e-05,
+      "loss": 0.4378,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9898001172274318e-05,
+      "loss": 0.4524,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9897515439845918e-05,
+      "loss": 0.4618,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9897028557565406e-05,
+      "loss": 0.4807,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989654052547036e-05,
+      "loss": 0.4266,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9896051343598457e-05,
+      "loss": 0.4476,
+      "step": 774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9895561011987443e-05,
+      "loss": 0.5141,
+      "step": 775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989506953067517e-05,
+      "loss": 0.4769,
+      "step": 776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9894576899699566e-05,
+      "loss": 0.4302,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9894083119098662e-05,
+      "loss": 0.4707,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.989358818891057e-05,
+      "loss": 0.4156,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9893092109173485e-05,
+      "loss": 0.4732,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.98925948799257e-05,
+      "loss": 0.459,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9892096501205594e-05,
+      "loss": 0.4544,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9891596973051625e-05,
+      "loss": 0.4627,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9891096295502363e-05,
+      "loss": 0.4741,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9890594468596443e-05,
+      "loss": 0.4607,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9890091492372604e-05,
+      "loss": 0.4197,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988958736686966e-05,
+      "loss": 0.4814,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9889082092126528e-05,
+      "loss": 0.4625,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9888575668182203e-05,
+      "loss": 0.4577,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9888068095075777e-05,
+      "loss": 0.413,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9887559372846423e-05,
+      "loss": 0.4848,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988704950153341e-05,
+      "loss": 0.4627,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988653848117609e-05,
+      "loss": 0.4597,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.98860263118139e-05,
+      "loss": 0.4076,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9885512993486384e-05,
+      "loss": 0.4551,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988499852623315e-05,
+      "loss": 0.4406,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9884482910093912e-05,
+      "loss": 0.459,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988396614510847e-05,
+      "loss": 0.4392,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9883448231316702e-05,
+      "loss": 0.479,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9882929168758597e-05,
+      "loss": 0.4384,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9882408957474198e-05,
+      "loss": 0.4561,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9881887597503675e-05,
+      "loss": 0.4461,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9881365088887264e-05,
+      "loss": 0.4996,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988084143166529e-05,
+      "loss": 0.5105,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.988031662587817e-05,
+      "loss": 0.4626,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9879790671566417e-05,
+      "loss": 0.4659,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9879263568770622e-05,
+      "loss": 0.4836,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987873531753147e-05,
+      "loss": 0.4621,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987820591788974e-05,
+      "loss": 0.3976,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987767536988628e-05,
+      "loss": 0.4734,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987714367356205e-05,
+      "loss": 0.4576,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987661082895809e-05,
+      "loss": 0.4871,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9876076836115517e-05,
+      "loss": 0.4484,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9875541695075553e-05,
+      "loss": 0.4676,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9875005405879504e-05,
+      "loss": 0.5239,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987446796856876e-05,
+      "loss": 0.4266,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9873929383184805e-05,
+      "loss": 0.4759,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9873389649769204e-05,
+      "loss": 0.4514,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9872848768363628e-05,
+      "loss": 0.4681,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987230673900981e-05,
+      "loss": 0.4964,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9871763561749597e-05,
+      "loss": 0.4973,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9871219236624903e-05,
+      "loss": 0.4348,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987067376367775e-05,
+      "loss": 0.5284,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.987012714295024e-05,
+      "loss": 0.5224,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9869579374484557e-05,
+      "loss": 0.4729,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9869030458322985e-05,
+      "loss": 0.4744,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9868480394507895e-05,
+      "loss": 0.5362,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986792918308173e-05,
+      "loss": 0.4689,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986737682408705e-05,
+      "loss": 0.4796,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986682331756648e-05,
+      "loss": 0.5453,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9866268663562746e-05,
+      "loss": 0.4404,
+      "step": 831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9865712862118652e-05,
+      "loss": 0.4166,
+      "step": 832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9865155913277104e-05,
+      "loss": 0.4876,
+      "step": 833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9864597817081083e-05,
+      "loss": 0.5147,
+      "step": 834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986403857357367e-05,
+      "loss": 0.5537,
+      "step": 835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9863478182798034e-05,
+      "loss": 0.455,
+      "step": 836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986291664479742e-05,
+      "loss": 0.4654,
+      "step": 837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9862353959615175e-05,
+      "loss": 0.487,
+      "step": 838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9861790127294722e-05,
+      "loss": 0.4494,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.986122514787959e-05,
+      "loss": 0.4473,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9860659021413377e-05,
+      "loss": 0.5178,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9860091747939787e-05,
+      "loss": 0.46,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9859523327502603e-05,
+      "loss": 0.4434,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.985895376014569e-05,
+      "loss": 0.4558,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.985838304591302e-05,
+      "loss": 0.4754,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9857811184848638e-05,
+      "loss": 0.4473,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9857238176996686e-05,
+      "loss": 0.4447,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985666402240139e-05,
+      "loss": 0.4512,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9856088721107057e-05,
+      "loss": 0.4372,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9855512273158102e-05,
+      "loss": 0.4586,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9854934678599017e-05,
+      "loss": 0.4455,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985435593747438e-05,
+      "loss": 0.4776,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9853776049828858e-05,
+      "loss": 0.4778,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985319501570721e-05,
+      "loss": 0.486,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9852612835154288e-05,
+      "loss": 0.5067,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9852029508215022e-05,
+      "loss": 0.4496,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9851445034934437e-05,
+      "loss": 0.4982,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985085941535765e-05,
+      "loss": 0.4597,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.985027264952985e-05,
+      "loss": 0.4858,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9849684737496337e-05,
+      "loss": 0.4786,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9849095679302485e-05,
+      "loss": 0.4611,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984850547499376e-05,
+      "loss": 0.4132,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9847914124615715e-05,
+      "loss": 0.4103,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984732162821399e-05,
+      "loss": 0.4547,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9846727985834326e-05,
+      "loss": 0.3954,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9846133197522533e-05,
+      "loss": 0.4326,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9845537263324524e-05,
+      "loss": 0.4572,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9844940183286297e-05,
+      "loss": 0.4716,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984434195745393e-05,
+      "loss": 0.4295,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9843742585873603e-05,
+      "loss": 0.4223,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9843142068591583e-05,
+      "loss": 0.442,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9842540405654203e-05,
+      "loss": 0.421,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9841937597107922e-05,
+      "loss": 0.4778,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9841333642999253e-05,
+      "loss": 0.4722,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984072854337482e-05,
+      "loss": 0.4081,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.984012229828132e-05,
+      "loss": 0.4939,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983951490776555e-05,
+      "loss": 0.4201,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9838906371874392e-05,
+      "loss": 0.4474,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983829669065481e-05,
+      "loss": 0.4472,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983768586415387e-05,
+      "loss": 0.4808,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983707389241871e-05,
+      "loss": 0.4744,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983646077549657e-05,
+      "loss": 0.4784,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983584651343477e-05,
+      "loss": 0.4364,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9835231106280717e-05,
+      "loss": 0.4031,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9834614554081922e-05,
+      "loss": 0.4724,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9833996856885968e-05,
+      "loss": 0.4376,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983337801474053e-05,
+      "loss": 0.4139,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9832758027693368e-05,
+      "loss": 0.4348,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.983213689579234e-05,
+      "loss": 0.4802,
+      "step": 889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9831514619085392e-05,
+      "loss": 0.4274,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9830891197620547e-05,
+      "loss": 0.5295,
+      "step": 891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9830266631445927e-05,
+      "loss": 0.393,
+      "step": 892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.982964092060974e-05,
+      "loss": 0.458,
+      "step": 893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9829014065160276e-05,
+      "loss": 0.4905,
+      "step": 894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9828386065145923e-05,
+      "loss": 0.421,
+      "step": 895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9827756920615146e-05,
+      "loss": 0.4693,
+      "step": 896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9827126631616513e-05,
+      "loss": 0.4392,
+      "step": 897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9826495198198665e-05,
+      "loss": 0.4801,
+      "step": 898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.982586262041035e-05,
+      "loss": 0.4103,
+      "step": 899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9825228898300382e-05,
+      "loss": 0.4536,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.982459403191767e-05,
+      "loss": 0.4408,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9823958021311237e-05,
+      "loss": 0.462,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9823320866530146e-05,
+      "loss": 0.4873,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9822682567623597e-05,
+      "loss": 0.4808,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9822043124640845e-05,
+      "loss": 0.5074,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9821402537631247e-05,
+      "loss": 0.4446,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9820760806644253e-05,
+      "loss": 0.4748,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9820117931729384e-05,
+      "loss": 0.4751,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9819473912936262e-05,
+      "loss": 0.4416,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9818828750314602e-05,
+      "loss": 0.4407,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9818182443914192e-05,
+      "loss": 0.3973,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9817534993784922e-05,
+      "loss": 0.4756,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981688639997676e-05,
+      "loss": 0.4677,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9816236662539775e-05,
+      "loss": 0.4503,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9815585781524112e-05,
+      "loss": 0.4885,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9814933756980004e-05,
+      "loss": 0.4401,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981428058895779e-05,
+      "loss": 0.4539,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9813626277507868e-05,
+      "loss": 0.4558,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9812970822680752e-05,
+      "loss": 0.4429,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9812314224527026e-05,
+      "loss": 0.4733,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.981165648309737e-05,
+      "loss": 0.4182,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9810997598442558e-05,
+      "loss": 0.4789,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9810337570613438e-05,
+      "loss": 0.4849,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9809676399660957e-05,
+      "loss": 0.4478,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9809014085636146e-05,
+      "loss": 0.4382,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9808350628590124e-05,
+      "loss": 0.4524,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9807686028574105e-05,
+      "loss": 0.4148,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9807020285639377e-05,
+      "loss": 0.5078,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.980635339983733e-05,
+      "loss": 0.4044,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9805685371219437e-05,
+      "loss": 0.4537,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9805016199837258e-05,
+      "loss": 0.4501,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9804345885742443e-05,
+      "loss": 0.4794,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9803674428986727e-05,
+      "loss": 0.4375,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9803001829621943e-05,
+      "loss": 0.4961,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9802328087699997e-05,
+      "loss": 0.4143,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9801653203272892e-05,
+      "loss": 0.5241,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9800977176392726e-05,
+      "loss": 0.4236,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9800300007111668e-05,
+      "loss": 0.4336,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979962169548199e-05,
+      "loss": 0.4329,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9798942241556048e-05,
+      "loss": 0.4566,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979826164538628e-05,
+      "loss": 0.4816,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9797579907025225e-05,
+      "loss": 0.469,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9796897026525492e-05,
+      "loss": 0.4927,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9796213003939798e-05,
+      "loss": 0.4093,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9795527839320936e-05,
+      "loss": 0.4973,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9794841532721782e-05,
+      "loss": 0.421,
+      "step": 946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.979415408419532e-05,
+      "loss": 0.4574,
+      "step": 947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9793465493794607e-05,
+      "loss": 0.4693,
+      "step": 948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9792775761572783e-05,
+      "loss": 0.4478,
+      "step": 949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9792084887583096e-05,
+      "loss": 0.452,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9791392871878862e-05,
+      "loss": 0.4393,
+      "step": 951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.97906997145135e-05,
+      "loss": 0.4872,
+      "step": 952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9790005415540508e-05,
+      "loss": 0.5156,
+      "step": 953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9789309975013467e-05,
+      "loss": 0.4253,
+      "step": 954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.978861339298607e-05,
+      "loss": 0.4612,
+      "step": 955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9787915669512068e-05,
+      "loss": 0.4846,
+      "step": 956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.978721680464532e-05,
+      "loss": 0.517,
+      "step": 957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.978651679843977e-05,
+      "loss": 0.4447,
+      "step": 958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9785815650949444e-05,
+      "loss": 0.4808,
+      "step": 959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9785113362228455e-05,
+      "loss": 0.4202,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.978440993233102e-05,
+      "loss": 0.4894,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9783705361311425e-05,
+      "loss": 0.4762,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.978299964922405e-05,
+      "loss": 0.5123,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9782292796123366e-05,
+      "loss": 0.4445,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9781584802063937e-05,
+      "loss": 0.4166,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.97808756671004e-05,
+      "loss": 0.4951,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9780165391287498e-05,
+      "loss": 0.4865,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9779453974680043e-05,
+      "loss": 0.4836,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9778741417332952e-05,
+      "loss": 0.4025,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9778027719301225e-05,
+      "loss": 0.4501,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.977731288063994e-05,
+      "loss": 0.4498,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9776596901404278e-05,
+      "loss": 0.4504,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.97758797816495e-05,
+      "loss": 0.4056,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9775161521430957e-05,
+      "loss": 0.412,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9774442120804082e-05,
+      "loss": 0.4765,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9773721579824407e-05,
+      "loss": 0.4232,
+      "step": 976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9772999898547545e-05,
+      "loss": 0.4278,
+      "step": 977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.97722770770292e-05,
+      "loss": 0.464,
+      "step": 978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9771553115325155e-05,
+      "loss": 0.4583,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9770828013491296e-05,
+      "loss": 0.431,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9770101771583587e-05,
+      "loss": 0.4585,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9769374389658082e-05,
+      "loss": 0.4791,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9768645867770924e-05,
+      "loss": 0.4079,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9767916205978345e-05,
+      "loss": 0.4272,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9767185404336657e-05,
+      "loss": 0.4485,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9766453462902274e-05,
+      "loss": 0.4613,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9765720381731684e-05,
+      "loss": 0.4462,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9764986160881476e-05,
+      "loss": 0.4522,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9764250800408316e-05,
+      "loss": 0.4669,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9763514300368958e-05,
+      "loss": 0.4743,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9762776660820255e-05,
+      "loss": 0.4774,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976203788181914e-05,
+      "loss": 0.4394,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.976129796342263e-05,
+      "loss": 0.4465,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9760556905687845e-05,
+      "loss": 0.4586,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9759814708671973e-05,
+      "loss": 0.4717,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9759071372432305e-05,
+      "loss": 0.4346,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9758326897026208e-05,
+      "loss": 0.4253,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9757581282511156e-05,
+      "loss": 0.4849,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975683452894469e-05,
+      "loss": 0.4205,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975608663638445e-05,
+      "loss": 0.4477,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9755337604888156e-05,
+      "loss": 0.4489,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975458743451363e-05,
+      "loss": 0.4524,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975383612531877e-05,
+      "loss": 0.4929,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.975308367736156e-05,
+      "loss": 0.4665,
+      "step": 1004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9752330090700087e-05,
+      "loss": 0.4272,
+      "step": 1005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9751575365392503e-05,
+      "loss": 0.4835,
+      "step": 1006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9750819501497073e-05,
+      "loss": 0.4945,
+      "step": 1007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9750062499072132e-05,
+      "loss": 0.4624,
+      "step": 1008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974930435817611e-05,
+      "loss": 0.4286,
+      "step": 1009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9748545078867524e-05,
+      "loss": 0.4414,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.974778466120498e-05,
+      "loss": 0.4985,
+      "step": 1011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9747023105247165e-05,
+      "loss": 0.4395,
+      "step": 1012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9746260411052855e-05,
+      "loss": 0.4889,
+      "step": 1013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9745496578680933e-05,
+      "loss": 0.4729,
+      "step": 1014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9744731608190346e-05,
+      "loss": 0.4473,
+      "step": 1015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9743965499640134e-05,
+      "loss": 0.4422,
+      "step": 1016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9743198253089434e-05,
+      "loss": 0.4425,
+      "step": 1017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9742429868597465e-05,
+      "loss": 0.4326,
+      "step": 1018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9741660346223532e-05,
+      "loss": 0.4494,
+      "step": 1019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9740889686027026e-05,
+      "loss": 0.3943,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.974011788806744e-05,
+      "loss": 0.5128,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9739344952404336e-05,
+      "loss": 0.4535,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9738570879097378e-05,
+      "loss": 0.464,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973779566820631e-05,
+      "loss": 0.4688,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9737019319790962e-05,
+      "loss": 0.4335,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973624183391126e-05,
+      "loss": 0.4528,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9735463210627213e-05,
+      "loss": 0.4555,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.973468344999892e-05,
+      "loss": 0.4658,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9733902552086563e-05,
+      "loss": 0.4302,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9733120516950414e-05,
+      "loss": 0.4977,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9732337344650843e-05,
+      "loss": 0.438,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9731553035248287e-05,
+      "loss": 0.446,
+      "step": 1032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9730767588803288e-05,
+      "loss": 0.4993,
+      "step": 1033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972998100537647e-05,
+      "loss": 0.4444,
+      "step": 1034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972919328502854e-05,
+      "loss": 0.4774,
+      "step": 1035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9728404427820305e-05,
+      "loss": 0.4128,
+      "step": 1036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9727614433812645e-05,
+      "loss": 0.408,
+      "step": 1037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972682330306654e-05,
+      "loss": 0.4158,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9726031035643053e-05,
+      "loss": 0.4202,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972523763160333e-05,
+      "loss": 0.445,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9724443091008612e-05,
+      "loss": 0.4756,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9723647413920224e-05,
+      "loss": 0.3993,
+      "step": 1042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972285060039959e-05,
+      "loss": 0.4679,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972205265050819e-05,
+      "loss": 0.4686,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9721253564307628e-05,
+      "loss": 0.4881,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.972045334185958e-05,
+      "loss": 0.4548,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9719651983225805e-05,
+      "loss": 0.4056,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9718849488468164e-05,
+      "loss": 0.4462,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9718045857648586e-05,
+      "loss": 0.3932,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9717241090829104e-05,
+      "loss": 0.4914,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.971643518807184e-05,
+      "loss": 0.4769,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9715628149438984e-05,
+      "loss": 0.4411,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9714819974992836e-05,
+      "loss": 0.4365,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9714010664795768e-05,
+      "loss": 0.4432,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.971320021891025e-05,
+      "loss": 0.4725,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9712388637398836e-05,
+      "loss": 0.5168,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9711575920324164e-05,
+      "loss": 0.4402,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9710762067748964e-05,
+      "loss": 0.4009,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9709947079736057e-05,
+      "loss": 0.4606,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9709130956348338e-05,
+      "loss": 0.4769,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9708313697648804e-05,
+      "loss": 0.4583,
+      "step": 1061
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9707495303700535e-05,
+      "loss": 0.381,
+      "step": 1062
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9706675774566703e-05,
+      "loss": 0.4124,
+      "step": 1063
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9705855110310554e-05,
+      "loss": 0.4242,
+      "step": 1064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970503331099543e-05,
+      "loss": 0.4618,
+      "step": 1065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9704210376684767e-05,
+      "loss": 0.4101,
+      "step": 1066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9703386307442083e-05,
+      "loss": 0.4392,
+      "step": 1067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.970256110333097e-05,
+      "loss": 0.3896,
+      "step": 1068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9701734764415143e-05,
+      "loss": 0.4796,
+      "step": 1069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9700907290758362e-05,
+      "loss": 0.4641,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.97000786824245e-05,
+      "loss": 0.3842,
+      "step": 1071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9699248939477527e-05,
+      "loss": 0.4142,
+      "step": 1072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9698418061981466e-05,
+      "loss": 0.4641,
+      "step": 1073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9697586050000457e-05,
+      "loss": 0.4424,
+      "step": 1074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9696752903598714e-05,
+      "loss": 0.4582,
+      "step": 1075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.969591862284055e-05,
+      "loss": 0.4935,
+      "step": 1076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.969508320779035e-05,
+      "loss": 0.441,
+      "step": 1077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9694246658512602e-05,
+      "loss": 0.4547,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969340897507187e-05,
+      "loss": 0.459,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969257015753281e-05,
+      "loss": 0.4876,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9691730205960165e-05,
+      "loss": 0.4874,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969088912041877e-05,
+      "loss": 0.4715,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.969004690097354e-05,
+      "loss": 0.3791,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9689203547689483e-05,
+      "loss": 0.4318,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968835906063169e-05,
+      "loss": 0.4473,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968751343986534e-05,
+      "loss": 0.4239,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968666668545571e-05,
+      "loss": 0.4809,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9685818797468146e-05,
+      "loss": 0.4086,
+      "step": 1088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.96849697759681e-05,
+      "loss": 0.4577,
+      "step": 1089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.96841196210211e-05,
+      "loss": 0.4458,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968326833269276e-05,
+      "loss": 0.4483,
+      "step": 1091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968241591104879e-05,
+      "loss": 0.4683,
+      "step": 1092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9681562356154986e-05,
+      "loss": 0.4607,
+      "step": 1093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.968070766807722e-05,
+      "loss": 0.4895,
+      "step": 1094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9679851846881474e-05,
+      "loss": 0.4997,
+      "step": 1095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967899489263379e-05,
+      "loss": 0.4341,
+      "step": 1096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9678136805400324e-05,
+      "loss": 0.4155,
+      "step": 1097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9677277585247296e-05,
+      "loss": 0.4531,
+      "step": 1098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967641723224103e-05,
+      "loss": 0.3975,
+      "step": 1099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967555574644793e-05,
+      "loss": 0.4719,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9674693127934487e-05,
+      "loss": 0.4577,
+      "step": 1101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967382937676729e-05,
+      "loss": 0.4722,
+      "step": 1102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9672964493012993e-05,
+      "loss": 0.4209,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967209847673836e-05,
+      "loss": 0.4589,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.967123132801024e-05,
+      "loss": 0.4627,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9670363046895547e-05,
+      "loss": 0.4795,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966949363346131e-05,
+      "loss": 0.4778,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9668623087774635e-05,
+      "loss": 0.4771,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9667751409902703e-05,
+      "loss": 0.4647,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9666878599912808e-05,
+      "loss": 0.4214,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9666004657872304e-05,
+      "loss": 0.437,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9665129583848658e-05,
+      "loss": 0.4701,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9664253377909406e-05,
+      "loss": 0.4261,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966337604012217e-05,
+      "loss": 0.4761,
+      "step": 1114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9662497570554683e-05,
+      "loss": 0.4152,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9661617969274734e-05,
+      "loss": 0.4757,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.966073723635022e-05,
+      "loss": 0.4771,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9659855371849118e-05,
+      "loss": 0.4222,
+      "step": 1118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9658972375839498e-05,
+      "loss": 0.444,
+      "step": 1119
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965808824838951e-05,
+      "loss": 0.447,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9657202989567393e-05,
+      "loss": 0.4382,
+      "step": 1121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965631659944148e-05,
+      "loss": 0.4003,
+      "step": 1122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.965542907808018e-05,
+      "loss": 0.4544,
+      "step": 1123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9654540425552e-05,
+      "loss": 0.4839,
+      "step": 1124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9653650641925532e-05,
+      "loss": 0.4426,
+      "step": 1125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9652759727269447e-05,
+      "loss": 0.445,
+      "step": 1126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9651867681652515e-05,
+      "loss": 0.4808,
+      "step": 1127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9650974505143585e-05,
+      "loss": 0.4795,
+      "step": 1128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9650080197811597e-05,
+      "loss": 0.4613,
+      "step": 1129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9649184759725575e-05,
+      "loss": 0.4559,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9648288190954632e-05,
+      "loss": 0.4305,
+      "step": 1131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9647390491567978e-05,
+      "loss": 0.4633,
+      "step": 1132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964649166163489e-05,
+      "loss": 0.4231,
+      "step": 1133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.964559170122475e-05,
+      "loss": 0.4832,
+      "step": 1134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9644690610407016e-05,
+      "loss": 0.4264,
+      "step": 1135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9643788389251243e-05,
+      "loss": 0.4568,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9642885037827064e-05,
+      "loss": 0.4601,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9641980556204206e-05,
+      "loss": 0.4345,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.964107494445248e-05,
+      "loss": 0.4627,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.964016820264178e-05,
+      "loss": 0.4581,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9639260330842104e-05,
+      "loss": 0.4506,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9638351329123515e-05,
+      "loss": 0.4038,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9637441197556175e-05,
+      "loss": 0.3972,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9636529936210337e-05,
+      "loss": 0.4047,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9635617545156332e-05,
+      "loss": 0.4402,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9634704024464582e-05,
+      "loss": 0.4824,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9633789374205594e-05,
+      "loss": 0.4538,
+      "step": 1147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9632873594449967e-05,
+      "loss": 0.4457,
+      "step": 1148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9631956685268387e-05,
+      "loss": 0.4491,
+      "step": 1149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9631038646731623e-05,
+      "loss": 0.4182,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9630119478910532e-05,
+      "loss": 0.4283,
+      "step": 1151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962919918187606e-05,
+      "loss": 0.4299,
+      "step": 1152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9628277755699244e-05,
+      "loss": 0.4944,
+      "step": 1153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9627355200451195e-05,
+      "loss": 0.4612,
+      "step": 1154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9626431516203118e-05,
+      "loss": 0.4461,
+      "step": 1155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962550670302632e-05,
+      "loss": 0.5041,
+      "step": 1156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9624580760992173e-05,
+      "loss": 0.4648,
+      "step": 1157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9623653690172143e-05,
+      "loss": 0.4231,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.962272549063779e-05,
+      "loss": 0.443,
+      "step": 1159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9621796162460754e-05,
+      "loss": 0.4581,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9620865705712765e-05,
+      "loss": 0.4233,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9619934120465643e-05,
+      "loss": 0.4477,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9619001406791285e-05,
+      "loss": 0.3911,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9618067564761684e-05,
+      "loss": 0.4501,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9617132594448924e-05,
+      "loss": 0.4262,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961619649592516e-05,
+      "loss": 0.387,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961525926926265e-05,
+      "loss": 0.4895,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961432091453373e-05,
+      "loss": 0.4506,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9613381431810832e-05,
+      "loss": 0.475,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.961244082116646e-05,
+      "loss": 0.4149,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9611499082673224e-05,
+      "loss": 0.3646,
+      "step": 1171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9610556216403806e-05,
+      "loss": 0.5021,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9609612222430975e-05,
+      "loss": 0.4239,
+      "step": 1173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9608667100827603e-05,
+      "loss": 0.4313,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9607720851666637e-05,
+      "loss": 0.4806,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9606773475021103e-05,
+      "loss": 0.4675,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9605824970964133e-05,
+      "loss": 0.4281,
+      "step": 1177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960487533956893e-05,
+      "loss": 0.4531,
+      "step": 1178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9603924580908794e-05,
+      "loss": 0.4787,
+      "step": 1179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960297269505711e-05,
+      "loss": 0.4403,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9602019682087346e-05,
+      "loss": 0.472,
+      "step": 1181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.960106554207306e-05,
+      "loss": 0.4616,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9600110275087894e-05,
+      "loss": 0.4279,
+      "step": 1183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9599153881205584e-05,
+      "loss": 0.4402,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.959819636049995e-05,
+      "loss": 0.4441,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9597237713044888e-05,
+      "loss": 0.3974,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.95962779389144e-05,
+      "loss": 0.4135,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.959531703818256e-05,
+      "loss": 0.4406,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9594355010923536e-05,
+      "loss": 0.3897,
+      "step": 1189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9593391857211582e-05,
+      "loss": 0.4202,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9592427577121035e-05,
+      "loss": 0.4486,
+      "step": 1191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.959146217072633e-05,
+      "loss": 0.387,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.959049563810197e-05,
+      "loss": 0.4798,
+      "step": 1193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9589527979322562e-05,
+      "loss": 0.4821,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9588559194462794e-05,
+      "loss": 0.4486,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9587589283597443e-05,
+      "loss": 0.5039,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9586618246801367e-05,
+      "loss": 0.4193,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9585646084149514e-05,
+      "loss": 0.4594,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9584672795716925e-05,
+      "loss": 0.4555,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9583698381578714e-05,
+      "loss": 0.4625,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95827228418101e-05,
+      "loss": 0.4312,
+      "step": 1201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.958174617648637e-05,
+      "loss": 0.4891,
+      "step": 1202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9580768385682916e-05,
+      "loss": 0.438,
+      "step": 1203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9579789469475198e-05,
+      "loss": 0.4266,
+      "step": 1204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957880942793878e-05,
+      "loss": 0.4374,
+      "step": 1205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9577828261149304e-05,
+      "loss": 0.4289,
+      "step": 1206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95768459691825e-05,
+      "loss": 0.4207,
+      "step": 1207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9575862552114187e-05,
+      "loss": 0.4247,
+      "step": 1208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9574878010020268e-05,
+      "loss": 0.4604,
+      "step": 1209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9573892342976732e-05,
+      "loss": 0.4266,
+      "step": 1210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957290555105966e-05,
+      "loss": 0.4228,
+      "step": 1211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9571917634345218e-05,
+      "loss": 0.4784,
+      "step": 1212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.957092859290965e-05,
+      "loss": 0.4644,
+      "step": 1213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9569938426829304e-05,
+      "loss": 0.4331,
+      "step": 1214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95689471361806e-05,
+      "loss": 0.426,
+      "step": 1215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956795472104005e-05,
+      "loss": 0.4295,
+      "step": 1216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956696118148425e-05,
+      "loss": 0.3976,
+      "step": 1217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9565966517589892e-05,
+      "loss": 0.4325,
+      "step": 1218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9564970729433746e-05,
+      "loss": 0.4362,
+      "step": 1219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9563973817092667e-05,
+      "loss": 0.4259,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9562975780643602e-05,
+      "loss": 0.4396,
+      "step": 1221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.956197662016359e-05,
+      "loss": 0.4452,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9560976335729747e-05,
+      "loss": 0.4288,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.955997492741927e-05,
+      "loss": 0.4582,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.955897239530947e-05,
+      "loss": 0.4131,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9557968739477713e-05,
+      "loss": 0.4557,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9556963960001464e-05,
+      "loss": 0.4547,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9555958056958285e-05,
+      "loss": 0.4354,
+      "step": 1228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9554951030425806e-05,
+      "loss": 0.466,
+      "step": 1229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9553942880481767e-05,
+      "loss": 0.4298,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.955293360720397e-05,
+      "loss": 0.4266,
+      "step": 1231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9551923210670315e-05,
+      "loss": 0.4654,
+      "step": 1232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9550911690958797e-05,
+      "loss": 0.4326,
+      "step": 1233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.954989904814748e-05,
+      "loss": 0.4141,
+      "step": 1234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9548885282314533e-05,
+      "loss": 0.4087,
+      "step": 1235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9547870393538196e-05,
+      "loss": 0.428,
+      "step": 1236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9546854381896802e-05,
+      "loss": 0.3984,
+      "step": 1237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9545837247468772e-05,
+      "loss": 0.4555,
+      "step": 1238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9544818990332617e-05,
+      "loss": 0.4011,
+      "step": 1239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9543799610566923e-05,
+      "loss": 0.4274,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9542779108250374e-05,
+      "loss": 0.4768,
+      "step": 1241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.954175748346174e-05,
+      "loss": 0.4836,
+      "step": 1242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.954073473627987e-05,
+      "loss": 0.5079,
+      "step": 1243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95397108667837e-05,
+      "loss": 0.4987,
+      "step": 1244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9538685875052266e-05,
+      "loss": 0.4503,
+      "step": 1245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953765976116468e-05,
+      "loss": 0.4579,
+      "step": 1246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9536632525200128e-05,
+      "loss": 0.3815,
+      "step": 1247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9535604167237914e-05,
+      "loss": 0.4399,
+      "step": 1248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.95345746873574e-05,
+      "loss": 0.4282,
+      "step": 1249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9533544085638048e-05,
+      "loss": 0.4055,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953251236215941e-05,
+      "loss": 0.5253,
+      "step": 1251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.953147951700111e-05,
+      "loss": 0.434,
+      "step": 1252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9530445550242873e-05,
+      "loss": 0.4213,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9529410461964498e-05,
+      "loss": 0.4672,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9528374252245885e-05,
+      "loss": 0.449,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9527336921167012e-05,
+      "loss": 0.4559,
+      "step": 1256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9526298468807942e-05,
+      "loss": 0.4281,
+      "step": 1257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9525258895248826e-05,
+      "loss": 0.4705,
+      "step": 1258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9524218200569904e-05,
+      "loss": 0.4341,
+      "step": 1259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9523176384851503e-05,
+      "loss": 0.4189,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.952213344817403e-05,
+      "loss": 0.4331,
+      "step": 1261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9521089390617993e-05,
+      "loss": 0.4553,
+      "step": 1262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9520044212263962e-05,
+      "loss": 0.3893,
+      "step": 1263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9518997913192622e-05,
+      "loss": 0.4427,
+      "step": 1264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9517950493484726e-05,
+      "loss": 0.4186,
+      "step": 1265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9516901953221112e-05,
+      "loss": 0.4686,
+      "step": 1266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9515852292482718e-05,
+      "loss": 0.4264,
+      "step": 1267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9514801511350558e-05,
+      "loss": 0.4097,
+      "step": 1268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.951374960990574e-05,
+      "loss": 0.4556,
+      "step": 1269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9512696588229445e-05,
+      "loss": 0.4461,
+      "step": 1270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9511642446402964e-05,
+      "loss": 0.4373,
+      "step": 1271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9510587184507645e-05,
+      "loss": 0.4311,
+      "step": 1272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9509530802624946e-05,
+      "loss": 0.4499,
+      "step": 1273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.95084733008364e-05,
+      "loss": 0.5196,
+      "step": 1274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9507414679223628e-05,
+      "loss": 0.456,
+      "step": 1275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9506354937868344e-05,
+      "loss": 0.4392,
+      "step": 1276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950529407685234e-05,
+      "loss": 0.4102,
+      "step": 1277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9504232096257497e-05,
+      "loss": 0.3848,
+      "step": 1278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9503168996165788e-05,
+      "loss": 0.3915,
+      "step": 1279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.950210477665926e-05,
+      "loss": 0.4638,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9501039437820056e-05,
+      "loss": 0.4321,
+      "step": 1281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949997297973041e-05,
+      "loss": 0.4389,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9498905402472625e-05,
+      "loss": 0.4599,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9497836706129104e-05,
+      "loss": 0.4571,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949676689078234e-05,
+      "loss": 0.506,
+      "step": 1285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9495695956514903e-05,
+      "loss": 0.4135,
+      "step": 1286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9494623903409448e-05,
+      "loss": 0.4202,
+      "step": 1287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949355073154872e-05,
+      "loss": 0.4943,
+      "step": 1288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9492476441015565e-05,
+      "loss": 0.443,
+      "step": 1289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.949140103189288e-05,
+      "loss": 0.4782,
+      "step": 1290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9490324504263684e-05,
+      "loss": 0.4035,
+      "step": 1291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9489246858211063e-05,
+      "loss": 0.4667,
+      "step": 1292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9488168093818195e-05,
+      "loss": 0.3938,
+      "step": 1293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9487088211168344e-05,
+      "loss": 0.4539,
+      "step": 1294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948600721034486e-05,
+      "loss": 0.4766,
+      "step": 1295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9484925091431177e-05,
+      "loss": 0.4617,
+      "step": 1296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948384185451082e-05,
+      "loss": 0.4291,
+      "step": 1297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9482757499667393e-05,
+      "loss": 0.4604,
+      "step": 1298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.94816720269846e-05,
+      "loss": 0.4132,
+      "step": 1299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.948058543654621e-05,
+      "loss": 0.4874,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9479497728436103e-05,
+      "loss": 0.4236,
+      "step": 1301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947840890273823e-05,
+      "loss": 0.4328,
+      "step": 1302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947731895953662e-05,
+      "loss": 0.427,
+      "step": 1303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947622789891541e-05,
+      "loss": 0.4216,
+      "step": 1304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947513572095881e-05,
+      "loss": 0.4487,
+      "step": 1305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9474042425751124e-05,
+      "loss": 0.4469,
+      "step": 1306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.947294801337673e-05,
+      "loss": 0.4727,
+      "step": 1307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.94718524839201e-05,
+      "loss": 0.4759,
+      "step": 1308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9470755837465794e-05,
+      "loss": 0.3502,
+      "step": 1309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9469658074098454e-05,
+      "loss": 0.4246,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9468559193902808e-05,
+      "loss": 0.4065,
+      "step": 1311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.946745919696368e-05,
+      "loss": 0.4206,
+      "step": 1312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9466358083365967e-05,
+      "loss": 0.3883,
+      "step": 1313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9465255853194652e-05,
+      "loss": 0.4468,
+      "step": 1314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.946415250653482e-05,
+      "loss": 0.4464,
+      "step": 1315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9463048043471625e-05,
+      "loss": 0.4444,
+      "step": 1316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9461942464090316e-05,
+      "loss": 0.4448,
+      "step": 1317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9460835768476225e-05,
+      "loss": 0.4347,
+      "step": 1318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945972795671478e-05,
+      "loss": 0.4513,
+      "step": 1319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9458619028891472e-05,
+      "loss": 0.5124,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9457508985091903e-05,
+      "loss": 0.4376,
+      "step": 1321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945639782540175e-05,
+      "loss": 0.4752,
+      "step": 1322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945528554990677e-05,
+      "loss": 0.4334,
+      "step": 1323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9454172158692822e-05,
+      "loss": 0.3983,
+      "step": 1324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9453057651845836e-05,
+      "loss": 0.4002,
+      "step": 1325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9451942029451834e-05,
+      "loss": 0.4369,
+      "step": 1326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.945082529159693e-05,
+      "loss": 0.4596,
+      "step": 1327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9449707438367312e-05,
+      "loss": 0.4396,
+      "step": 1328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9448588469849266e-05,
+      "loss": 0.4257,
+      "step": 1329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9447468386129152e-05,
+      "loss": 0.4214,
+      "step": 1330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944634718729343e-05,
+      "loss": 0.4339,
+      "step": 1331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.944522487342864e-05,
+      "loss": 0.4317,
+      "step": 1332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9444101444621395e-05,
+      "loss": 0.3709,
+      "step": 1333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9442976900958416e-05,
+      "loss": 0.4663,
+      "step": 1334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9441851242526498e-05,
+      "loss": 0.426,
+      "step": 1335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9440724469412523e-05,
+      "loss": 0.4473,
+      "step": 1336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9439596581703462e-05,
+      "loss": 0.4596,
+      "step": 1337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943846757948637e-05,
+      "loss": 0.4275,
+      "step": 1338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9437337462848382e-05,
+      "loss": 0.36,
+      "step": 1339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9436206231876734e-05,
+      "loss": 0.424,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943507388665873e-05,
+      "loss": 0.4517,
+      "step": 1341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9433940427281778e-05,
+      "loss": 0.516,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9432805853833352e-05,
+      "loss": 0.445,
+      "step": 1343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9431670166401038e-05,
+      "loss": 0.4492,
+      "step": 1344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.943053336507248e-05,
+      "loss": 0.3952,
+      "step": 1345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942939544993543e-05,
+      "loss": 0.4373,
+      "step": 1346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942825642107771e-05,
+      "loss": 0.4048,
+      "step": 1347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9427116278587244e-05,
+      "loss": 0.4411,
+      "step": 1348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942597502255202e-05,
+      "loss": 0.4587,
+      "step": 1349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9424832653060136e-05,
+      "loss": 0.3937,
+      "step": 1350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942368917019976e-05,
+      "loss": 0.4017,
+      "step": 1351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.942254457405915e-05,
+      "loss": 0.4082,
+      "step": 1352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9421398864726653e-05,
+      "loss": 0.433,
+      "step": 1353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.94202520422907e-05,
+      "loss": 0.4285,
+      "step": 1354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9419104106839808e-05,
+      "loss": 0.4353,
+      "step": 1355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9417955058462572e-05,
+      "loss": 0.4617,
+      "step": 1356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.941680489724769e-05,
+      "loss": 0.3951,
+      "step": 1357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9415653623283927e-05,
+      "loss": 0.3939,
+      "step": 1358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9414501236660155e-05,
+      "loss": 0.3787,
+      "step": 1359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9413347737465316e-05,
+      "loss": 0.4571,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9412193125788435e-05,
+      "loss": 0.4149,
+      "step": 1361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9411037401718628e-05,
+      "loss": 0.4685,
+      "step": 1362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9409880565345113e-05,
+      "loss": 0.448,
+      "step": 1363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9408722616757166e-05,
+      "loss": 0.3675,
+      "step": 1364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9407563556044172e-05,
+      "loss": 0.439,
+      "step": 1365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9406403383295587e-05,
+      "loss": 0.4134,
+      "step": 1366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9405242098600953e-05,
+      "loss": 0.445,
+      "step": 1367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9404079702049914e-05,
+      "loss": 0.4164,
+      "step": 1368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.940291619373218e-05,
+      "loss": 0.4181,
+      "step": 1369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.940175157373756e-05,
+      "loss": 0.4095,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9400585842155938e-05,
+      "loss": 0.4769,
+      "step": 1371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9399418999077297e-05,
+      "loss": 0.4416,
+      "step": 1372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9398251044591696e-05,
+      "loss": 0.4778,
+      "step": 1373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9397081978789283e-05,
+      "loss": 0.4234,
+      "step": 1374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9395911801760287e-05,
+      "loss": 0.4722,
+      "step": 1375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9394740513595037e-05,
+      "loss": 0.4607,
+      "step": 1376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939356811438393e-05,
+      "loss": 0.4039,
+      "step": 1377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.939239460421746e-05,
+      "loss": 0.4253,
+      "step": 1378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.93912199831862e-05,
+      "loss": 0.3932,
+      "step": 1379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9390044251380817e-05,
+      "loss": 0.4808,
+      "step": 1380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9388867408892056e-05,
+      "loss": 0.4134,
+      "step": 1381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.938768945581075e-05,
+      "loss": 0.4523,
+      "step": 1382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9386510392227816e-05,
+      "loss": 0.41,
+      "step": 1383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9385330218234265e-05,
+      "loss": 0.3889,
+      "step": 1384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9384148933921184e-05,
+      "loss": 0.4687,
+      "step": 1385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9382966539379752e-05,
+      "loss": 0.4087,
+      "step": 1386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.938178303470123e-05,
+      "loss": 0.4385,
+      "step": 1387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.938059841997696e-05,
+      "loss": 0.4296,
+      "step": 1388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.937941269529838e-05,
+      "loss": 0.3991,
+      "step": 1389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9378225860757013e-05,
+      "loss": 0.4472,
+      "step": 1390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.937703791644446e-05,
+      "loss": 0.3586,
+      "step": 1391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9375848862452412e-05,
+      "loss": 0.4952,
+      "step": 1392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9374658698872648e-05,
+      "loss": 0.4248,
+      "step": 1393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9373467425797024e-05,
+      "loss": 0.4344,
+      "step": 1394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.937227504331749e-05,
+      "loss": 0.3923,
+      "step": 1395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.937108155152608e-05,
+      "loss": 0.4634,
+      "step": 1396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9369886950514918e-05,
+      "loss": 0.4653,
+      "step": 1397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9368691240376195e-05,
+      "loss": 0.4179,
+      "step": 1398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.936749442120221e-05,
+      "loss": 0.4749,
+      "step": 1399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9366296493085335e-05,
+      "loss": 0.4928,
+      "step": 1400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9365097456118033e-05,
+      "loss": 0.4263,
+      "step": 1401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9363897310392852e-05,
+      "loss": 0.4589,
+      "step": 1402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9362696056002425e-05,
+      "loss": 0.5166,
+      "step": 1403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9361493693039465e-05,
+      "loss": 0.4073,
+      "step": 1404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9360290221596775e-05,
+      "loss": 0.4045,
+      "step": 1405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935908564176725e-05,
+      "loss": 0.4667,
+      "step": 1406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9357879953643863e-05,
+      "loss": 0.4335,
+      "step": 1407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935667315731967e-05,
+      "loss": 0.418,
+      "step": 1408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9355465252887813e-05,
+      "loss": 0.3733,
+      "step": 1409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9354256240441538e-05,
+      "loss": 0.4008,
+      "step": 1410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9353046120074145e-05,
+      "loss": 0.5145,
+      "step": 1411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935183489187905e-05,
+      "loss": 0.4384,
+      "step": 1412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.935062255594973e-05,
+      "loss": 0.4293,
+      "step": 1413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9349409112379766e-05,
+      "loss": 0.4753,
+      "step": 1414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9348194561262806e-05,
+      "loss": 0.5074,
+      "step": 1415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9346978902692606e-05,
+      "loss": 0.4732,
+      "step": 1416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9345762136762992e-05,
+      "loss": 0.4371,
+      "step": 1417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9344544263567875e-05,
+      "loss": 0.4352,
+      "step": 1418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9343325283201257e-05,
+      "loss": 0.4318,
+      "step": 1419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9342105195757227e-05,
+      "loss": 0.4365,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.934088400132996e-05,
+      "loss": 0.4147,
+      "step": 1421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9339661700013696e-05,
+      "loss": 0.4453,
+      "step": 1422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9338438291902794e-05,
+      "loss": 0.4528,
+      "step": 1423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9337213777091675e-05,
+      "loss": 0.4491,
+      "step": 1424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.933598815567486e-05,
+      "loss": 0.4765,
+      "step": 1425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9334761427746935e-05,
+      "loss": 0.3745,
+      "step": 1426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9333533593402592e-05,
+      "loss": 0.3759,
+      "step": 1427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9332304652736594e-05,
+      "loss": 0.4613,
+      "step": 1428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9331074605843805e-05,
+      "loss": 0.4266,
+      "step": 1429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.932984345281916e-05,
+      "loss": 0.4078,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9328611193757685e-05,
+      "loss": 0.4274,
+      "step": 1431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9327377828754488e-05,
+      "loss": 0.488,
+      "step": 1432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9326143357904768e-05,
+      "loss": 0.4021,
+      "step": 1433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.932490778130381e-05,
+      "loss": 0.4618,
+      "step": 1434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9323671099046977e-05,
+      "loss": 0.432,
+      "step": 1435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.932243331122972e-05,
+      "loss": 0.4454,
+      "step": 1436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9321194417947578e-05,
+      "loss": 0.4772,
+      "step": 1437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.931995441929617e-05,
+      "loss": 0.4395,
+      "step": 1438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9318713315371213e-05,
+      "loss": 0.4211,
+      "step": 1439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.93174711062685e-05,
+      "loss": 0.4471,
+      "step": 1440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9316227792083897e-05,
+      "loss": 0.4548,
+      "step": 1441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9314983372913384e-05,
+      "loss": 0.4534,
+      "step": 1442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9313737848853e-05,
+      "loss": 0.4087,
+      "step": 1443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9312491219998885e-05,
+      "loss": 0.4395,
+      "step": 1444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9311243486447253e-05,
+      "loss": 0.4225,
+      "step": 1445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.930999464829442e-05,
+      "loss": 0.4395,
+      "step": 1446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9308744705636768e-05,
+      "loss": 0.4266,
+      "step": 1447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.930749365857077e-05,
+      "loss": 0.4397,
+      "step": 1448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9306241507192995e-05,
+      "loss": 0.4768,
+      "step": 1449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9304988251600084e-05,
+      "loss": 0.418,
+      "step": 1450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9303733891888775e-05,
+      "loss": 0.4271,
+      "step": 1451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9302478428155876e-05,
+      "loss": 0.4459,
+      "step": 1452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.93012218604983e-05,
+      "loss": 0.4035,
+      "step": 1453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9299964189013016e-05,
+      "loss": 0.4079,
+      "step": 1454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9298705413797114e-05,
+      "loss": 0.4054,
+      "step": 1455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9297445534947742e-05,
+      "loss": 0.4045,
+      "step": 1456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9296184552562152e-05,
+      "loss": 0.4798,
+      "step": 1457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9294922466737658e-05,
+      "loss": 0.4099,
+      "step": 1458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9293659277571685e-05,
+      "loss": 0.454,
+      "step": 1459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9292394985161725e-05,
+      "loss": 0.4074,
+      "step": 1460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.929112958960536e-05,
+      "loss": 0.4353,
+      "step": 1461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9289863091000267e-05,
+      "loss": 0.3885,
+      "step": 1462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928859548944419e-05,
+      "loss": 0.4382,
+      "step": 1463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9287326785034975e-05,
+      "loss": 0.4285,
+      "step": 1464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9286056977870542e-05,
+      "loss": 0.4066,
+      "step": 1465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.92847860680489e-05,
+      "loss": 0.466,
+      "step": 1466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9283514055668144e-05,
+      "loss": 0.4369,
+      "step": 1467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928224094082645e-05,
+      "loss": 0.4545,
+      "step": 1468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.928096672362209e-05,
+      "loss": 0.4125,
+      "step": 1469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9279691404153408e-05,
+      "loss": 0.4743,
+      "step": 1470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9278414982518834e-05,
+      "loss": 0.4627,
+      "step": 1471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9277137458816903e-05,
+      "loss": 0.4122,
+      "step": 1472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9275858833146202e-05,
+      "loss": 0.3674,
+      "step": 1473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9274579105605427e-05,
+      "loss": 0.4323,
+      "step": 1474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9273298276293356e-05,
+      "loss": 0.4305,
+      "step": 1475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9272016345308848e-05,
+      "loss": 0.3776,
+      "step": 1476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9270733312750844e-05,
+      "loss": 0.4102,
+      "step": 1477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9269449178718374e-05,
+      "loss": 0.3546,
+      "step": 1478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9268163943310555e-05,
+      "loss": 0.41,
+      "step": 1479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926687760662659e-05,
+      "loss": 0.4846,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9265590168765757e-05,
+      "loss": 0.3805,
+      "step": 1481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.926430162982743e-05,
+      "loss": 0.3861,
+      "step": 1482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9263011989911065e-05,
+      "loss": 0.4366,
+      "step": 1483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9261721249116197e-05,
+      "loss": 0.4283,
+      "step": 1484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9260429407542453e-05,
+      "loss": 0.4271,
+      "step": 1485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9259136465289544e-05,
+      "loss": 0.3997,
+      "step": 1486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9257842422457264e-05,
+      "loss": 0.4593,
+      "step": 1487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925654727914549e-05,
+      "loss": 0.4401,
+      "step": 1488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9255251035454194e-05,
+      "loss": 0.4316,
+      "step": 1489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9253953691483417e-05,
+      "loss": 0.4467,
+      "step": 1490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9252655247333294e-05,
+      "loss": 0.3899,
+      "step": 1491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9251355703104055e-05,
+      "loss": 0.3683,
+      "step": 1492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.925005505889599e-05,
+      "loss": 0.4252,
+      "step": 1493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9248753314809497e-05,
+      "loss": 0.4566,
+      "step": 1494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9247450470945044e-05,
+      "loss": 0.4223,
+      "step": 1495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9246146527403195e-05,
+      "loss": 0.4172,
+      "step": 1496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.924484148428459e-05,
+      "loss": 0.4319,
+      "step": 1497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9243535341689965e-05,
+      "loss": 0.3735,
+      "step": 1498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9242228099720124e-05,
+      "loss": 0.3844,
+      "step": 1499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9240919758475973e-05,
+      "loss": 0.4404,
+      "step": 1500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923961031805849e-05,
+      "loss": 0.4733,
+      "step": 1501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9238299778568743e-05,
+      "loss": 0.4269,
+      "step": 1502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923698814010789e-05,
+      "loss": 0.4511,
+      "step": 1503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923567540277716e-05,
+      "loss": 0.4394,
+      "step": 1504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9234361566677885e-05,
+      "loss": 0.4245,
+      "step": 1505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.923304663191147e-05,
+      "loss": 0.4477,
+      "step": 1506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.92317305985794e-05,
+      "loss": 0.3621,
+      "step": 1507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9230413466783265e-05,
+      "loss": 0.4652,
+      "step": 1508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9229095236624713e-05,
+      "loss": 0.4105,
+      "step": 1509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9227775908205497e-05,
+      "loss": 0.4288,
+      "step": 1510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.922645548162745e-05,
+      "loss": 0.4796,
+      "step": 1511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9225133956992484e-05,
+      "loss": 0.4118,
+      "step": 1512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.92238113344026e-05,
+      "loss": 0.4666,
+      "step": 1513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.922248761395989e-05,
+      "loss": 0.4565,
+      "step": 1514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9221162795766515e-05,
+      "loss": 0.4494,
+      "step": 1515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9219836879924737e-05,
+      "loss": 0.403,
+      "step": 1516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.921850986653689e-05,
+      "loss": 0.4112,
+      "step": 1517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9217181755705404e-05,
+      "loss": 0.4126,
+      "step": 1518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9215852547532785e-05,
+      "loss": 0.4421,
+      "step": 1519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9214522242121627e-05,
+      "loss": 0.4684,
+      "step": 1520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9213190839574605e-05,
+      "loss": 0.4242,
+      "step": 1521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9211858339994487e-05,
+      "loss": 0.3879,
+      "step": 1522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9210524743484125e-05,
+      "loss": 0.4661,
+      "step": 1523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9209190050146442e-05,
+      "loss": 0.4482,
+      "step": 1524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9207854260084458e-05,
+      "loss": 0.4644,
+      "step": 1525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9206517373401284e-05,
+      "loss": 0.4352,
+      "step": 1526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9205179390200088e-05,
+      "loss": 0.4037,
+      "step": 1527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9203840310584162e-05,
+      "loss": 0.4065,
+      "step": 1528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9202500134656843e-05,
+      "loss": 0.4108,
+      "step": 1529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9201158862521588e-05,
+      "loss": 0.4735,
+      "step": 1530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9199816494281907e-05,
+      "loss": 0.4724,
+      "step": 1531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.919847303004142e-05,
+      "loss": 0.4344,
+      "step": 1532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9197128469903818e-05,
+      "loss": 0.4327,
+      "step": 1533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.919578281397288e-05,
+      "loss": 0.4148,
+      "step": 1534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.919443606235246e-05,
+      "loss": 0.388,
+      "step": 1535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9193088215146525e-05,
+      "loss": 0.3931,
+      "step": 1536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.919173927245909e-05,
+      "loss": 0.4188,
+      "step": 1537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9190389234394285e-05,
+      "loss": 0.4691,
+      "step": 1538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9189038101056304e-05,
+      "loss": 0.4105,
+      "step": 1539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9187685872549433e-05,
+      "loss": 0.4533,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9186332548978045e-05,
+      "loss": 0.4051,
+      "step": 1541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9184978130446593e-05,
+      "loss": 0.4572,
+      "step": 1542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.918362261705962e-05,
+      "loss": 0.3961,
+      "step": 1543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9182266008921748e-05,
+      "loss": 0.4444,
+      "step": 1544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9180908306137683e-05,
+      "loss": 0.4288,
+      "step": 1545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9179549508812222e-05,
+      "loss": 0.4288,
+      "step": 1546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.917818961705024e-05,
+      "loss": 0.4191,
+      "step": 1547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.91768286309567e-05,
+      "loss": 0.4224,
+      "step": 1548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9175466550636657e-05,
+      "loss": 0.4419,
+      "step": 1549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.917410337619523e-05,
+      "loss": 0.4749,
+      "step": 1550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9172739107737633e-05,
+      "loss": 0.438,
+      "step": 1551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.917137374536918e-05,
+      "loss": 0.4744,
+      "step": 1552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.917000728919524e-05,
+      "loss": 0.4446,
+      "step": 1553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9168639739321294e-05,
+      "loss": 0.4071,
+      "step": 1554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9167271095852886e-05,
+      "loss": 0.3987,
+      "step": 1555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916590135889566e-05,
+      "loss": 0.4494,
+      "step": 1556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9164530528555332e-05,
+      "loss": 0.429,
+      "step": 1557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9163158604937718e-05,
+      "loss": 0.4373,
+      "step": 1558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.91617855881487e-05,
+      "loss": 0.4661,
+      "step": 1559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.916041147829425e-05,
+      "loss": 0.3731,
+      "step": 1560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.915903627548044e-05,
+      "loss": 0.435,
+      "step": 1561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9157659979813407e-05,
+      "loss": 0.446,
+      "step": 1562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9156282591399374e-05,
+      "loss": 0.5041,
+      "step": 1563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9154904110344665e-05,
+      "loss": 0.4476,
+      "step": 1564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9153524536755665e-05,
+      "loss": 0.3857,
+      "step": 1565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9152143870738864e-05,
+      "loss": 0.4358,
+      "step": 1566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.915076211240083e-05,
+      "loss": 0.4172,
+      "step": 1567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.91493792618482e-05,
+      "loss": 0.4301,
+      "step": 1568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9147995319187716e-05,
+      "loss": 0.4961,
+      "step": 1569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9146610284526197e-05,
+      "loss": 0.4046,
+      "step": 1570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9145224157970543e-05,
+      "loss": 0.443,
+      "step": 1571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9143836939627745e-05,
+      "loss": 0.4163,
+      "step": 1572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.914244862960487e-05,
+      "loss": 0.4206,
+      "step": 1573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9141059228009078e-05,
+      "loss": 0.4559,
+      "step": 1574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9139668734947606e-05,
+      "loss": 0.4192,
+      "step": 1575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9138277150527776e-05,
+      "loss": 0.4465,
+      "step": 1576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9136884474857e-05,
+      "loss": 0.4833,
+      "step": 1577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.913549070804277e-05,
+      "loss": 0.4391,
+      "step": 1578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.913409585019266e-05,
+      "loss": 0.4022,
+      "step": 1579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9132699901414333e-05,
+      "loss": 0.5099,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9131302861815535e-05,
+      "loss": 0.4516,
+      "step": 1581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9129904731504092e-05,
+      "loss": 0.5956,
+      "step": 1582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912850551058792e-05,
+      "loss": 0.5467,
+      "step": 1583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9127105199175018e-05,
+      "loss": 0.4213,
+      "step": 1584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9125703797373465e-05,
+      "loss": 0.3577,
+      "step": 1585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9124301305291428e-05,
+      "loss": 0.3983,
+      "step": 1586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912289772303716e-05,
+      "loss": 0.477,
+      "step": 1587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.912149305071899e-05,
+      "loss": 0.3851,
+      "step": 1588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9120087288445337e-05,
+      "loss": 0.3853,
+      "step": 1589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.911868043632471e-05,
+      "loss": 0.4591,
+      "step": 1590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.911727249446569e-05,
+      "loss": 0.405,
+      "step": 1591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9115863462976946e-05,
+      "loss": 0.4059,
+      "step": 1592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9114453341967237e-05,
+      "loss": 0.4354,
+      "step": 1593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9113042131545405e-05,
+      "loss": 0.474,
+      "step": 1594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.911162983182037e-05,
+      "loss": 0.4465,
+      "step": 1595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9110216442901133e-05,
+      "loss": 0.446,
+      "step": 1596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9108801964896793e-05,
+      "loss": 0.3882,
+      "step": 1597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9107386397916523e-05,
+      "loss": 0.4495,
+      "step": 1598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9105969742069577e-05,
+      "loss": 0.4033,
+      "step": 1599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.910455199746531e-05,
+      "loss": 0.378,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9103133164213142e-05,
+      "loss": 0.4425,
+      "step": 1601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9101713242422583e-05,
+      "loss": 0.4494,
+      "step": 1602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9100292232203235e-05,
+      "loss": 0.4285,
+      "step": 1603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.909887013366477e-05,
+      "loss": 0.4111,
+      "step": 1604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9097446946916956e-05,
+      "loss": 0.4431,
+      "step": 1605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9096022672069636e-05,
+      "loss": 0.4743,
+      "step": 1606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9094597309232748e-05,
+      "loss": 0.4591,
+      "step": 1607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9093170858516304e-05,
+      "loss": 0.4077,
+      "step": 1608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9091743320030402e-05,
+      "loss": 0.4796,
+      "step": 1609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.909031469388523e-05,
+      "loss": 0.443,
+      "step": 1610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9088884980191048e-05,
+      "loss": 0.452,
+      "step": 1611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908745417905821e-05,
+      "loss": 0.4084,
+      "step": 1612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9086022290597156e-05,
+      "loss": 0.4266,
+      "step": 1613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9084589314918402e-05,
+      "loss": 0.4387,
+      "step": 1614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908315525213255e-05,
+      "loss": 0.4241,
+      "step": 1615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9081720102350292e-05,
+      "loss": 0.4622,
+      "step": 1616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.908028386568239e-05,
+      "loss": 0.4266,
+      "step": 1617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.90788465422397e-05,
+      "loss": 0.4146,
+      "step": 1618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9077408132133164e-05,
+      "loss": 0.4164,
+      "step": 1619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.907596863547381e-05,
+      "loss": 0.4538,
+      "step": 1620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.907452805237274e-05,
+      "loss": 0.4455,
+      "step": 1621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9073086382941135e-05,
+      "loss": 0.4143,
+      "step": 1622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9071643627290282e-05,
+      "loss": 0.3973,
+      "step": 1623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9070199785531527e-05,
+      "loss": 0.4331,
+      "step": 1624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9068754857776323e-05,
+      "loss": 0.4064,
+      "step": 1625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9067308844136193e-05,
+      "loss": 0.4725,
+      "step": 1626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.906586174472274e-05,
+      "loss": 0.3528,
+      "step": 1627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9064413559647663e-05,
+      "loss": 0.4206,
+      "step": 1628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.906296428902274e-05,
+      "loss": 0.4976,
+      "step": 1629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9061513932959827e-05,
+      "loss": 0.4353,
+      "step": 1630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9060062491570867e-05,
+      "loss": 0.4008,
+      "step": 1631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9058609964967897e-05,
+      "loss": 0.4234,
+      "step": 1632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9057156353263022e-05,
+      "loss": 0.4482,
+      "step": 1633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9055701656568437e-05,
+      "loss": 0.4065,
+      "step": 1634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.905424587499643e-05,
+      "loss": 0.4213,
+      "step": 1635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9052789008659354e-05,
+      "loss": 0.427,
+      "step": 1636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9051331057669658e-05,
+      "loss": 0.5335,
+      "step": 1637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9049872022139877e-05,
+      "loss": 0.4407,
+      "step": 1638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9048411902182627e-05,
+      "loss": 0.4078,
+      "step": 1639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9046950697910603e-05,
+      "loss": 0.3792,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904548840943658e-05,
+      "loss": 0.3993,
+      "step": 1641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9044025036873435e-05,
+      "loss": 0.465,
+      "step": 1642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904256058033411e-05,
+      "loss": 0.4053,
+      "step": 1643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.904109503993164e-05,
+      "loss": 0.4452,
+      "step": 1644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.903962841577914e-05,
+      "loss": 0.4774,
+      "step": 1645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9038160707989816e-05,
+      "loss": 0.4522,
+      "step": 1646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9036691916676943e-05,
+      "loss": 0.4231,
+      "step": 1647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9035222041953897e-05,
+      "loss": 0.4866,
+      "step": 1648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.903375108393412e-05,
+      "loss": 0.4573,
+      "step": 1649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9032279042731154e-05,
+      "loss": 0.3824,
+      "step": 1650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9030805918458615e-05,
+      "loss": 0.4287,
+      "step": 1651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9029331711230206e-05,
+      "loss": 0.4358,
+      "step": 1652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.902785642115971e-05,
+      "loss": 0.4339,
+      "step": 1653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9026380048360996e-05,
+      "loss": 0.4296,
+      "step": 1654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9024902592948016e-05,
+      "loss": 0.412,
+      "step": 1655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9023424055034813e-05,
+      "loss": 0.4572,
+      "step": 1656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9021944434735497e-05,
+      "loss": 0.3802,
+      "step": 1657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9020463732164278e-05,
+      "loss": 0.4571,
+      "step": 1658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9018981947435438e-05,
+      "loss": 0.4233,
+      "step": 1659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9017499080663353e-05,
+      "loss": 0.4367,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9016015131962474e-05,
+      "loss": 0.4653,
+      "step": 1661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9014530101447334e-05,
+      "loss": 0.4293,
+      "step": 1662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.901304398923256e-05,
+      "loss": 0.4201,
+      "step": 1663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.901155679543285e-05,
+      "loss": 0.4339,
+      "step": 1664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9010068520163e-05,
+      "loss": 0.3893,
+      "step": 1665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.900857916353788e-05,
+      "loss": 0.458,
+      "step": 1666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9007088725672438e-05,
+      "loss": 0.4238,
+      "step": 1667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9005597206681715e-05,
+      "loss": 0.4213,
+      "step": 1668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9004104606680833e-05,
+      "loss": 0.4353,
+      "step": 1669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9002610925785002e-05,
+      "loss": 0.4206,
+      "step": 1670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9001116164109503e-05,
+      "loss": 0.4496,
+      "step": 1671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.899962032176971e-05,
+      "loss": 0.4236,
+      "step": 1672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8998123398881085e-05,
+      "loss": 0.4207,
+      "step": 1673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8996625395559156e-05,
+      "loss": 0.4758,
+      "step": 1674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8995126311919548e-05,
+      "loss": 0.3932,
+      "step": 1675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8993626148077975e-05,
+      "loss": 0.428,
+      "step": 1676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8992124904150217e-05,
+      "loss": 0.4562,
+      "step": 1677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.899062258025215e-05,
+      "loss": 0.422,
+      "step": 1678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8989119176499727e-05,
+      "loss": 0.4169,
+      "step": 1679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8987614693008993e-05,
+      "loss": 0.4277,
+      "step": 1680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8986109129896058e-05,
+      "loss": 0.4007,
+      "step": 1681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8984602487277142e-05,
+      "loss": 0.5029,
+      "step": 1682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8983094765268522e-05,
+      "loss": 0.4303,
+      "step": 1683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.898158596398658e-05,
+      "loss": 0.4455,
+      "step": 1684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.898007608354776e-05,
+      "loss": 0.4379,
+      "step": 1685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8978565124068614e-05,
+      "loss": 0.4606,
+      "step": 1686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.897705308566576e-05,
+      "loss": 0.4076,
+      "step": 1687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89755399684559e-05,
+      "loss": 0.3725,
+      "step": 1688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8974025772555823e-05,
+      "loss": 0.4138,
+      "step": 1689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89725104980824e-05,
+      "loss": 0.4159,
+      "step": 1690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8970994145152588e-05,
+      "loss": 0.4264,
+      "step": 1691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896947671388343e-05,
+      "loss": 0.4162,
+      "step": 1692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896795820439204e-05,
+      "loss": 0.4637,
+      "step": 1693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8966438616795624e-05,
+      "loss": 0.4433,
+      "step": 1694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896491795121148e-05,
+      "loss": 0.4276,
+      "step": 1695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8963396207756965e-05,
+      "loss": 0.4075,
+      "step": 1696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.896187338654954e-05,
+      "loss": 0.4142,
+      "step": 1697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8960349487706746e-05,
+      "loss": 0.3876,
+      "step": 1698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89588245113462e-05,
+      "loss": 0.4915,
+      "step": 1699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8957298457585605e-05,
+      "loss": 0.4098,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895577132654275e-05,
+      "loss": 0.4146,
+      "step": 1701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8954243118335506e-05,
+      "loss": 0.4124,
+      "step": 1702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.895271383308182e-05,
+      "loss": 0.4307,
+      "step": 1703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8951183470899742e-05,
+      "loss": 0.4365,
+      "step": 1704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894965203190738e-05,
+      "loss": 0.4151,
+      "step": 1705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8948119516222942e-05,
+      "loss": 0.4345,
+      "step": 1706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.894658592396471e-05,
+      "loss": 0.378,
+      "step": 1707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8945051255251055e-05,
+      "loss": 0.3954,
+      "step": 1708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8943515510200436e-05,
+      "loss": 0.4676,
+      "step": 1709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8941978688931373e-05,
+      "loss": 0.4139,
+      "step": 1710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8940440791562502e-05,
+      "loss": 0.4295,
+      "step": 1711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.893890181821251e-05,
+      "loss": 0.4223,
+      "step": 1712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8937361769000188e-05,
+      "loss": 0.4068,
+      "step": 1713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8935820644044398e-05,
+      "loss": 0.4987,
+      "step": 1714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.89342784434641e-05,
+      "loss": 0.4031,
+      "step": 1715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.893273516737832e-05,
+      "loss": 0.3994,
+      "step": 1716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8931190815906175e-05,
+      "loss": 0.4327,
+      "step": 1717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8929645389166867e-05,
+      "loss": 0.4067,
+      "step": 1718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.892809888727968e-05,
+      "loss": 0.3985,
+      "step": 1719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8926551310363973e-05,
+      "loss": 0.3722,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8925002658539203e-05,
+      "loss": 0.4327,
+      "step": 1721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.892345293192489e-05,
+      "loss": 0.4209,
+      "step": 1722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8921902130640663e-05,
+      "loss": 0.4326,
+      "step": 1723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8920350254806205e-05,
+      "loss": 0.4151,
+      "step": 1724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.89187973045413e-05,
+      "loss": 0.3882,
+      "step": 1725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8917243279965818e-05,
+      "loss": 0.3959,
+      "step": 1726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8915688181199703e-05,
+      "loss": 0.4689,
+      "step": 1727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.891413200836298e-05,
+      "loss": 0.3918,
+      "step": 1728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.891257476157576e-05,
+      "loss": 0.4088,
+      "step": 1729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8911016440958243e-05,
+      "loss": 0.4133,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8909457046630703e-05,
+      "loss": 0.4293,
+      "step": 1731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.89078965787135e-05,
+      "loss": 0.3906,
+      "step": 1732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8906335037327075e-05,
+      "loss": 0.4467,
+      "step": 1733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8904772422591966e-05,
+      "loss": 0.4163,
+      "step": 1734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.890320873462877e-05,
+      "loss": 0.4103,
+      "step": 1735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8901643973558188e-05,
+      "loss": 0.4616,
+      "step": 1736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8900078139500984e-05,
+      "loss": 0.4093,
+      "step": 1737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8898511232578023e-05,
+      "loss": 0.4432,
+      "step": 1738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8896943252910244e-05,
+      "loss": 0.4001,
+      "step": 1739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8895374200618673e-05,
+      "loss": 0.3964,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.889380407582441e-05,
+      "loss": 0.3959,
+      "step": 1741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8892232878648647e-05,
+      "loss": 0.4308,
+      "step": 1742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8890660609212657e-05,
+      "loss": 0.4037,
+      "step": 1743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888908726763779e-05,
+      "loss": 0.4821,
+      "step": 1744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888751285404549e-05,
+      "loss": 0.4481,
+      "step": 1745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8885937368557265e-05,
+      "loss": 0.3918,
+      "step": 1746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.888436081129473e-05,
+      "loss": 0.4498,
+      "step": 1747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8882783182379567e-05,
+      "loss": 0.405,
+      "step": 1748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8881204481933536e-05,
+      "loss": 0.4402,
+      "step": 1749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8879624710078497e-05,
+      "loss": 0.3924,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887804386693638e-05,
+      "loss": 0.4064,
+      "step": 1751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.88764619526292e-05,
+      "loss": 0.4144,
+      "step": 1752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887487896727906e-05,
+      "loss": 0.3939,
+      "step": 1753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.887329491100814e-05,
+      "loss": 0.3943,
+      "step": 1754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8871709783938694e-05,
+      "loss": 0.4431,
+      "step": 1755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8870123586193084e-05,
+      "loss": 0.4259,
+      "step": 1756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8868536317893727e-05,
+      "loss": 0.4537,
+      "step": 1757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8866947979163146e-05,
+      "loss": 0.4608,
+      "step": 1758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8865358570123926e-05,
+      "loss": 0.4757,
+      "step": 1759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.886376809089875e-05,
+      "loss": 0.3672,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8862176541610377e-05,
+      "loss": 0.3759,
+      "step": 1761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8860583922381647e-05,
+      "loss": 0.5061,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.885899023333548e-05,
+      "loss": 0.4471,
+      "step": 1763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8857395474594897e-05,
+      "loss": 0.4229,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.885579964628298e-05,
+      "loss": 0.4268,
+      "step": 1765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.88542027485229e-05,
+      "loss": 0.4317,
+      "step": 1766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8852604781437916e-05,
+      "loss": 0.4631,
+      "step": 1767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8851005745151366e-05,
+      "loss": 0.3753,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8849405639786668e-05,
+      "loss": 0.424,
+      "step": 1769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884780446546732e-05,
+      "loss": 0.4024,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884620222231692e-05,
+      "loss": 0.4173,
+      "step": 1771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.884459891045913e-05,
+      "loss": 0.4032,
+      "step": 1772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8842994530017692e-05,
+      "loss": 0.4108,
+      "step": 1773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8841389081116447e-05,
+      "loss": 0.4804,
+      "step": 1774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.883978256387931e-05,
+      "loss": 0.4388,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8838174978430278e-05,
+      "loss": 0.4559,
+      "step": 1776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.883656632489343e-05,
+      "loss": 0.393,
+      "step": 1777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.883495660339293e-05,
+      "loss": 0.441,
+      "step": 1778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.883334581405302e-05,
+      "loss": 0.5091,
+      "step": 1779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8831733956998028e-05,
+      "loss": 0.4233,
+      "step": 1780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.883012103235237e-05,
+      "loss": 0.3911,
+      "step": 1781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.882850704024053e-05,
+      "loss": 0.4106,
+      "step": 1782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.882689198078709e-05,
+      "loss": 0.3752,
+      "step": 1783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.88252758541167e-05,
+      "loss": 0.4349,
+      "step": 1784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.88236586603541e-05,
+      "loss": 0.4018,
+      "step": 1785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8822040399624116e-05,
+      "loss": 0.4686,
+      "step": 1786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.882042107205166e-05,
+      "loss": 0.4507,
+      "step": 1787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8818800677761696e-05,
+      "loss": 0.4088,
+      "step": 1788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8817179216879308e-05,
+      "loss": 0.4451,
+      "step": 1789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.881555668952965e-05,
+      "loss": 0.4274,
+      "step": 1790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8813933095837945e-05,
+      "loss": 0.4289,
+      "step": 1791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8812308435929515e-05,
+      "loss": 0.3907,
+      "step": 1792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8810682709929754e-05,
+      "loss": 0.4438,
+      "step": 1793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8809055917964152e-05,
+      "loss": 0.3946,
+      "step": 1794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.880742806015826e-05,
+      "loss": 0.4215,
+      "step": 1795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.880579913663773e-05,
+      "loss": 0.3545,
+      "step": 1796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8804169147528284e-05,
+      "loss": 0.4729,
+      "step": 1797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.880253809295573e-05,
+      "loss": 0.3906,
+      "step": 1798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.880090597304597e-05,
+      "loss": 0.4237,
+      "step": 1799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.879927278792497e-05,
+      "loss": 0.4782,
+      "step": 1800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8797638537718783e-05,
+      "loss": 0.497,
+      "step": 1801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8796003222553558e-05,
+      "loss": 0.3915,
+      "step": 1802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.87943668425555e-05,
+      "loss": 0.3933,
+      "step": 1803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8792729397850927e-05,
+      "loss": 0.3804,
+      "step": 1804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8791090888566214e-05,
+      "loss": 0.4059,
+      "step": 1805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8789451314827837e-05,
+      "loss": 0.4061,
+      "step": 1806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8787810676762333e-05,
+      "loss": 0.4258,
+      "step": 1807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8786168974496344e-05,
+      "loss": 0.4251,
+      "step": 1808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8784526208156575e-05,
+      "loss": 0.4153,
+      "step": 1809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8782882377869827e-05,
+      "loss": 0.4833,
+      "step": 1810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8781237483762975e-05,
+      "loss": 0.3962,
+      "step": 1811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.877959152596298e-05,
+      "loss": 0.4037,
+      "step": 1812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.877794450459688e-05,
+      "loss": 0.4035,
+      "step": 1813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8776296419791812e-05,
+      "loss": 0.4937,
+      "step": 1814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8774647271674968e-05,
+      "loss": 0.43,
+      "step": 1815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.877299706037364e-05,
+      "loss": 0.4425,
+      "step": 1816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8771345786015197e-05,
+      "loss": 0.452,
+      "step": 1817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8769693448727096e-05,
+      "loss": 0.4117,
+      "step": 1818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.876804004863687e-05,
+      "loss": 0.4093,
+      "step": 1819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8766385585872132e-05,
+      "loss": 0.4137,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8764730060560584e-05,
+      "loss": 0.4428,
+      "step": 1821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.876307347283e-05,
+      "loss": 0.4681,
+      "step": 1822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8761415822808245e-05,
+      "loss": 0.4472,
+      "step": 1823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8759757110623273e-05,
+      "loss": 0.4801,
+      "step": 1824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8758097336403096e-05,
+      "loss": 0.4481,
+      "step": 1825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875643650027583e-05,
+      "loss": 0.4455,
+      "step": 1826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8754774602369666e-05,
+      "loss": 0.4435,
+      "step": 1827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.875311164281287e-05,
+      "loss": 0.469,
+      "step": 1828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8751447621733804e-05,
+      "loss": 0.4568,
+      "step": 1829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8749782539260894e-05,
+      "loss": 0.3686,
+      "step": 1830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8748116395522672e-05,
+      "loss": 0.4923,
+      "step": 1831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8746449190647723e-05,
+      "loss": 0.3994,
+      "step": 1832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.874478092476474e-05,
+      "loss": 0.4361,
+      "step": 1833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8743111598002485e-05,
+      "loss": 0.4119,
+      "step": 1834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8741441210489796e-05,
+      "loss": 0.4734,
+      "step": 1835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8739769762355606e-05,
+      "loss": 0.4742,
+      "step": 1836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8738097253728923e-05,
+      "loss": 0.4521,
+      "step": 1837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8736423684738844e-05,
+      "loss": 0.44,
+      "step": 1838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.873474905551453e-05,
+      "loss": 0.4272,
+      "step": 1839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.873307336618525e-05,
+      "loss": 0.4092,
+      "step": 1840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.873139661688033e-05,
+      "loss": 0.4433,
+      "step": 1841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.872971880772919e-05,
+      "loss": 0.4083,
+      "step": 1842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.872803993886133e-05,
+      "loss": 0.389,
+      "step": 1843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.872636001040634e-05,
+      "loss": 0.3714,
+      "step": 1844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.872467902249388e-05,
+      "loss": 0.4443,
+      "step": 1845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.872299697525369e-05,
+      "loss": 0.4923,
+      "step": 1846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.87213138688156e-05,
+      "loss": 0.4281,
+      "step": 1847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8719629703309523e-05,
+      "loss": 0.3974,
+      "step": 1848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8717944478865447e-05,
+      "loss": 0.446,
+      "step": 1849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8716258195613443e-05,
+      "loss": 0.3891,
+      "step": 1850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8714570853683666e-05,
+      "loss": 0.419,
+      "step": 1851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8712882453206356e-05,
+      "loss": 0.4452,
+      "step": 1852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.871119299431183e-05,
+      "loss": 0.4931,
+      "step": 1853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.870950247713048e-05,
+      "loss": 0.4934,
+      "step": 1854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8707810901792798e-05,
+      "loss": 0.4318,
+      "step": 1855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8706118268429334e-05,
+      "loss": 0.4155,
+      "step": 1856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8704424577170745e-05,
+      "loss": 0.3941,
+      "step": 1857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8702729828147753e-05,
+      "loss": 0.3893,
+      "step": 1858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8701034021491162e-05,
+      "loss": 0.4535,
+      "step": 1859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8699337157331863e-05,
+      "loss": 0.427,
+      "step": 1860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.869763923580083e-05,
+      "loss": 0.418,
+      "step": 1861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8695940257029116e-05,
+      "loss": 0.457,
+      "step": 1862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.869424022114785e-05,
+      "loss": 0.4373,
+      "step": 1863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.869253912828825e-05,
+      "loss": 0.4354,
+      "step": 1864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.869083697858162e-05,
+      "loss": 0.4866,
+      "step": 1865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8689133772159327e-05,
+      "loss": 0.51,
+      "step": 1866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.868742950915284e-05,
+      "loss": 0.3885,
+      "step": 1867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.86857241896937e-05,
+      "loss": 0.4186,
+      "step": 1868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.868401781391353e-05,
+      "loss": 0.45,
+      "step": 1869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.868231038194403e-05,
+      "loss": 0.4279,
+      "step": 1870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8680601893917e-05,
+      "loss": 0.4145,
+      "step": 1871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8678892349964295e-05,
+      "loss": 0.4598,
+      "step": 1872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.867718175021787e-05,
+      "loss": 0.4368,
+      "step": 1873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.867547009480975e-05,
+      "loss": 0.4115,
+      "step": 1874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8673757383872058e-05,
+      "loss": 0.3989,
+      "step": 1875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8672043617536984e-05,
+      "loss": 0.3768,
+      "step": 1876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8670328795936805e-05,
+      "loss": 0.3898,
+      "step": 1877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8668612919203873e-05,
+      "loss": 0.428,
+      "step": 1878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8666895987470625e-05,
+      "loss": 0.4209,
+      "step": 1879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8665178000869592e-05,
+      "loss": 0.4135,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8663458959533366e-05,
+      "loss": 0.3794,
+      "step": 1881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8661738863594628e-05,
+      "loss": 0.4178,
+      "step": 1882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8660017713186153e-05,
+      "loss": 0.4166,
+      "step": 1883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.865829550844078e-05,
+      "loss": 0.4512,
+      "step": 1884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8656572249491436e-05,
+      "loss": 0.4303,
+      "step": 1885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8654847936471126e-05,
+      "loss": 0.4605,
+      "step": 1886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8653122569512945e-05,
+      "loss": 0.3957,
+      "step": 1887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8651396148750062e-05,
+      "loss": 0.4089,
+      "step": 1888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8649668674315726e-05,
+      "loss": 0.4391,
+      "step": 1889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8647940146343278e-05,
+      "loss": 0.3921,
+      "step": 1890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8646210564966132e-05,
+      "loss": 0.3797,
+      "step": 1891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8644479930317776e-05,
+      "loss": 0.4186,
+      "step": 1892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8642748242531795e-05,
+      "loss": 0.4306,
+      "step": 1893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8641015501741843e-05,
+      "loss": 0.4347,
+      "step": 1894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8639281708081665e-05,
+      "loss": 0.3682,
+      "step": 1895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.863754686168508e-05,
+      "loss": 0.4458,
+      "step": 1896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8635810962685987e-05,
+      "loss": 0.4166,
+      "step": 1897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8634074011218382e-05,
+      "loss": 0.4445,
+      "step": 1898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8632336007416314e-05,
+      "loss": 0.4112,
+      "step": 1899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.863059695141394e-05,
+      "loss": 0.4248,
+      "step": 1900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8628856843345487e-05,
+      "loss": 0.4166,
+      "step": 1901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8627115683345263e-05,
+      "loss": 0.4029,
+      "step": 1902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862537347154765e-05,
+      "loss": 0.4192,
+      "step": 1903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862363020808713e-05,
+      "loss": 0.4101,
+      "step": 1904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8621885893098245e-05,
+      "loss": 0.4752,
+      "step": 1905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.862014052671564e-05,
+      "loss": 0.411,
+      "step": 1906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8618394109074022e-05,
+      "loss": 0.417,
+      "step": 1907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8616646640308192e-05,
+      "loss": 0.4359,
+      "step": 1908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8614898120553023e-05,
+      "loss": 0.4263,
+      "step": 1909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.861314854994347e-05,
+      "loss": 0.4606,
+      "step": 1910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.861139792861458e-05,
+      "loss": 0.4527,
+      "step": 1911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8609646256701463e-05,
+      "loss": 0.4228,
+      "step": 1912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8607893534339332e-05,
+      "loss": 0.5343,
+      "step": 1913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8606139761663465e-05,
+      "loss": 0.3719,
+      "step": 1914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8604384938809217e-05,
+      "loss": 0.4148,
+      "step": 1915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8602629065912044e-05,
+      "loss": 0.403,
+      "step": 1916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.860087214310747e-05,
+      "loss": 0.4144,
+      "step": 1917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8599114170531094e-05,
+      "loss": 0.4249,
+      "step": 1918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.859735514831861e-05,
+      "loss": 0.4046,
+      "step": 1919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8595595076605786e-05,
+      "loss": 0.4302,
+      "step": 1920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.859383395552847e-05,
+      "loss": 0.4308,
+      "step": 1921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8592071785222596e-05,
+      "loss": 0.3805,
+      "step": 1922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8590308565824167e-05,
+      "loss": 0.416,
+      "step": 1923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.858854429746929e-05,
+      "loss": 0.3843,
+      "step": 1924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8586778980294125e-05,
+      "loss": 0.4084,
+      "step": 1925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8585012614434933e-05,
+      "loss": 0.528,
+      "step": 1926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8583245200028044e-05,
+      "loss": 0.4159,
+      "step": 1927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8581476737209886e-05,
+      "loss": 0.38,
+      "step": 1928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8579707226116945e-05,
+      "loss": 0.3939,
+      "step": 1929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8577936666885805e-05,
+      "loss": 0.4431,
+      "step": 1930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.857616505965312e-05,
+      "loss": 0.4742,
+      "step": 1931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8574392404555634e-05,
+      "loss": 0.4463,
+      "step": 1932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.857261870173017e-05,
+      "loss": 0.3926,
+      "step": 1933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8570843951313625e-05,
+      "loss": 0.4357,
+      "step": 1934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.856906815344298e-05,
+      "loss": 0.4636,
+      "step": 1935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.856729130825531e-05,
+      "loss": 0.3885,
+      "step": 1936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8565513415887744e-05,
+      "loss": 0.4359,
+      "step": 1937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8563734476477516e-05,
+      "loss": 0.4162,
+      "step": 1938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.856195449016193e-05,
+      "loss": 0.422,
+      "step": 1939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8560173457078373e-05,
+      "loss": 0.37,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8558391377364318e-05,
+      "loss": 0.3757,
+      "step": 1941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8556608251157302e-05,
+      "loss": 0.4185,
+      "step": 1942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.855482407859496e-05,
+      "loss": 0.433,
+      "step": 1943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8553038859815003e-05,
+      "loss": 0.3901,
+      "step": 1944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8551252594955218e-05,
+      "loss": 0.4132,
+      "step": 1945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8549465284153485e-05,
+      "loss": 0.4301,
+      "step": 1946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8547676927547744e-05,
+      "loss": 0.4149,
+      "step": 1947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8545887525276032e-05,
+      "loss": 0.418,
+      "step": 1948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8544097077476472e-05,
+      "loss": 0.4368,
+      "step": 1949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8542305584287243e-05,
+      "loss": 0.4089,
+      "step": 1950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.854051304584663e-05,
+      "loss": 0.3866,
+      "step": 1951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.8538719462292985e-05,
+      "loss": 0.3758,
+      "step": 1952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8536924833764745e-05,
+      "loss": 0.3991,
+      "step": 1953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8535129160400427e-05,
+      "loss": 0.4064,
+      "step": 1954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8533332442338627e-05,
+      "loss": 0.4422,
+      "step": 1955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8531534679718027e-05,
+      "loss": 0.3895,
+      "step": 1956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8529735872677383e-05,
+      "loss": 0.4088,
+      "step": 1957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8527936021355534e-05,
+      "loss": 0.4206,
+      "step": 1958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.85261351258914e-05,
+      "loss": 0.413,
+      "step": 1959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852433318642398e-05,
+      "loss": 0.4192,
+      "step": 1960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.852253020309236e-05,
+      "loss": 0.4028,
+      "step": 1961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.85207261760357e-05,
+      "loss": 0.4211,
+      "step": 1962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8518921105393243e-05,
+      "loss": 0.3526,
+      "step": 1963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.851711499130431e-05,
+      "loss": 0.4498,
+      "step": 1964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.85153078339083e-05,
+      "loss": 0.4823,
+      "step": 1965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8513499633344706e-05,
+      "loss": 0.456,
+      "step": 1966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8511690389753086e-05,
+      "loss": 0.4181,
+      "step": 1967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8509880103273084e-05,
+      "loss": 0.4056,
+      "step": 1968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8508068774044438e-05,
+      "loss": 0.4113,
+      "step": 1969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8506256402206935e-05,
+      "loss": 0.4405,
+      "step": 1970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8504442987900475e-05,
+      "loss": 0.4314,
+      "step": 1971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850262853126502e-05,
+      "loss": 0.3756,
+      "step": 1972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.850081303244062e-05,
+      "loss": 0.3982,
+      "step": 1973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8498996491567394e-05,
+      "loss": 0.414,
+      "step": 1974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8497178908785564e-05,
+      "loss": 0.4078,
+      "step": 1975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8495360284235406e-05,
+      "loss": 0.4299,
+      "step": 1976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8493540618057296e-05,
+      "loss": 0.4144,
+      "step": 1977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8491719910391685e-05,
+      "loss": 0.433,
+      "step": 1978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8489898161379093e-05,
+      "loss": 0.3948,
+      "step": 1979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.848807537116014e-05,
+      "loss": 0.4106,
+      "step": 1980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8486251539875515e-05,
+      "loss": 0.4306,
+      "step": 1981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8484426667665985e-05,
+      "loss": 0.3762,
+      "step": 1982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8482600754672408e-05,
+      "loss": 0.3593,
+      "step": 1983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8480773801035707e-05,
+      "loss": 0.4402,
+      "step": 1984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8478945806896897e-05,
+      "loss": 0.3939,
+      "step": 1985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8477116772397072e-05,
+      "loss": 0.3886,
+      "step": 1986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.84752866976774e-05,
+      "loss": 0.4035,
+      "step": 1987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.847345558287914e-05,
+      "loss": 0.3885,
+      "step": 1988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8471623428143625e-05,
+      "loss": 0.4029,
+      "step": 1989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8469790233612262e-05,
+      "loss": 0.4196,
+      "step": 1990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8467955999426547e-05,
+      "loss": 0.4386,
+      "step": 1991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8466120725728058e-05,
+      "loss": 0.4265,
+      "step": 1992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8464284412658444e-05,
+      "loss": 0.4345,
+      "step": 1993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.846244706035944e-05,
+      "loss": 0.4075,
+      "step": 1994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8460608668972868e-05,
+      "loss": 0.4551,
+      "step": 1995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.845876923864061e-05,
+      "loss": 0.4071,
+      "step": 1996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.845692876950465e-05,
+      "loss": 0.438,
+      "step": 1997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8455087261707037e-05,
+      "loss": 0.4283,
+      "step": 1998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.845324471538991e-05,
+      "loss": 0.4405,
+      "step": 1999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.845140113069549e-05,
+      "loss": 0.424,
+      "step": 2000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8449556507766063e-05,
+      "loss": 0.4183,
+      "step": 2001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8447710846744003e-05,
+      "loss": 0.4281,
+      "step": 2002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8445864147771775e-05,
+      "loss": 0.4137,
+      "step": 2003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.844401641099191e-05,
+      "loss": 0.4486,
+      "step": 2004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8442167636547025e-05,
+      "loss": 0.3882,
+      "step": 2005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.844031782457982e-05,
+      "loss": 0.4639,
+      "step": 2006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.843846697523306e-05,
+      "loss": 0.4077,
+      "step": 2007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8436615088649614e-05,
+      "loss": 0.3585,
+      "step": 2008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8434762164972406e-05,
+      "loss": 0.4228,
+      "step": 2009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.8432908204344464e-05,
+      "loss": 0.4473,
+      "step": 2010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.843105320690888e-05,
+      "loss": 0.4237,
+      "step": 2011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8429197172808827e-05,
+      "loss": 0.4701,
+      "step": 2012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8427340102187567e-05,
+      "loss": 0.4499,
+      "step": 2013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.842548199518843e-05,
+      "loss": 0.4147,
+      "step": 2014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8423622851954832e-05,
+      "loss": 0.3887,
+      "step": 2015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8421762672630285e-05,
+      "loss": 0.4174,
+      "step": 2016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8419901457358345e-05,
+      "loss": 0.4048,
+      "step": 2017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.841803920628268e-05,
+      "loss": 0.382,
+      "step": 2018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8416175919547017e-05,
+      "loss": 0.3848,
+      "step": 2019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.841431159729519e-05,
+      "loss": 0.4744,
+      "step": 2020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8412446239671078e-05,
+      "loss": 0.4092,
+      "step": 2021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.841057984681866e-05,
+      "loss": 0.3994,
+      "step": 2022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8408712418881998e-05,
+      "loss": 0.4364,
+      "step": 2023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8406843956005225e-05,
+      "loss": 0.438,
+      "step": 2024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8404974458332552e-05,
+      "loss": 0.4286,
+      "step": 2025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8403103926008284e-05,
+      "loss": 0.4714,
+      "step": 2026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8401232359176782e-05,
+      "loss": 0.4125,
+      "step": 2027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8399359757982523e-05,
+      "loss": 0.4282,
+      "step": 2028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8397486122570023e-05,
+      "loss": 0.3947,
+      "step": 2029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8395611453083903e-05,
+      "loss": 0.4335,
+      "step": 2030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8393735749668863e-05,
+      "loss": 0.3855,
+      "step": 2031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8391859012469663e-05,
+      "loss": 0.3738,
+      "step": 2032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8389981241631177e-05,
+      "loss": 0.4193,
+      "step": 2033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8388102437298325e-05,
+      "loss": 0.4007,
+      "step": 2034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8386222599616127e-05,
+      "loss": 0.4383,
+      "step": 2035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8384341728729676e-05,
+      "loss": 0.4396,
+      "step": 2036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8382459824784142e-05,
+      "loss": 0.4499,
+      "step": 2037
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8380576887924782e-05,
+      "loss": 0.4171,
+      "step": 2038
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.837869291829693e-05,
+      "loss": 0.4102,
+      "step": 2039
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8376807916045995e-05,
+      "loss": 0.4523,
+      "step": 2040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.837492188131747e-05,
+      "loss": 0.435,
+      "step": 2041
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8373034814256923e-05,
+      "loss": 0.3581,
+      "step": 2042
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8371146715010014e-05,
+      "loss": 0.4334,
+      "step": 2043
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.836925758372247e-05,
+      "loss": 0.3802,
+      "step": 2044
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8367367420540107e-05,
+      "loss": 0.4083,
+      "step": 2045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8365476225608807e-05,
+      "loss": 0.41,
+      "step": 2046
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8363583999074542e-05,
+      "loss": 0.3535,
+      "step": 2047
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.836169074108337e-05,
+      "loss": 0.4712,
+      "step": 2048
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8359796451781412e-05,
+      "loss": 0.3772,
+      "step": 2049
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.835790113131488e-05,
+      "loss": 0.4522,
+      "step": 2050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.835600477983006e-05,
+      "loss": 0.3939,
+      "step": 2051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8354107397473327e-05,
+      "loss": 0.383,
+      "step": 2052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.835220898439112e-05,
+      "loss": 0.4253,
+      "step": 2053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8350309540729977e-05,
+      "loss": 0.4107,
+      "step": 2054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8348409066636498e-05,
+      "loss": 0.4087,
+      "step": 2055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8346507562257366e-05,
+      "loss": 0.3804,
+      "step": 2056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.834460502773936e-05,
+      "loss": 0.4091,
+      "step": 2057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.834270146322931e-05,
+      "loss": 0.4196,
+      "step": 2058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.834079686887415e-05,
+      "loss": 0.425,
+      "step": 2059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.833889124482088e-05,
+      "loss": 0.4207,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8336984591216593e-05,
+      "loss": 0.4028,
+      "step": 2061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.833507690820844e-05,
+      "loss": 0.4078,
+      "step": 2062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8333168195943675e-05,
+      "loss": 0.4111,
+      "step": 2063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8331258454569612e-05,
+      "loss": 0.396,
+      "step": 2064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8329347684233654e-05,
+      "loss": 0.4596,
+      "step": 2065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8327435885083292e-05,
+      "loss": 0.3928,
+      "step": 2066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8325523057266073e-05,
+      "loss": 0.4005,
+      "step": 2067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.832360920092964e-05,
+      "loss": 0.4194,
+      "step": 2068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.8321694316221723e-05,
+      "loss": 0.4152,
+      "step": 2069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8319778403290112e-05,
+      "loss": 0.4085,
+      "step": 2070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831786146228268e-05,
+      "loss": 0.4026,
+      "step": 2071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8315943493347397e-05,
+      "loss": 0.4188,
+      "step": 2072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.831402449663229e-05,
+      "loss": 0.3943,
+      "step": 2073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8312104472285477e-05,
+      "loss": 0.4435,
+      "step": 2074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8310183420455157e-05,
+      "loss": 0.401,
+      "step": 2075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8308261341289603e-05,
+      "loss": 0.4134,
+      "step": 2076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8306338234937166e-05,
+      "loss": 0.3658,
+      "step": 2077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.830441410154629e-05,
+      "loss": 0.4188,
+      "step": 2078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.830248894126547e-05,
+      "loss": 0.4162,
+      "step": 2079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8300562754243314e-05,
+      "loss": 0.41,
+      "step": 2080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.829863554062848e-05,
+      "loss": 0.4657,
+      "step": 2081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.829670730056973e-05,
+      "loss": 0.3676,
+      "step": 2082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8294778034215888e-05,
+      "loss": 0.4304,
+      "step": 2083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.829284774171587e-05,
+      "loss": 0.4375,
+      "step": 2084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8290916423218648e-05,
+      "loss": 0.385,
+      "step": 2085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8288984078873302e-05,
+      "loss": 0.4441,
+      "step": 2086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8287050708828977e-05,
+      "loss": 0.4333,
+      "step": 2087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8285116313234893e-05,
+      "loss": 0.4463,
+      "step": 2088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8283180892240355e-05,
+      "loss": 0.3917,
+      "step": 2089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8281244445994756e-05,
+      "loss": 0.3954,
+      "step": 2090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8279306974647553e-05,
+      "loss": 0.4409,
+      "step": 2091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8277368478348288e-05,
+      "loss": 0.399,
+      "step": 2092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.827542895724658e-05,
+      "loss": 0.444,
+      "step": 2093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8273488411492133e-05,
+      "loss": 0.3622,
+      "step": 2094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8271546841234727e-05,
+      "loss": 0.3766,
+      "step": 2095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8269604246624216e-05,
+      "loss": 0.4241,
+      "step": 2096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826766062781054e-05,
+      "loss": 0.494,
+      "step": 2097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826571598494371e-05,
+      "loss": 0.5445,
+      "step": 2098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8263770318173835e-05,
+      "loss": 0.403,
+      "step": 2099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.826182362765108e-05,
+      "loss": 0.4278,
+      "step": 2100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8259875913525696e-05,
+      "loss": 0.4333,
+      "step": 2101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8257927175948025e-05,
+      "loss": 0.3965,
+      "step": 2102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8255977415068474e-05,
+      "loss": 0.4018,
+      "step": 2103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.825402663103753e-05,
+      "loss": 0.4189,
+      "step": 2104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.825207482400577e-05,
+      "loss": 0.4268,
+      "step": 2105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8250121994123837e-05,
+      "loss": 0.4206,
+      "step": 2106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8248168141542456e-05,
+      "loss": 0.4459,
+      "step": 2107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.824621326641244e-05,
+      "loss": 0.3695,
+      "step": 2108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8244257368884674e-05,
+      "loss": 0.4032,
+      "step": 2109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.824230044911012e-05,
+      "loss": 0.4478,
+      "step": 2110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.824034250723982e-05,
+      "loss": 0.357,
+      "step": 2111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8238383543424894e-05,
+      "loss": 0.3982,
+      "step": 2112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8236423557816555e-05,
+      "loss": 0.4184,
+      "step": 2113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8234462550566067e-05,
+      "loss": 0.4116,
+      "step": 2114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.82325005218248e-05,
+      "loss": 0.3936,
+      "step": 2115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8230537471744186e-05,
+      "loss": 0.4196,
+      "step": 2116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8228573400475743e-05,
+      "loss": 0.3933,
+      "step": 2117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8226608308171064e-05,
+      "loss": 0.3783,
+      "step": 2118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.822464219498183e-05,
+      "loss": 0.4052,
+      "step": 2119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.822267506105978e-05,
+      "loss": 0.4269,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.822070690655676e-05,
+      "loss": 0.4197,
+      "step": 2121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.821873773162468e-05,
+      "loss": 0.4153,
+      "step": 2122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8216767536415517e-05,
+      "loss": 0.3678,
+      "step": 2123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8214796321081346e-05,
+      "loss": 0.4339,
+      "step": 2124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8212824085774315e-05,
+      "loss": 0.4927,
+      "step": 2125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.821085083064665e-05,
+      "loss": 0.4267,
+      "step": 2126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.8208876555850648e-05,
+      "loss": 0.4022,
+      "step": 2127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8206901261538697e-05,
+      "loss": 0.4468,
+      "step": 2128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.820492494786326e-05,
+      "loss": 0.4377,
+      "step": 2129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8202947614976873e-05,
+      "loss": 0.3946,
+      "step": 2130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8200969263032155e-05,
+      "loss": 0.4737,
+      "step": 2131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8198989892181808e-05,
+      "loss": 0.4503,
+      "step": 2132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8197009502578603e-05,
+      "loss": 0.4285,
+      "step": 2133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8195028094375398e-05,
+      "loss": 0.3932,
+      "step": 2134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8193045667725124e-05,
+      "loss": 0.415,
+      "step": 2135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8191062222780793e-05,
+      "loss": 0.3845,
+      "step": 2136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.81890777596955e-05,
+      "loss": 0.4343,
+      "step": 2137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.818709227862241e-05,
+      "loss": 0.4481,
+      "step": 2138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.818510577971477e-05,
+      "loss": 0.4165,
+      "step": 2139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8183118263125907e-05,
+      "loss": 0.4514,
+      "step": 2140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8181129729009226e-05,
+      "loss": 0.4,
+      "step": 2141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8179140177518216e-05,
+      "loss": 0.3793,
+      "step": 2142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8177149608806424e-05,
+      "loss": 0.4154,
+      "step": 2143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8175158023027505e-05,
+      "loss": 0.4609,
+      "step": 2144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.817316542033517e-05,
+      "loss": 0.4043,
+      "step": 2145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8171171800883224e-05,
+      "loss": 0.4156,
+      "step": 2146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.816917716482554e-05,
+      "loss": 0.406,
+      "step": 2147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.816718151231606e-05,
+      "loss": 0.3933,
+      "step": 2148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8165184843508835e-05,
+      "loss": 0.3975,
+      "step": 2149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8163187158557964e-05,
+      "loss": 0.3837,
+      "step": 2150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8161188457617645e-05,
+      "loss": 0.4515,
+      "step": 2151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8159188740842138e-05,
+      "loss": 0.3754,
+      "step": 2152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8157188008385795e-05,
+      "loss": 0.3843,
+      "step": 2153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.815518626040304e-05,
+      "loss": 0.4295,
+      "step": 2154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8153183497048373e-05,
+      "loss": 0.4101,
+      "step": 2155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.815117971847638e-05,
+      "loss": 0.3781,
+      "step": 2156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8149174924841716e-05,
+      "loss": 0.3694,
+      "step": 2157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8147169116299124e-05,
+      "loss": 0.4201,
+      "step": 2158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8145162293003413e-05,
+      "loss": 0.402,
+      "step": 2159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8143154455109494e-05,
+      "loss": 0.4426,
+      "step": 2160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8141145602772325e-05,
+      "loss": 0.4023,
+      "step": 2161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.813913573614696e-05,
+      "loss": 0.4908,
+      "step": 2162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8137124855388535e-05,
+      "loss": 0.4121,
+      "step": 2163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8135112960652253e-05,
+      "loss": 0.4209,
+      "step": 2164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8133100052093404e-05,
+      "loss": 0.4395,
+      "step": 2165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.813108612986735e-05,
+      "loss": 0.4147,
+      "step": 2166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8129071194129532e-05,
+      "loss": 0.3887,
+      "step": 2167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8127055245035472e-05,
+      "loss": 0.3608,
+      "step": 2168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8125038282740777e-05,
+      "loss": 0.4154,
+      "step": 2169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8123020307401114e-05,
+      "loss": 0.3774,
+      "step": 2170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8121001319172247e-05,
+      "loss": 0.4219,
+      "step": 2171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.811898131821e-05,
+      "loss": 0.4389,
+      "step": 2172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8116960304670296e-05,
+      "loss": 0.3858,
+      "step": 2173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8114938278709118e-05,
+      "loss": 0.4178,
+      "step": 2174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8112915240482538e-05,
+      "loss": 0.4166,
+      "step": 2175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.81108911901467e-05,
+      "loss": 0.4372,
+      "step": 2176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.810886612785783e-05,
+      "loss": 0.4376,
+      "step": 2177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.810684005377223e-05,
+      "loss": 0.4161,
+      "step": 2178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8104812968046283e-05,
+      "loss": 0.4204,
+      "step": 2179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8102784870836448e-05,
+      "loss": 0.3697,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.810075576229925e-05,
+      "loss": 0.4668,
+      "step": 2181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8098725642591324e-05,
+      "loss": 0.3963,
+      "step": 2182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.809669451186935e-05,
+      "loss": 0.385,
+      "step": 2183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8094662370290106e-05,
+      "loss": 0.4802,
+      "step": 2184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8092629218010433e-05,
+      "loss": 0.4114,
+      "step": 2185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.8090595055187264e-05,
+      "loss": 0.4023,
+      "step": 2186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8088559881977605e-05,
+      "loss": 0.3561,
+      "step": 2187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8086523698538533e-05,
+      "loss": 0.4096,
+      "step": 2188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8084486505027217e-05,
+      "loss": 0.4722,
+      "step": 2189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8082448301600886e-05,
+      "loss": 0.4164,
+      "step": 2190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8080409088416868e-05,
+      "loss": 0.4577,
+      "step": 2191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8078368865632552e-05,
+      "loss": 0.4283,
+      "step": 2192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8076327633405412e-05,
+      "loss": 0.3615,
+      "step": 2193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8074285391893e-05,
+      "loss": 0.3822,
+      "step": 2194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.807224214125294e-05,
+      "loss": 0.4274,
+      "step": 2195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8070197881642943e-05,
+      "loss": 0.4356,
+      "step": 2196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8068152613220794e-05,
+      "loss": 0.4622,
+      "step": 2197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.806610633614435e-05,
+      "loss": 0.358,
+      "step": 2198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.806405905057156e-05,
+      "loss": 0.4538,
+      "step": 2199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.806201075666043e-05,
+      "loss": 0.4177,
+      "step": 2200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8059961454569066e-05,
+      "loss": 0.3874,
+      "step": 2201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.805791114445564e-05,
+      "loss": 0.4438,
+      "step": 2202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8055859826478403e-05,
+      "loss": 0.4034,
+      "step": 2203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.805380750079568e-05,
+      "loss": 0.3786,
+      "step": 2204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.805175416756588e-05,
+      "loss": 0.3671,
+      "step": 2205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.804969982694749e-05,
+      "loss": 0.4232,
+      "step": 2206
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.804764447909907e-05,
+      "loss": 0.4066,
+      "step": 2207
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8045588124179264e-05,
+      "loss": 0.4437,
+      "step": 2208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.804353076234679e-05,
+      "loss": 0.4415,
+      "step": 2209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8041472393760432e-05,
+      "loss": 0.3895,
+      "step": 2210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8039413018579082e-05,
+      "loss": 0.3773,
+      "step": 2211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8037352636961677e-05,
+      "loss": 0.3968,
+      "step": 2212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8035291249067254e-05,
+      "loss": 0.4793,
+      "step": 2213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8033228855054913e-05,
+      "loss": 0.4381,
+      "step": 2214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8031165455083844e-05,
+      "loss": 0.4249,
+      "step": 2215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8029101049313305e-05,
+      "loss": 0.4258,
+      "step": 2216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8027035637902637e-05,
+      "loss": 0.3275,
+      "step": 2217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8024969221011254e-05,
+      "loss": 0.4494,
+      "step": 2218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8022901798798654e-05,
+      "loss": 0.4322,
+      "step": 2219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8020833371424413e-05,
+      "loss": 0.4089,
+      "step": 2220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.801876393904817e-05,
+      "loss": 0.3781,
+      "step": 2221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8016693501829664e-05,
+      "loss": 0.3798,
+      "step": 2222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.801462205992869e-05,
+      "loss": 0.4407,
+      "step": 2223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8012549613505138e-05,
+      "loss": 0.4195,
+      "step": 2224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8010476162718963e-05,
+      "loss": 0.4141,
+      "step": 2225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8008401707730207e-05,
+      "loss": 0.4008,
+      "step": 2226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.800632624869898e-05,
+      "loss": 0.3902,
+      "step": 2227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8004249785785475e-05,
+      "loss": 0.4589,
+      "step": 2228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.800217231914997e-05,
+      "loss": 0.4358,
+      "step": 2229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.8000093848952802e-05,
+      "loss": 0.3943,
+      "step": 2230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7998014375354407e-05,
+      "loss": 0.3891,
+      "step": 2231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7995933898515275e-05,
+      "loss": 0.3657,
+      "step": 2232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7993852418595997e-05,
+      "loss": 0.3869,
+      "step": 2233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7991769935757223e-05,
+      "loss": 0.4404,
+      "step": 2234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7989686450159687e-05,
+      "loss": 0.4055,
+      "step": 2235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7987601961964207e-05,
+      "loss": 0.4002,
+      "step": 2236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.798551647133167e-05,
+      "loss": 0.3986,
+      "step": 2237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.798342997842304e-05,
+      "loss": 0.3833,
+      "step": 2238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7981342483399363e-05,
+      "loss": 0.4267,
+      "step": 2239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7979253986421763e-05,
+      "loss": 0.4337,
+      "step": 2240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7977164487651435e-05,
+      "loss": 0.3376,
+      "step": 2241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.797507398724966e-05,
+      "loss": 0.4524,
+      "step": 2242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7972982485377786e-05,
+      "loss": 0.4273,
+      "step": 2243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7970889982197247e-05,
+      "loss": 0.4192,
+      "step": 2244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.796879647786955e-05,
+      "loss": 0.4313,
+      "step": 2245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.796670197255628e-05,
+      "loss": 0.364,
+      "step": 2246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7964606466419102e-05,
+      "loss": 0.4807,
+      "step": 2247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7962509959619757e-05,
+      "loss": 0.3969,
+      "step": 2248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7960412452320056e-05,
+      "loss": 0.4176,
+      "step": 2249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7958313944681896e-05,
+      "loss": 0.4371,
+      "step": 2250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7956214436867248e-05,
+      "loss": 0.4213,
+      "step": 2251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7954113929038164e-05,
+      "loss": 0.4328,
+      "step": 2252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.795201242135677e-05,
+      "loss": 0.4186,
+      "step": 2253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7949909913985267e-05,
+      "loss": 0.385,
+      "step": 2254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7947806407085932e-05,
+      "loss": 0.3894,
+      "step": 2255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7945701900821127e-05,
+      "loss": 0.4201,
+      "step": 2256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7943596395353282e-05,
+      "loss": 0.4084,
+      "step": 2257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7941489890844917e-05,
+      "loss": 0.4321,
+      "step": 2258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7939382387458613e-05,
+      "loss": 0.3256,
+      "step": 2259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7937273885357038e-05,
+      "loss": 0.3987,
+      "step": 2260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.793516438470294e-05,
+      "loss": 0.4163,
+      "step": 2261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7933053885659127e-05,
+      "loss": 0.4244,
+      "step": 2262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.793094238838851e-05,
+      "loss": 0.412,
+      "step": 2263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792882989305405e-05,
+      "loss": 0.4343,
+      "step": 2264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792671639981881e-05,
+      "loss": 0.417,
+      "step": 2265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7924601908845908e-05,
+      "loss": 0.4068,
+      "step": 2266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.792248642029856e-05,
+      "loss": 0.4077,
+      "step": 2267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7920369934340038e-05,
+      "loss": 0.4189,
+      "step": 2268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7918252451133707e-05,
+      "loss": 0.4199,
+      "step": 2269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7916133970843003e-05,
+      "loss": 0.4299,
+      "step": 2270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7914014493631437e-05,
+      "loss": 0.4024,
+      "step": 2271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7911894019662602e-05,
+      "loss": 0.3786,
+      "step": 2272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.790977254910016e-05,
+      "loss": 0.3884,
+      "step": 2273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7907650082107857e-05,
+      "loss": 0.4109,
+      "step": 2274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7905526618849515e-05,
+      "loss": 0.3966,
+      "step": 2275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7903402159489036e-05,
+      "loss": 0.3984,
+      "step": 2276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7901276704190386e-05,
+      "loss": 0.3803,
+      "step": 2277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.789915025311762e-05,
+      "loss": 0.4171,
+      "step": 2278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7897022806434866e-05,
+      "loss": 0.4236,
+      "step": 2279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7894894364306326e-05,
+      "loss": 0.4159,
+      "step": 2280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7892764926896292e-05,
+      "loss": 0.402,
+      "step": 2281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.789063449436912e-05,
+      "loss": 0.3645,
+      "step": 2282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7888503066889234e-05,
+      "loss": 0.3894,
+      "step": 2283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.788637064462116e-05,
+      "loss": 0.3945,
+      "step": 2284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.788423722772948e-05,
+      "loss": 0.4529,
+      "step": 2285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.788210281637886e-05,
+      "loss": 0.4465,
+      "step": 2286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.787996741073405e-05,
+      "loss": 0.4336,
+      "step": 2287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.787783101095986e-05,
+      "loss": 0.422,
+      "step": 2288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7875693617221188e-05,
+      "loss": 0.3873,
+      "step": 2289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7873555229683013e-05,
+      "loss": 0.4198,
+      "step": 2290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7871415848510383e-05,
+      "loss": 0.404,
+      "step": 2291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7869275473868414e-05,
+      "loss": 0.472,
+      "step": 2292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7867134105922323e-05,
+      "loss": 0.4596,
+      "step": 2293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7864991744837382e-05,
+      "loss": 0.4436,
+      "step": 2294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7862848390778953e-05,
+      "loss": 0.4225,
+      "step": 2295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.786070404391246e-05,
+      "loss": 0.4284,
+      "step": 2296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7858558704403424e-05,
+      "loss": 0.4084,
+      "step": 2297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.785641237241742e-05,
+      "loss": 0.4347,
+      "step": 2298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7854265048120114e-05,
+      "loss": 0.4188,
+      "step": 2299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7852116731677256e-05,
+      "loss": 0.4261,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7849967423254645e-05,
+      "loss": 0.3888,
+      "step": 2301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7847817123018187e-05,
+      "loss": 0.4188,
+      "step": 2302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7845665831133838e-05,
+      "loss": 0.4288,
+      "step": 2303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7843513547767656e-05,
+      "loss": 0.3805,
+      "step": 2304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.784136027308576e-05,
+      "loss": 0.4123,
+      "step": 2305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.783920600725435e-05,
+      "loss": 0.3765,
+      "step": 2306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.783705075043969e-05,
+      "loss": 0.3935,
+      "step": 2307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7834894502808146e-05,
+      "loss": 0.4486,
+      "step": 2308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7832737264526143e-05,
+      "loss": 0.4087,
+      "step": 2309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7830579035760176e-05,
+      "loss": 0.4124,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7828419816676836e-05,
+      "loss": 0.4174,
+      "step": 2311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7826259607442776e-05,
+      "loss": 0.3974,
+      "step": 2312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7824098408224735e-05,
+      "loss": 0.4276,
+      "step": 2313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7821936219189518e-05,
+      "loss": 0.4035,
+      "step": 2314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7819773040504015e-05,
+      "loss": 0.4152,
+      "step": 2315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7817608872335185e-05,
+      "loss": 0.3973,
+      "step": 2316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.781544371485007e-05,
+      "loss": 0.3825,
+      "step": 2317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7813277568215787e-05,
+      "loss": 0.3935,
+      "step": 2318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7811110432599526e-05,
+      "loss": 0.4337,
+      "step": 2319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7808942308168556e-05,
+      "loss": 0.4798,
+      "step": 2320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.780677319509023e-05,
+      "loss": 0.449,
+      "step": 2321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.780460309353195e-05,
+      "loss": 0.4196,
+      "step": 2322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7802432003661233e-05,
+      "loss": 0.3725,
+      "step": 2323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7800259925645645e-05,
+      "loss": 0.3757,
+      "step": 2324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7798086859652833e-05,
+      "loss": 0.3979,
+      "step": 2325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7795912805850523e-05,
+      "loss": 0.4254,
+      "step": 2326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7793737764406522e-05,
+      "loss": 0.4556,
+      "step": 2327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7791561735488717e-05,
+      "loss": 0.4164,
+      "step": 2328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7789384719265042e-05,
+      "loss": 0.3794,
+      "step": 2329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7787206715903543e-05,
+      "loss": 0.4242,
+      "step": 2330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7785027725572322e-05,
+      "loss": 0.4108,
+      "step": 2331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7782847748439565e-05,
+      "loss": 0.382,
+      "step": 2332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7780666784673535e-05,
+      "loss": 0.3855,
+      "step": 2333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.777848483444256e-05,
+      "loss": 0.4148,
+      "step": 2334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.777630189791506e-05,
+      "loss": 0.4122,
+      "step": 2335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7774117975259514e-05,
+      "loss": 0.3993,
+      "step": 2336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7771933066644492e-05,
+      "loss": 0.3928,
+      "step": 2337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7769747172238632e-05,
+      "loss": 0.3596,
+      "step": 2338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7767560292210655e-05,
+      "loss": 0.4573,
+      "step": 2339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.776537242672935e-05,
+      "loss": 0.4092,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7763183575963585e-05,
+      "loss": 0.4441,
+      "step": 2341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7760993740082303e-05,
+      "loss": 0.4077,
+      "step": 2342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7758802919254532e-05,
+      "loss": 0.4172,
+      "step": 2343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.775661111364936e-05,
+      "loss": 0.4314,
+      "step": 2344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7754418323435966e-05,
+      "loss": 0.3799,
+      "step": 2345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7752224548783594e-05,
+      "loss": 0.4512,
+      "step": 2346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7750029789861574e-05,
+      "loss": 0.3813,
+      "step": 2347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.77478340468393e-05,
+      "loss": 0.4202,
+      "step": 2348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7745637319886253e-05,
+      "loss": 0.4405,
+      "step": 2349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7743439609171983e-05,
+      "loss": 0.3818,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.774124091486612e-05,
+      "loss": 0.3541,
+      "step": 2351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773904123713837e-05,
+      "loss": 0.3786,
+      "step": 2352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773684057615851e-05,
+      "loss": 0.3854,
+      "step": 2353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.77346389320964e-05,
+      "loss": 0.4104,
+      "step": 2354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.773243630512197e-05,
+      "loss": 0.3938,
+      "step": 2355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7730232695405222e-05,
+      "loss": 0.4081,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.7728028103116252e-05,
+      "loss": 0.4045,
+      "step": 2357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.772582252842521e-05,
+      "loss": 0.3867,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.772361597150233e-05,
+      "loss": 0.416,
+      "step": 2359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.772140843251793e-05,
+      "loss": 0.4659,
+      "step": 2360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.77191999116424e-05,
+      "loss": 0.3499,
+      "step": 2361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7716990409046195e-05,
+      "loss": 0.3881,
+      "step": 2362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.771477992489985e-05,
+      "loss": 0.4214,
+      "step": 2363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7712568459373993e-05,
+      "loss": 0.413,
+      "step": 2364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.77103560126393e-05,
+      "loss": 0.3915,
+      "step": 2365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7708142584866548e-05,
+      "loss": 0.4057,
+      "step": 2366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7705928176226572e-05,
+      "loss": 0.3484,
+      "step": 2367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7703712786890288e-05,
+      "loss": 0.4565,
+      "step": 2368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7701496417028694e-05,
+      "loss": 0.4177,
+      "step": 2369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7699279066812858e-05,
+      "loss": 0.4396,
+      "step": 2370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7697060736413924e-05,
+      "loss": 0.4569,
+      "step": 2371
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7694841426003108e-05,
+      "loss": 0.4042,
+      "step": 2372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7692621135751708e-05,
+      "loss": 0.396,
+      "step": 2373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7690399865831098e-05,
+      "loss": 0.4142,
+      "step": 2374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.768817761641272e-05,
+      "loss": 0.3996,
+      "step": 2375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7685954387668104e-05,
+      "loss": 0.3886,
+      "step": 2376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7683730179768837e-05,
+      "loss": 0.3951,
+      "step": 2377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7681504992886603e-05,
+      "loss": 0.378,
+      "step": 2378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7679278827193143e-05,
+      "loss": 0.46,
+      "step": 2379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.767705168286029e-05,
+      "loss": 0.454,
+      "step": 2380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7674823560059934e-05,
+      "loss": 0.4103,
+      "step": 2381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7672594458964057e-05,
+      "loss": 0.4055,
+      "step": 2382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7670364379744715e-05,
+      "loss": 0.4293,
+      "step": 2383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7668133322574026e-05,
+      "loss": 0.4503,
+      "step": 2384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7665901287624198e-05,
+      "loss": 0.4423,
+      "step": 2385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7663668275067502e-05,
+      "loss": 0.3481,
+      "step": 2386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.76614342850763e-05,
+      "loss": 0.4159,
+      "step": 2387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7659199317823016e-05,
+      "loss": 0.4457,
+      "step": 2388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7656963373480152e-05,
+      "loss": 0.3692,
+      "step": 2389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.765472645222029e-05,
+      "loss": 0.3795,
+      "step": 2390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7652488554216087e-05,
+      "loss": 0.4338,
+      "step": 2391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.765024967964027e-05,
+      "loss": 0.408,
+      "step": 2392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7648009828665646e-05,
+      "loss": 0.3934,
+      "step": 2393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7645769001465093e-05,
+      "loss": 0.3876,
+      "step": 2394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7643527198211572e-05,
+      "loss": 0.5092,
+      "step": 2395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7641284419078115e-05,
+      "loss": 0.4106,
+      "step": 2396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7639040664237823e-05,
+      "loss": 0.3632,
+      "step": 2397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7636795933863886e-05,
+      "loss": 0.3485,
+      "step": 2398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7634550228129555e-05,
+      "loss": 0.4131,
+      "step": 2399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7632303547208164e-05,
+      "loss": 0.3643,
+      "step": 2400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.763005589127312e-05,
+      "loss": 0.4115,
+      "step": 2401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7627807260497918e-05,
+      "loss": 0.4126,
+      "step": 2402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.76255576550561e-05,
+      "loss": 0.4244,
+      "step": 2403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.762330707512131e-05,
+      "loss": 0.3863,
+      "step": 2404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7621055520867253e-05,
+      "loss": 0.3996,
+      "step": 2405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7618802992467718e-05,
+      "loss": 0.366,
+      "step": 2406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.761654949009656e-05,
+      "loss": 0.39,
+      "step": 2407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7614295013927716e-05,
+      "loss": 0.3445,
+      "step": 2408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7612039564135194e-05,
+      "loss": 0.4116,
+      "step": 2409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7609783140893083e-05,
+      "loss": 0.4131,
+      "step": 2410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.760752574437554e-05,
+      "loss": 0.4634,
+      "step": 2411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7605267374756793e-05,
+      "loss": 0.4539,
+      "step": 2412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7603008032211167e-05,
+      "loss": 0.3931,
+      "step": 2413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7600747716913037e-05,
+      "loss": 0.4307,
+      "step": 2414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.759848642903687e-05,
+      "loss": 0.4144,
+      "step": 2415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7596224168757197e-05,
+      "loss": 0.413,
+      "step": 2416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7593960936248625e-05,
+      "loss": 0.4532,
+      "step": 2417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.759169673168585e-05,
+      "loss": 0.4981,
+      "step": 2418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.7589431555243628e-05,
+      "loss": 0.4194,
+      "step": 2419
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7587165407096794e-05,
+      "loss": 0.4641,
+      "step": 2420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7584898287420253e-05,
+      "loss": 0.436,
+      "step": 2421
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7582630196389e-05,
+      "loss": 0.3799,
+      "step": 2422
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.758036113417809e-05,
+      "loss": 0.4163,
+      "step": 2423
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7578091100962666e-05,
+      "loss": 0.4301,
+      "step": 2424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.757582009691793e-05,
+      "loss": 0.4118,
+      "step": 2425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.757354812221917e-05,
+      "loss": 0.4087,
+      "step": 2426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7571275177041748e-05,
+      "loss": 0.4305,
+      "step": 2427
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.75690012615611e-05,
+      "loss": 0.4687,
+      "step": 2428
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7566726375952737e-05,
+      "loss": 0.446,
+      "step": 2429
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7564450520392238e-05,
+      "loss": 0.3961,
+      "step": 2430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7562173695055268e-05,
+      "loss": 0.3789,
+      "step": 2431
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7559895900117562e-05,
+      "loss": 0.4195,
+      "step": 2432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7557617135754928e-05,
+      "loss": 0.3924,
+      "step": 2433
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.755533740214325e-05,
+      "loss": 0.4317,
+      "step": 2434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7553056699458488e-05,
+      "loss": 0.4125,
+      "step": 2435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.755077502787668e-05,
+      "loss": 0.3871,
+      "step": 2436
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.754849238757393e-05,
+      "loss": 0.4122,
+      "step": 2437
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7546208778726417e-05,
+      "loss": 0.3857,
+      "step": 2438
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7543924201510413e-05,
+      "loss": 0.3755,
+      "step": 2439
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7541638656102242e-05,
+      "loss": 0.3958,
+      "step": 2440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7539352142678314e-05,
+      "loss": 0.401,
+      "step": 2441
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.753706466141511e-05,
+      "loss": 0.4032,
+      "step": 2442
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7534776212489186e-05,
+      "loss": 0.4257,
+      "step": 2443
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7532486796077176e-05,
+      "loss": 0.4507,
+      "step": 2444
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7530196412355793e-05,
+      "loss": 0.4128,
+      "step": 2445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7527905061501807e-05,
+      "loss": 0.4369,
+      "step": 2446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.752561274369208e-05,
+      "loss": 0.3956,
+      "step": 2447
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7523319459103543e-05,
+      "loss": 0.4304,
+      "step": 2448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7521025207913203e-05,
+      "loss": 0.3926,
+      "step": 2449
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7518729990298132e-05,
+      "loss": 0.408,
+      "step": 2450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7516433806435495e-05,
+      "loss": 0.4074,
+      "step": 2451
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7514136656502508e-05,
+      "loss": 0.3943,
+      "step": 2452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7511838540676486e-05,
+      "loss": 0.368,
+      "step": 2453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7509539459134803e-05,
+      "loss": 0.3896,
+      "step": 2454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7507239412054915e-05,
+      "loss": 0.3994,
+      "step": 2455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7504938399614344e-05,
+      "loss": 0.428,
+      "step": 2456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.750263642199069e-05,
+      "loss": 0.4125,
+      "step": 2457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7500333479361636e-05,
+      "loss": 0.4056,
+      "step": 2458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.749802957190493e-05,
+      "loss": 0.3565,
+      "step": 2459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7495724699798395e-05,
+      "loss": 0.4482,
+      "step": 2460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7493418863219933e-05,
+      "loss": 0.4298,
+      "step": 2461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7491112062347518e-05,
+      "loss": 0.4645,
+      "step": 2462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7488804297359198e-05,
+      "loss": 0.442,
+      "step": 2463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7486495568433096e-05,
+      "loss": 0.3857,
+      "step": 2464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7484185875747404e-05,
+      "loss": 0.3808,
+      "step": 2465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7481875219480406e-05,
+      "loss": 0.463,
+      "step": 2466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7479563599810436e-05,
+      "loss": 0.4018,
+      "step": 2467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7477251016915924e-05,
+      "loss": 0.3549,
+      "step": 2468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7474937470975355e-05,
+      "loss": 0.3656,
+      "step": 2469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7472622962167305e-05,
+      "loss": 0.4186,
+      "step": 2470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.747030749067041e-05,
+      "loss": 0.413,
+      "step": 2471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.74679910566634e-05,
+      "loss": 0.3959,
+      "step": 2472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7465673660325056e-05,
+      "loss": 0.4305,
+      "step": 2473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7463355301834254e-05,
+      "loss": 0.4006,
+      "step": 2474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7461035981369926e-05,
+      "loss": 0.3856,
+      "step": 2475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7458715699111087e-05,
+      "loss": 0.48,
+      "step": 2476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.7456394455236828e-05,
+      "loss": 0.4167,
+      "step": 2477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7454072249926316e-05,
+      "loss": 0.4095,
+      "step": 2478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7451749083358783e-05,
+      "loss": 0.4148,
+      "step": 2479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7449424955713547e-05,
+      "loss": 0.4025,
+      "step": 2480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7447099867169987e-05,
+      "loss": 0.4185,
+      "step": 2481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7444773817907568e-05,
+      "loss": 0.4167,
+      "step": 2482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7442446808105816e-05,
+      "loss": 0.4655,
+      "step": 2483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7440118837944356e-05,
+      "loss": 0.5266,
+      "step": 2484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.743778990760285e-05,
+      "loss": 0.3708,
+      "step": 2485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7435460017261067e-05,
+      "loss": 0.4281,
+      "step": 2486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7433129167098834e-05,
+      "loss": 0.437,
+      "step": 2487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.743079735729606e-05,
+      "loss": 0.3761,
+      "step": 2488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7428464588032717e-05,
+      "loss": 0.418,
+      "step": 2489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7426130859488862e-05,
+      "loss": 0.4346,
+      "step": 2490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.742379617184462e-05,
+      "loss": 0.4521,
+      "step": 2491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7421460525280197e-05,
+      "loss": 0.4168,
+      "step": 2492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.741912391997586e-05,
+      "loss": 0.363,
+      "step": 2493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7416786356111956e-05,
+      "loss": 0.3905,
+      "step": 2494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7414447833868922e-05,
+      "loss": 0.4191,
+      "step": 2495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.741210835342724e-05,
+      "loss": 0.4086,
+      "step": 2496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7409767914967493e-05,
+      "loss": 0.408,
+      "step": 2497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7407426518670318e-05,
+      "loss": 0.3957,
+      "step": 2498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7405084164716435e-05,
+      "loss": 0.4038,
+      "step": 2499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.740274085328664e-05,
+      "loss": 0.3814,
+      "step": 2500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7400396584561796e-05,
+      "loss": 0.4141,
+      "step": 2501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.739805135872284e-05,
+      "loss": 0.4212,
+      "step": 2502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7395705175950793e-05,
+      "loss": 0.4046,
+      "step": 2503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.739335803642674e-05,
+      "loss": 0.3891,
+      "step": 2504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7391009940331845e-05,
+      "loss": 0.4627,
+      "step": 2505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7388660887847343e-05,
+      "loss": 0.4237,
+      "step": 2506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.738631087915454e-05,
+      "loss": 0.3633,
+      "step": 2507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7383959914434823e-05,
+      "loss": 0.4217,
+      "step": 2508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7381607993869654e-05,
+      "loss": 0.4309,
+      "step": 2509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7379255117640548e-05,
+      "loss": 0.478,
+      "step": 2510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.737690128592913e-05,
+      "loss": 0.4416,
+      "step": 2511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7374546498917062e-05,
+      "loss": 0.4334,
+      "step": 2512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.737219075678611e-05,
+      "loss": 0.4277,
+      "step": 2513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.736983405971809e-05,
+      "loss": 0.4212,
+      "step": 2514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7367476407894905e-05,
+      "loss": 0.4126,
+      "step": 2515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7365117801498523e-05,
+      "loss": 0.4,
+      "step": 2516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7362758240711002e-05,
+      "loss": 0.4823,
+      "step": 2517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7360397725714453e-05,
+      "loss": 0.3595,
+      "step": 2518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7358036256691078e-05,
+      "loss": 0.4075,
+      "step": 2519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.735567383382314e-05,
+      "loss": 0.3358,
+      "step": 2520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7353310457292978e-05,
+      "loss": 0.4403,
+      "step": 2521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.735094612728302e-05,
+      "loss": 0.4177,
+      "step": 2522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7348580843975737e-05,
+      "loss": 0.4195,
+      "step": 2523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.73462146075537e-05,
+      "loss": 0.4272,
+      "step": 2524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7343847418199552e-05,
+      "loss": 0.3727,
+      "step": 2525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.734147927609599e-05,
+      "loss": 0.4124,
+      "step": 2526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.733911018142581e-05,
+      "loss": 0.3728,
+      "step": 2527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.733674013437185e-05,
+      "loss": 0.4543,
+      "step": 2528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.733436913511706e-05,
+      "loss": 0.4169,
+      "step": 2529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7331997183844432e-05,
+      "loss": 0.4474,
+      "step": 2530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7329624280737048e-05,
+      "loss": 0.3949,
+      "step": 2531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7327250425978055e-05,
+      "loss": 0.3707,
+      "step": 2532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7324875619750675e-05,
+      "loss": 0.4373,
+      "step": 2533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7322499862238214e-05,
+      "loss": 0.3883,
+      "step": 2534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.7320123153624034e-05,
+      "loss": 0.4134,
+      "step": 2535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7317745494091585e-05,
+      "loss": 0.4909,
+      "step": 2536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7315366883824382e-05,
+      "loss": 0.4156,
+      "step": 2537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7312987323006015e-05,
+      "loss": 0.4209,
+      "step": 2538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7310606811820153e-05,
+      "loss": 0.3824,
+      "step": 2539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.730822535045053e-05,
+      "loss": 0.37,
+      "step": 2540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7305842939080953e-05,
+      "loss": 0.4006,
+      "step": 2541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7303459577895318e-05,
+      "loss": 0.3607,
+      "step": 2542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7301075267077576e-05,
+      "loss": 0.4061,
+      "step": 2543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7298690006811754e-05,
+      "loss": 0.4681,
+      "step": 2544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.729630379728196e-05,
+      "loss": 0.3415,
+      "step": 2545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7293916638672373e-05,
+      "loss": 0.3587,
+      "step": 2546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.729152853116724e-05,
+      "loss": 0.4373,
+      "step": 2547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7289139474950895e-05,
+      "loss": 0.4803,
+      "step": 2548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7286749470207723e-05,
+      "loss": 0.3892,
+      "step": 2549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7284358517122197e-05,
+      "loss": 0.392,
+      "step": 2550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7281966615878867e-05,
+      "loss": 0.4123,
+      "step": 2551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7279573766662343e-05,
+      "loss": 0.4396,
+      "step": 2552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7277179969657317e-05,
+      "loss": 0.4102,
+      "step": 2553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7274785225048557e-05,
+      "loss": 0.4141,
+      "step": 2554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7272389533020893e-05,
+      "loss": 0.415,
+      "step": 2555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7269992893759238e-05,
+      "loss": 0.4156,
+      "step": 2556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7267595307448578e-05,
+      "loss": 0.3711,
+      "step": 2557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7265196774273956e-05,
+      "loss": 0.4079,
+      "step": 2558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7262797294420515e-05,
+      "loss": 0.4341,
+      "step": 2559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7260396868073452e-05,
+      "loss": 0.4367,
+      "step": 2560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7257995495418036e-05,
+      "loss": 0.4056,
+      "step": 2561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7255593176639622e-05,
+      "loss": 0.4283,
+      "step": 2562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7253189911923628e-05,
+      "loss": 0.382,
+      "step": 2563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.725078570145555e-05,
+      "loss": 0.4228,
+      "step": 2564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7248380545420947e-05,
+      "loss": 0.4122,
+      "step": 2565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.724597444400547e-05,
+      "loss": 0.4082,
+      "step": 2566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7243567397394825e-05,
+      "loss": 0.415,
+      "step": 2567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7241159405774803e-05,
+      "loss": 0.3576,
+      "step": 2568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7238750469331263e-05,
+      "loss": 0.3711,
+      "step": 2569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.723634058825013e-05,
+      "loss": 0.422,
+      "step": 2570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7233929762717414e-05,
+      "loss": 0.4133,
+      "step": 2571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.723151799291919e-05,
+      "loss": 0.4101,
+      "step": 2572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7229105279041605e-05,
+      "loss": 0.344,
+      "step": 2573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7226691621270888e-05,
+      "loss": 0.4299,
+      "step": 2574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7224277019793337e-05,
+      "loss": 0.4077,
+      "step": 2575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7221861474795314e-05,
+      "loss": 0.398,
+      "step": 2576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.721944498646327e-05,
+      "loss": 0.3384,
+      "step": 2577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.721702755498371e-05,
+      "loss": 0.4022,
+      "step": 2578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7214609180543223e-05,
+      "loss": 0.4558,
+      "step": 2579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7212189863328475e-05,
+      "loss": 0.391,
+      "step": 2580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7209769603526195e-05,
+      "loss": 0.3936,
+      "step": 2581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7207348401323188e-05,
+      "loss": 0.4197,
+      "step": 2582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.720492625690634e-05,
+      "loss": 0.3945,
+      "step": 2583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7202503170462586e-05,
+      "loss": 0.3776,
+      "step": 2584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7200079142178963e-05,
+      "loss": 0.444,
+      "step": 2585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7197654172242564e-05,
+      "loss": 0.3525,
+      "step": 2586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.719522826084056e-05,
+      "loss": 0.4307,
+      "step": 2587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.719280140816019e-05,
+      "loss": 0.4144,
+      "step": 2588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7190373614388775e-05,
+      "loss": 0.4441,
+      "step": 2589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7187944879713692e-05,
+      "loss": 0.4026,
+      "step": 2590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.718551520432241e-05,
+      "loss": 0.4017,
+      "step": 2591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7183084588402455e-05,
+      "loss": 0.3897,
+      "step": 2592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7180653032141435e-05,
+      "loss": 0.3815,
+      "step": 2593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7178220535727026e-05,
+      "loss": 0.3355,
+      "step": 2594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.717578709934698e-05,
+      "loss": 0.3692,
+      "step": 2595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7173352723189116e-05,
+      "loss": 0.4147,
+      "step": 2596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7170917407441334e-05,
+      "loss": 0.4429,
+      "step": 2597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.71684811522916e-05,
+      "loss": 0.4088,
+      "step": 2598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.716604395792796e-05,
+      "loss": 0.3913,
+      "step": 2599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7163605824538516e-05,
+      "loss": 0.4271,
+      "step": 2600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.716116675231146e-05,
+      "loss": 0.4403,
+      "step": 2601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7158726741435046e-05,
+      "loss": 0.4021,
+      "step": 2602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7156285792097607e-05,
+      "loss": 0.3929,
+      "step": 2603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7153843904487546e-05,
+      "loss": 0.3609,
+      "step": 2604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7151401078793337e-05,
+      "loss": 0.4006,
+      "step": 2605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7148957315203527e-05,
+      "loss": 0.3885,
+      "step": 2606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7146512613906737e-05,
+      "loss": 0.4214,
+      "step": 2607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7144066975091654e-05,
+      "loss": 0.3887,
+      "step": 2608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7141620398947052e-05,
+      "loss": 0.4207,
+      "step": 2609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7139172885661762e-05,
+      "loss": 0.3723,
+      "step": 2610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7136724435424693e-05,
+      "loss": 0.3825,
+      "step": 2611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.713427504842483e-05,
+      "loss": 0.4062,
+      "step": 2612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7131824724851222e-05,
+      "loss": 0.4497,
+      "step": 2613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7129373464892994e-05,
+      "loss": 0.432,
+      "step": 2614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7126921268739353e-05,
+      "loss": 0.373,
+      "step": 2615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7124468136579567e-05,
+      "loss": 0.4704,
+      "step": 2616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7122014068602974e-05,
+      "loss": 0.3773,
+      "step": 2617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.711955906499899e-05,
+      "loss": 0.3537,
+      "step": 2618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7117103125957106e-05,
+      "loss": 0.392,
+      "step": 2619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.711464625166688e-05,
+      "loss": 0.4219,
+      "step": 2620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7112188442317944e-05,
+      "loss": 0.4087,
+      "step": 2621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.71097296981e-05,
+      "loss": 0.4561,
+      "step": 2622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7107270019202825e-05,
+      "loss": 0.4168,
+      "step": 2623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.710480940581627e-05,
+      "loss": 0.4547,
+      "step": 2624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7102347858130253e-05,
+      "loss": 0.4192,
+      "step": 2625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7099885376334763e-05,
+      "loss": 0.4025,
+      "step": 2626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7097421960619872e-05,
+      "loss": 0.399,
+      "step": 2627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.709495761117571e-05,
+      "loss": 0.4227,
+      "step": 2628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7092492328192494e-05,
+      "loss": 0.4035,
+      "step": 2629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7090026111860497e-05,
+      "loss": 0.3752,
+      "step": 2630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7087558962370073e-05,
+      "loss": 0.432,
+      "step": 2631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7085090879911644e-05,
+      "loss": 0.4079,
+      "step": 2632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7082621864675717e-05,
+      "loss": 0.3683,
+      "step": 2633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7080151916852855e-05,
+      "loss": 0.4061,
+      "step": 2634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7077681036633695e-05,
+      "loss": 0.3515,
+      "step": 2635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7075209224208953e-05,
+      "loss": 0.3876,
+      "step": 2636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.707273647976942e-05,
+      "loss": 0.4092,
+      "step": 2637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.707026280350594e-05,
+      "loss": 0.3718,
+      "step": 2638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.706778819560945e-05,
+      "loss": 0.4104,
+      "step": 2639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7065312656270952e-05,
+      "loss": 0.3987,
+      "step": 2640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7062836185681514e-05,
+      "loss": 0.3669,
+      "step": 2641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.706035878403228e-05,
+      "loss": 0.4232,
+      "step": 2642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.705788045151447e-05,
+      "loss": 0.4267,
+      "step": 2643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7055401188319366e-05,
+      "loss": 0.4219,
+      "step": 2644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7052920994638336e-05,
+      "loss": 0.4276,
+      "step": 2645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7050439870662806e-05,
+      "loss": 0.4162,
+      "step": 2646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.704795781658428e-05,
+      "loss": 0.4099,
+      "step": 2647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7045474832594335e-05,
+      "loss": 0.3899,
+      "step": 2648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7042990918884612e-05,
+      "loss": 0.3975,
+      "step": 2649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.704050607564684e-05,
+      "loss": 0.402,
+      "step": 2650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7038020303072802e-05,
+      "loss": 0.3984,
+      "step": 2651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7035533601354367e-05,
+      "loss": 0.4395,
+      "step": 2652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.703304597068346e-05,
+      "loss": 0.4016,
+      "step": 2653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7030557411252095e-05,
+      "loss": 0.3879,
+      "step": 2654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7028067923252342e-05,
+      "loss": 0.3873,
+      "step": 2655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7025577506876355e-05,
+      "loss": 0.4186,
+      "step": 2656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.702308616231635e-05,
+      "loss": 0.3607,
+      "step": 2657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7020593889764625e-05,
+      "loss": 0.443,
+      "step": 2658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7018100689413544e-05,
+      "loss": 0.3871,
+      "step": 2659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7015606561455537e-05,
+      "loss": 0.4018,
+      "step": 2660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.701311150608311e-05,
+      "loss": 0.4184,
+      "step": 2661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7010615523488855e-05,
+      "loss": 0.3775,
+      "step": 2662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7008118613865406e-05,
+      "loss": 0.3872,
+      "step": 2663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7005620777405496e-05,
+      "loss": 0.3909,
+      "step": 2664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7003122014301913e-05,
+      "loss": 0.4529,
+      "step": 2665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7000622324747526e-05,
+      "loss": 0.4087,
+      "step": 2666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6998121708935266e-05,
+      "loss": 0.3871,
+      "step": 2667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6995620167058144e-05,
+      "loss": 0.4178,
+      "step": 2668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6993117699309236e-05,
+      "loss": 0.3928,
+      "step": 2669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6990614305881704e-05,
+      "loss": 0.4433,
+      "step": 2670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.698810998696875e-05,
+      "loss": 0.4609,
+      "step": 2671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6985604742763692e-05,
+      "loss": 0.4116,
+      "step": 2672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.698309857345988e-05,
+      "loss": 0.3994,
+      "step": 2673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6980591479250753e-05,
+      "loss": 0.4391,
+      "step": 2674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6978083460329815e-05,
+      "loss": 0.381,
+      "step": 2675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6975574516890657e-05,
+      "loss": 0.4127,
+      "step": 2676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.697306464912692e-05,
+      "loss": 0.4352,
+      "step": 2677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6970553857232325e-05,
+      "loss": 0.4211,
+      "step": 2678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.696804214140067e-05,
+      "loss": 0.4,
+      "step": 2679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6965529501825822e-05,
+      "loss": 0.4325,
+      "step": 2680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6963015938701713e-05,
+      "loss": 0.3902,
+      "step": 2681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.696050145222235e-05,
+      "loss": 0.4018,
+      "step": 2682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6957986042581813e-05,
+      "loss": 0.3707,
+      "step": 2683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6955469709974253e-05,
+      "loss": 0.4066,
+      "step": 2684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6952952454593885e-05,
+      "loss": 0.4683,
+      "step": 2685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6950434276635006e-05,
+      "loss": 0.4144,
+      "step": 2686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694791517629198e-05,
+      "loss": 0.3985,
+      "step": 2687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6945395153759242e-05,
+      "loss": 0.4186,
+      "step": 2688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694287420923129e-05,
+      "loss": 0.4113,
+      "step": 2689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6940352342902714e-05,
+      "loss": 0.3854,
+      "step": 2690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6937829554968155e-05,
+      "loss": 0.3957,
+      "step": 2691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.693530584562233e-05,
+      "loss": 0.4216,
+      "step": 2692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6932781215060034e-05,
+      "loss": 0.3802,
+      "step": 2693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6930255663476122e-05,
+      "loss": 0.4088,
+      "step": 2694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6927729191065536e-05,
+      "loss": 0.3827,
+      "step": 2695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6925201798023272e-05,
+      "loss": 0.4448,
+      "step": 2696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6922673484544406e-05,
+      "loss": 0.4083,
+      "step": 2697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6920144250824084e-05,
+      "loss": 0.3711,
+      "step": 2698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6917614097057534e-05,
+      "loss": 0.3994,
+      "step": 2699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.691508302344002e-05,
+      "loss": 0.4387,
+      "step": 2700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6912551030166922e-05,
+      "loss": 0.444,
+      "step": 2701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.691001811743366e-05,
+      "loss": 0.4156,
+      "step": 2702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.690748428543574e-05,
+      "loss": 0.3886,
+      "step": 2703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6904949534368724e-05,
+      "loss": 0.3814,
+      "step": 2704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6902413864428264e-05,
+      "loss": 0.3886,
+      "step": 2705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6899877275810072e-05,
+      "loss": 0.4043,
+      "step": 2706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.689733976870993e-05,
+      "loss": 0.3814,
+      "step": 2707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6894801343323695e-05,
+      "loss": 0.4011,
+      "step": 2708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6892261999847293e-05,
+      "loss": 0.3692,
+      "step": 2709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.688972173847672e-05,
+      "loss": 0.401,
+      "step": 2710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6887180559408044e-05,
+      "loss": 0.3709,
+      "step": 2711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6884638462837407e-05,
+      "loss": 0.3871,
+      "step": 2712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6882095448961012e-05,
+      "loss": 0.4281,
+      "step": 2713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6879551517975147e-05,
+      "loss": 0.3888,
+      "step": 2714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6877006670076157e-05,
+      "loss": 0.3617,
+      "step": 2715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.687446090546047e-05,
+      "loss": 0.3608,
+      "step": 2716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.687191422432457e-05,
+      "loss": 0.4395,
+      "step": 2717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6869366626865026e-05,
+      "loss": 0.39,
+      "step": 2718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6866818113278474e-05,
+      "loss": 0.381,
+      "step": 2719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.686426868376162e-05,
+      "loss": 0.3979,
+      "step": 2720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.686171833851123e-05,
+      "loss": 0.3363,
+      "step": 2721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6859167077724155e-05,
+      "loss": 0.3786,
+      "step": 2722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.685661490159732e-05,
+      "loss": 0.3826,
+      "step": 2723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6854061810327704e-05,
+      "loss": 0.3819,
+      "step": 2724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.685150780411236e-05,
+      "loss": 0.4055,
+      "step": 2725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6848952883148435e-05,
+      "loss": 0.3502,
+      "step": 2726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6846397047633107e-05,
+      "loss": 0.4252,
+      "step": 2727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.684384029776366e-05,
+      "loss": 0.4549,
+      "step": 2728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6841282633737434e-05,
+      "loss": 0.4262,
+      "step": 2729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.683872405575183e-05,
+      "loss": 0.4163,
+      "step": 2730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6836164564004346e-05,
+      "loss": 0.4336,
+      "step": 2731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6833604158692518e-05,
+      "loss": 0.4182,
+      "step": 2732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6831042840013978e-05,
+      "loss": 0.4003,
+      "step": 2733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6828480608166415e-05,
+      "loss": 0.387,
+      "step": 2734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6825917463347595e-05,
+      "loss": 0.4524,
+      "step": 2735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682335340575535e-05,
+      "loss": 0.4055,
+      "step": 2736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6820788435587587e-05,
+      "loss": 0.3692,
+      "step": 2737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6818222553042282e-05,
+      "loss": 0.3767,
+      "step": 2738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.681565575831748e-05,
+      "loss": 0.416,
+      "step": 2739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.681308805161129e-05,
+      "loss": 0.3912,
+      "step": 2740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6810519433121907e-05,
+      "loss": 0.3728,
+      "step": 2741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6807949903047587e-05,
+      "loss": 0.3877,
+      "step": 2742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6805379461586652e-05,
+      "loss": 0.4096,
+      "step": 2743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.68028081089375e-05,
+      "loss": 0.4278,
+      "step": 2744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6800235845298604e-05,
+      "loss": 0.4172,
+      "step": 2745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6797662670868497e-05,
+      "loss": 0.4014,
+      "step": 2746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6795088585845785e-05,
+      "loss": 0.3993,
+      "step": 2747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6792513590429158e-05,
+      "loss": 0.3924,
+      "step": 2748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6789937684817353e-05,
+      "loss": 0.3969,
+      "step": 2749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6787360869209192e-05,
+      "loss": 0.4075,
+      "step": 2750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6784783143803567e-05,
+      "loss": 0.4073,
+      "step": 2751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6782204508799437e-05,
+      "loss": 0.4076,
+      "step": 2752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.677962496439583e-05,
+      "loss": 0.3592,
+      "step": 2753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6777044510791847e-05,
+      "loss": 0.4063,
+      "step": 2754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6774463148186657e-05,
+      "loss": 0.3907,
+      "step": 2755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6771880876779503e-05,
+      "loss": 0.3676,
+      "step": 2756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6769297696769692e-05,
+      "loss": 0.4589,
+      "step": 2757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.676671360835661e-05,
+      "loss": 0.3776,
+      "step": 2758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6764128611739704e-05,
+      "loss": 0.3938,
+      "step": 2759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.676154270711849e-05,
+      "loss": 0.4186,
+      "step": 2760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6758955894692567e-05,
+      "loss": 0.4078,
+      "step": 2761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6756368174661588e-05,
+      "loss": 0.3584,
+      "step": 2762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.675377954722529e-05,
+      "loss": 0.4459,
+      "step": 2763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6751190012583476e-05,
+      "loss": 0.3965,
+      "step": 2764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6748599570936013e-05,
+      "loss": 0.3907,
+      "step": 2765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.674600822248284e-05,
+      "loss": 0.4113,
+      "step": 2766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6743415967423973e-05,
+      "loss": 0.4169,
+      "step": 2767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.674082280595949e-05,
+      "loss": 0.3778,
+      "step": 2768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6738228738289544e-05,
+      "loss": 0.4246,
+      "step": 2769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.673563376461435e-05,
+      "loss": 0.3808,
+      "step": 2770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6733037885134205e-05,
+      "loss": 0.3651,
+      "step": 2771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.673044110004947e-05,
+      "loss": 0.3822,
+      "step": 2772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6727843409560572e-05,
+      "loss": 0.4225,
+      "step": 2773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6725244813868016e-05,
+      "loss": 0.3861,
+      "step": 2774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6722645313172367e-05,
+      "loss": 0.4322,
+      "step": 2775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6720044907674267e-05,
+      "loss": 0.3889,
+      "step": 2776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.671744359757443e-05,
+      "loss": 0.41,
+      "step": 2777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6714841383073635e-05,
+      "loss": 0.4429,
+      "step": 2778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6712238264372724e-05,
+      "loss": 0.4106,
+      "step": 2779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6709634241672624e-05,
+      "loss": 0.3659,
+      "step": 2780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6707029315174325e-05,
+      "loss": 0.3737,
+      "step": 2781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.670442348507888e-05,
+      "loss": 0.4017,
+      "step": 2782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6701816751587428e-05,
+      "loss": 0.4077,
+      "step": 2783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6699209114901156e-05,
+      "loss": 0.3843,
+      "step": 2784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.669660057522134e-05,
+      "loss": 0.388,
+      "step": 2785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6693991132749312e-05,
+      "loss": 0.4183,
+      "step": 2786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.669138078768648e-05,
+      "loss": 0.4309,
+      "step": 2787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6688769540234327e-05,
+      "loss": 0.4097,
+      "step": 2788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6686157390594398e-05,
+      "loss": 0.4303,
+      "step": 2789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6683544338968308e-05,
+      "loss": 0.3992,
+      "step": 2790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6680930385557742e-05,
+      "loss": 0.3897,
+      "step": 2791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.667831553056446e-05,
+      "loss": 0.3705,
+      "step": 2792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6675699774190282e-05,
+      "loss": 0.4062,
+      "step": 2793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.667308311663711e-05,
+      "loss": 0.4338,
+      "step": 2794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.66704655581069e-05,
+      "loss": 0.4307,
+      "step": 2795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6667847098801688e-05,
+      "loss": 0.3917,
+      "step": 2796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6665227738923585e-05,
+      "loss": 0.3579,
+      "step": 2797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6662607478674758e-05,
+      "loss": 0.3439,
+      "step": 2798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6659986318257445e-05,
+      "loss": 0.4181,
+      "step": 2799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6657364257873968e-05,
+      "loss": 0.3972,
+      "step": 2800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6654741297726707e-05,
+      "loss": 0.3773,
+      "step": 2801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6652117438018106e-05,
+      "loss": 0.3478,
+      "step": 2802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6649492678950688e-05,
+      "loss": 0.4367,
+      "step": 2803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6646867020727042e-05,
+      "loss": 0.397,
+      "step": 2804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6644240463549837e-05,
+      "loss": 0.4213,
+      "step": 2805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.664161300762179e-05,
+      "loss": 0.393,
+      "step": 2806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6638984653145703e-05,
+      "loss": 0.3769,
+      "step": 2807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6636355400324442e-05,
+      "loss": 0.3855,
+      "step": 2808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6633725249360944e-05,
+      "loss": 0.3771,
+      "step": 2809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6631094200458218e-05,
+      "loss": 0.4469,
+      "step": 2810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6628462253819342e-05,
+      "loss": 0.3596,
+      "step": 2811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.662582940964745e-05,
+      "loss": 0.4095,
+      "step": 2812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.662319566814576e-05,
+      "loss": 0.3383,
+      "step": 2813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.662056102951756e-05,
+      "loss": 0.3898,
+      "step": 2814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.66179254939662e-05,
+      "loss": 0.3874,
+      "step": 2815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6615289061695094e-05,
+      "loss": 0.398,
+      "step": 2816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6612651732907746e-05,
+      "loss": 0.5003,
+      "step": 2817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6610013507807707e-05,
+      "loss": 0.4629,
+      "step": 2818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.660737438659861e-05,
+      "loss": 0.399,
+      "step": 2819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.660473436948415e-05,
+      "loss": 0.4164,
+      "step": 2820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6602093456668094e-05,
+      "loss": 0.392,
+      "step": 2821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6599451648354286e-05,
+      "loss": 0.3906,
+      "step": 2822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6596808944746625e-05,
+      "loss": 0.3865,
+      "step": 2823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.659416534604908e-05,
+      "loss": 0.3651,
+      "step": 2824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6591520852465714e-05,
+      "loss": 0.4306,
+      "step": 2825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6588875464200625e-05,
+      "loss": 0.4355,
+      "step": 2826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6586229181457994e-05,
+      "loss": 0.3936,
+      "step": 2827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.658358200444208e-05,
+      "loss": 0.3794,
+      "step": 2828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.65809339333572e-05,
+      "loss": 0.4042,
+      "step": 2829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6578284968407743e-05,
+      "loss": 0.4037,
+      "step": 2830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6575635109798163e-05,
+      "loss": 0.4218,
+      "step": 2831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6572984357732995e-05,
+      "loss": 0.4016,
+      "step": 2832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6570332712416833e-05,
+      "loss": 0.3736,
+      "step": 2833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.656768017405434e-05,
+      "loss": 0.3527,
+      "step": 2834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6565026742850244e-05,
+      "loss": 0.4104,
+      "step": 2835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.656237241900936e-05,
+      "loss": 0.3604,
+      "step": 2836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6559717202736545e-05,
+      "loss": 0.4018,
+      "step": 2837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6557061094236756e-05,
+      "loss": 0.3634,
+      "step": 2838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6554404093714992e-05,
+      "loss": 0.4341,
+      "step": 2839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6551746201376338e-05,
+      "loss": 0.3299,
+      "step": 2840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.654908741742593e-05,
+      "loss": 0.4285,
+      "step": 2841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6546427742068998e-05,
+      "loss": 0.4585,
+      "step": 2842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.654376717551082e-05,
+      "loss": 0.3954,
+      "step": 2843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6541105717956743e-05,
+      "loss": 0.398,
+      "step": 2844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.65384433696122e-05,
+      "loss": 0.4404,
+      "step": 2845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6535780130682675e-05,
+      "loss": 0.4156,
+      "step": 2846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6533116001373737e-05,
+      "loss": 0.4123,
+      "step": 2847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6530450981891e-05,
+      "loss": 0.3671,
+      "step": 2848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6527785072440178e-05,
+      "loss": 0.3409,
+      "step": 2849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6525118273227022e-05,
+      "loss": 0.4089,
+      "step": 2850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.652245058445738e-05,
+      "loss": 0.4367,
+      "step": 2851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6519782006337142e-05,
+      "loss": 0.4179,
+      "step": 2852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6517112539072293e-05,
+      "loss": 0.416,
+      "step": 2853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6514442182868863e-05,
+      "loss": 0.4012,
+      "step": 2854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6511770937932963e-05,
+      "loss": 0.3948,
+      "step": 2855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6509098804470776e-05,
+      "loss": 0.3679,
+      "step": 2856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6506425782688546e-05,
+      "loss": 0.3874,
+      "step": 2857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.650375187279259e-05,
+      "loss": 0.3233,
+      "step": 2858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6501077074989284e-05,
+      "loss": 0.439,
+      "step": 2859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6498401389485085e-05,
+      "loss": 0.3695,
+      "step": 2860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6495724816486517e-05,
+      "loss": 0.3989,
+      "step": 2861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.649304735620016e-05,
+      "loss": 0.4255,
+      "step": 2862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.649036900883268e-05,
+      "loss": 0.3643,
+      "step": 2863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.64876897745908e-05,
+      "loss": 0.4275,
+      "step": 2864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6485009653681315e-05,
+      "loss": 0.4303,
+      "step": 2865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6482328646311088e-05,
+      "loss": 0.354,
+      "step": 2866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6479646752687045e-05,
+      "loss": 0.4019,
+      "step": 2867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6476963973016192e-05,
+      "loss": 0.4019,
+      "step": 2868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6474280307505595e-05,
+      "loss": 0.4758,
+      "step": 2869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6471595756362392e-05,
+      "loss": 0.3926,
+      "step": 2870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6468910319793787e-05,
+      "loss": 0.4342,
+      "step": 2871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.646622399800705e-05,
+      "loss": 0.4029,
+      "step": 2872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6463536791209526e-05,
+      "loss": 0.4014,
+      "step": 2873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.646084869960862e-05,
+      "loss": 0.4336,
+      "step": 2874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.645815972341182e-05,
+      "loss": 0.3869,
+      "step": 2875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6455469862826666e-05,
+      "loss": 0.3686,
+      "step": 2876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.645277911806077e-05,
+      "loss": 0.3973,
+      "step": 2877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.645008748932182e-05,
+      "loss": 0.3851,
+      "step": 2878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6447394976817565e-05,
+      "loss": 0.4035,
+      "step": 2879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.644470158075582e-05,
+      "loss": 0.3844,
+      "step": 2880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.644200730134448e-05,
+      "loss": 0.434,
+      "step": 2881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6439312138791494e-05,
+      "loss": 0.3829,
+      "step": 2882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.643661609330489e-05,
+      "loss": 0.4179,
+      "step": 2883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6433919165092766e-05,
+      "loss": 0.4085,
+      "step": 2884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6431221354363267e-05,
+      "loss": 0.363,
+      "step": 2885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6428522661324636e-05,
+      "loss": 0.3946,
+      "step": 2886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.642582308618516e-05,
+      "loss": 0.4187,
+      "step": 2887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.64231226291532e-05,
+      "loss": 0.4184,
+      "step": 2888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6420421290437202e-05,
+      "loss": 0.4228,
+      "step": 2889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6417719070245657e-05,
+      "loss": 0.4222,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6415015968787135e-05,
+      "loss": 0.3459,
+      "step": 2891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6412311986270274e-05,
+      "loss": 0.4063,
+      "step": 2892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640960712290378e-05,
+      "loss": 0.3987,
+      "step": 2893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640690137889642e-05,
+      "loss": 0.3914,
+      "step": 2894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640419475445704e-05,
+      "loss": 0.3873,
+      "step": 2895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.640148724979455e-05,
+      "loss": 0.3231,
+      "step": 2896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.639877886511792e-05,
+      "loss": 0.3536,
+      "step": 2897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6396069600636194e-05,
+      "loss": 0.3624,
+      "step": 2898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.639335945655849e-05,
+      "loss": 0.423,
+      "step": 2899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.639064843309399e-05,
+      "loss": 0.3948,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6387936530451934e-05,
+      "loss": 0.4098,
+      "step": 2901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.638522374884164e-05,
+      "loss": 0.4048,
+      "step": 2902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.63825100884725e-05,
+      "loss": 0.3635,
+      "step": 2903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6379795549553954e-05,
+      "loss": 0.421,
+      "step": 2904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.637708013229553e-05,
+      "loss": 0.3749,
+      "step": 2905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.637436383690681e-05,
+      "loss": 0.4039,
+      "step": 2906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6371646663597452e-05,
+      "loss": 0.3435,
+      "step": 2907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6368928612577176e-05,
+      "loss": 0.3651,
+      "step": 2908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6366209684055775e-05,
+      "loss": 0.4284,
+      "step": 2909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6363489878243105e-05,
+      "loss": 0.4045,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.636076919534909e-05,
+      "loss": 0.3766,
+      "step": 2911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6358047635583734e-05,
+      "loss": 0.3939,
+      "step": 2912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6355325199157087e-05,
+      "loss": 0.4047,
+      "step": 2913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6352601886279282e-05,
+      "loss": 0.403,
+      "step": 2914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6349877697160513e-05,
+      "loss": 0.3938,
+      "step": 2915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6347152632011055e-05,
+      "loss": 0.4473,
+      "step": 2916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6344426691041224e-05,
+      "loss": 0.4303,
+      "step": 2917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.634169987446143e-05,
+      "loss": 0.429,
+      "step": 2918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6338972182482135e-05,
+      "loss": 0.4204,
+      "step": 2919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6336243615313876e-05,
+      "loss": 0.4302,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6333514173167256e-05,
+      "loss": 0.4147,
+      "step": 2921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6330783856252944e-05,
+      "loss": 0.354,
+      "step": 2922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6328052664781674e-05,
+      "loss": 0.3855,
+      "step": 2923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6325320598964256e-05,
+      "loss": 0.4104,
+      "step": 2924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6322587659011554e-05,
+      "loss": 0.4421,
+      "step": 2925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6319853845134523e-05,
+      "loss": 0.4453,
+      "step": 2926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.631711915754415e-05,
+      "loss": 0.4184,
+      "step": 2927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6314383596451528e-05,
+      "loss": 0.3795,
+      "step": 2928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.631164716206779e-05,
+      "loss": 0.3577,
+      "step": 2929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.630890985460414e-05,
+      "loss": 0.3676,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6306171674271866e-05,
+      "loss": 0.4164,
+      "step": 2931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6303432621282306e-05,
+      "loss": 0.3737,
+      "step": 2932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6300692695846872e-05,
+      "loss": 0.3986,
+      "step": 2933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6297951898177045e-05,
+      "loss": 0.4257,
+      "step": 2934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.629521022848437e-05,
+      "loss": 0.3958,
+      "step": 2935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6292467686980453e-05,
+      "loss": 0.4329,
+      "step": 2936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.628972427387699e-05,
+      "loss": 0.3866,
+      "step": 2937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6286979989385718e-05,
+      "loss": 0.4048,
+      "step": 2938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6284234833718455e-05,
+      "loss": 0.448,
+      "step": 2939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6281488807087083e-05,
+      "loss": 0.3623,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.627874190970355e-05,
+      "loss": 0.3508,
+      "step": 2941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6275994141779882e-05,
+      "loss": 0.446,
+      "step": 2942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6273245503528152e-05,
+      "loss": 0.4095,
+      "step": 2943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6270495995160516e-05,
+      "loss": 0.411,
+      "step": 2944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.626774561688919e-05,
+      "loss": 0.3744,
+      "step": 2945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6264994368926464e-05,
+      "loss": 0.3785,
+      "step": 2946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.626224225148469e-05,
+      "loss": 0.3985,
+      "step": 2947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6259489264776287e-05,
+      "loss": 0.4303,
+      "step": 2948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.625673540901374e-05,
+      "loss": 0.3761,
+      "step": 2949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.62539806844096e-05,
+      "loss": 0.4183,
+      "step": 2950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.62512250911765e-05,
+      "loss": 0.4099,
+      "step": 2951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.624846862952711e-05,
+      "loss": 0.3925,
+      "step": 2952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6245711299674204e-05,
+      "loss": 0.3754,
+      "step": 2953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6242953101830593e-05,
+      "loss": 0.4274,
+      "step": 2954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.624019403620917e-05,
+      "loss": 0.389,
+      "step": 2955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.623743410302289e-05,
+      "loss": 0.357,
+      "step": 2956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.623467330248478e-05,
+      "loss": 0.3882,
+      "step": 2957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6231911634807923e-05,
+      "loss": 0.4017,
+      "step": 2958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6229149100205478e-05,
+      "loss": 0.4124,
+      "step": 2959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6226385698890672e-05,
+      "loss": 0.4292,
+      "step": 2960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6223621431076793e-05,
+      "loss": 0.3996,
+      "step": 2961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6220856296977203e-05,
+      "loss": 0.3993,
+      "step": 2962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6218090296805317e-05,
+      "loss": 0.3657,
+      "step": 2963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.621532343077464e-05,
+      "loss": 0.3683,
+      "step": 2964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6212555699098715e-05,
+      "loss": 0.3817,
+      "step": 2965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6209787101991182e-05,
+      "loss": 0.3841,
+      "step": 2966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.620701763966572e-05,
+      "loss": 0.4037,
+      "step": 2967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6204247312336095e-05,
+      "loss": 0.4057,
+      "step": 2968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.620147612021613e-05,
+      "loss": 0.3654,
+      "step": 2969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.619870406351972e-05,
+      "loss": 0.366,
+      "step": 2970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6195931142460824e-05,
+      "loss": 0.3117,
+      "step": 2971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.619315735725346e-05,
+      "loss": 0.3983,
+      "step": 2972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6190382708111727e-05,
+      "loss": 0.3925,
+      "step": 2973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6187607195249784e-05,
+      "loss": 0.3839,
+      "step": 2974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6184830818881854e-05,
+      "loss": 0.3737,
+      "step": 2975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.618205357922223e-05,
+      "loss": 0.4102,
+      "step": 2976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6179275476485273e-05,
+      "loss": 0.4201,
+      "step": 2977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6176496510885403e-05,
+      "loss": 0.3981,
+      "step": 2978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.617371668263712e-05,
+      "loss": 0.4342,
+      "step": 2979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.617093599195497e-05,
+      "loss": 0.3637,
+      "step": 2980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6168154439053594e-05,
+      "loss": 0.3691,
+      "step": 2981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6165372024147678e-05,
+      "loss": 0.458,
+      "step": 2982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6162588747451976e-05,
+      "loss": 0.3893,
+      "step": 2983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6159804609181312e-05,
+      "loss": 0.4267,
+      "step": 2984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6157019609550587e-05,
+      "loss": 0.3634,
+      "step": 2985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.615423374877475e-05,
+      "loss": 0.3778,
+      "step": 2986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6151447027068825e-05,
+      "loss": 0.4229,
+      "step": 2987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6148659444647913e-05,
+      "loss": 0.373,
+      "step": 2988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.614587100172716e-05,
+      "loss": 0.4214,
+      "step": 2989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6143081698521795e-05,
+      "loss": 0.3891,
+      "step": 2990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6140291535247105e-05,
+      "loss": 0.4608,
+      "step": 2991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.613750051211845e-05,
+      "loss": 0.3999,
+      "step": 2992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.613470862935125e-05,
+      "loss": 0.3573,
+      "step": 2993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.613191588716099e-05,
+      "loss": 0.4823,
+      "step": 2994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.612912228576324e-05,
+      "loss": 0.3671,
+      "step": 2995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6126327825373608e-05,
+      "loss": 0.3891,
+      "step": 2996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6123532506207786e-05,
+      "loss": 0.4291,
+      "step": 2997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.612073632848153e-05,
+      "loss": 0.43,
+      "step": 2998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6117939292410657e-05,
+      "loss": 0.3727,
+      "step": 2999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6115141398211055e-05,
+      "loss": 0.3946,
+      "step": 3000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.6112342646098682e-05,
+      "loss": 0.4157,
+      "step": 3001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.610954303628955e-05,
+      "loss": 0.3955,
+      "step": 3002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6106742568999753e-05,
+      "loss": 0.4038,
+      "step": 3003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6103941244445433e-05,
+      "loss": 0.3959,
+      "step": 3004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.610113906284281e-05,
+      "loss": 0.4039,
+      "step": 3005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6098336024408175e-05,
+      "loss": 0.384,
+      "step": 3006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6095532129357874e-05,
+      "loss": 0.4083,
+      "step": 3007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6092727377908322e-05,
+      "loss": 0.3626,
+      "step": 3008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6089921770276e-05,
+      "loss": 0.3755,
+      "step": 3009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6087115306677464e-05,
+      "loss": 0.384,
+      "step": 3010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6084307987329318e-05,
+      "loss": 0.3626,
+      "step": 3011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6081499812448248e-05,
+      "loss": 0.4052,
+      "step": 3012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6078690782251008e-05,
+      "loss": 0.3649,
+      "step": 3013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6075880896954393e-05,
+      "loss": 0.3864,
+      "step": 3014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6073070156775296e-05,
+      "loss": 0.421,
+      "step": 3015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.607025856193066e-05,
+      "loss": 0.4052,
+      "step": 3016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6067446112637492e-05,
+      "loss": 0.3728,
+      "step": 3017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6064632809112866e-05,
+      "loss": 0.394,
+      "step": 3018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.606181865157393e-05,
+      "loss": 0.3279,
+      "step": 3019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6059003640237887e-05,
+      "loss": 0.3921,
+      "step": 3020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.605618777532202e-05,
+      "loss": 0.3772,
+      "step": 3021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.605337105704366e-05,
+      "loss": 0.4224,
+      "step": 3022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.605055348562022e-05,
+      "loss": 0.3926,
+      "step": 3023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6047735061269163e-05,
+      "loss": 0.428,
+      "step": 3024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6044915784208035e-05,
+      "loss": 0.3763,
+      "step": 3025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6042095654654434e-05,
+      "loss": 0.3735,
+      "step": 3026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6039274672826038e-05,
+      "loss": 0.4018,
+      "step": 3027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.603645283894057e-05,
+      "loss": 0.3506,
+      "step": 3028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.603363015321584e-05,
+      "loss": 0.3798,
+      "step": 3029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6030806615869712e-05,
+      "loss": 0.3819,
+      "step": 3030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6027982227120117e-05,
+      "loss": 0.4271,
+      "step": 3031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6025156987185057e-05,
+      "loss": 0.3984,
+      "step": 3032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6022330896282586e-05,
+      "loss": 0.3264,
+      "step": 3033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6019503954630847e-05,
+      "loss": 0.3721,
+      "step": 3034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6016676162448026e-05,
+      "loss": 0.3775,
+      "step": 3035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6013847519952384e-05,
+      "loss": 0.4081,
+      "step": 3036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.601101802736225e-05,
+      "loss": 0.4048,
+      "step": 3037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.600818768489602e-05,
+      "loss": 0.3629,
+      "step": 3038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6005356492772137e-05,
+      "loss": 0.4077,
+      "step": 3039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.600252445120914e-05,
+      "loss": 0.3979,
+      "step": 3040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5999691560425614e-05,
+      "loss": 0.3966,
+      "step": 3041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.599685782064021e-05,
+      "loss": 0.4266,
+      "step": 3042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5994023232071644e-05,
+      "loss": 0.4174,
+      "step": 3043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5991187794938714e-05,
+      "loss": 0.4143,
+      "step": 3044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5988351509460254e-05,
+      "loss": 0.3983,
+      "step": 3045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.59855143758552e-05,
+      "loss": 0.4213,
+      "step": 3046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5982676394342516e-05,
+      "loss": 0.3784,
+      "step": 3047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.597983756514126e-05,
+      "loss": 0.3992,
+      "step": 3048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5976997888470535e-05,
+      "loss": 0.4248,
+      "step": 3049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5974157364549528e-05,
+      "loss": 0.3702,
+      "step": 3050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5971315993597478e-05,
+      "loss": 0.3887,
+      "step": 3051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5968473775833702e-05,
+      "loss": 0.4409,
+      "step": 3052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.596563071147756e-05,
+      "loss": 0.3652,
+      "step": 3053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5962786800748502e-05,
+      "loss": 0.4137,
+      "step": 3054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5959942043866027e-05,
+      "loss": 0.4444,
+      "step": 3055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.595709644104971e-05,
+      "loss": 0.4277,
+      "step": 3056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5954249992519184e-05,
+      "loss": 0.4212,
+      "step": 3057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5951402698494152e-05,
+      "loss": 0.4217,
+      "step": 3058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.594855455919438e-05,
+      "loss": 0.4465,
+      "step": 3059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.5945705574839687e-05,
+      "loss": 0.4572,
+      "step": 3060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.594285574564999e-05,
+      "loss": 0.4124,
+      "step": 3061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.594000507184524e-05,
+      "loss": 0.4024,
+      "step": 3062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.593715355364546e-05,
+      "loss": 0.4506,
+      "step": 3063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5934301191270745e-05,
+      "loss": 0.4236,
+      "step": 3064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5931447984941254e-05,
+      "loss": 0.3677,
+      "step": 3065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5928593934877208e-05,
+      "loss": 0.43,
+      "step": 3066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5925739041298902e-05,
+      "loss": 0.3698,
+      "step": 3067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5922883304426673e-05,
+      "loss": 0.4115,
+      "step": 3068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.592002672448095e-05,
+      "loss": 0.4205,
+      "step": 3069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5917169301682217e-05,
+      "loss": 0.3723,
+      "step": 3070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.591431103625101e-05,
+      "loss": 0.4651,
+      "step": 3071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5911451928407958e-05,
+      "loss": 0.3784,
+      "step": 3072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5908591978373727e-05,
+      "loss": 0.3698,
+      "step": 3073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5905731186369064e-05,
+      "loss": 0.4135,
+      "step": 3074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5902869552614772e-05,
+      "loss": 0.3473,
+      "step": 3075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5900007077331735e-05,
+      "loss": 0.3406,
+      "step": 3076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5897143760740877e-05,
+      "loss": 0.3808,
+      "step": 3077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.589427960306321e-05,
+      "loss": 0.4408,
+      "step": 3078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5891414604519797e-05,
+      "loss": 0.4259,
+      "step": 3079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.588854876533177e-05,
+      "loss": 0.4495,
+      "step": 3080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5885682085720333e-05,
+      "loss": 0.3959,
+      "step": 3081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.588281456590674e-05,
+      "loss": 0.4157,
+      "step": 3082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5879946206112327e-05,
+      "loss": 0.4007,
+      "step": 3083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5877077006558478e-05,
+      "loss": 0.4054,
+      "step": 3084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5874206967466653e-05,
+      "loss": 0.4179,
+      "step": 3085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.587133608905837e-05,
+      "loss": 0.3424,
+      "step": 3086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5868464371555224e-05,
+      "loss": 0.4585,
+      "step": 3087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.586559181517886e-05,
+      "loss": 0.3854,
+      "step": 3088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5862718420150993e-05,
+      "loss": 0.4133,
+      "step": 3089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5859844186693403e-05,
+      "loss": 0.3924,
+      "step": 3090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5856969115027937e-05,
+      "loss": 0.3251,
+      "step": 3091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.585409320537651e-05,
+      "loss": 0.3749,
+      "step": 3092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5851216457961087e-05,
+      "loss": 0.353,
+      "step": 3093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5848338873003718e-05,
+      "loss": 0.3822,
+      "step": 3094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.584546045072649e-05,
+      "loss": 0.4018,
+      "step": 3095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.584258119135159e-05,
+      "loss": 0.3657,
+      "step": 3096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5839701095101246e-05,
+      "loss": 0.4284,
+      "step": 3097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.583682016219775e-05,
+      "loss": 0.3425,
+      "step": 3098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5833938392863467e-05,
+      "loss": 0.3739,
+      "step": 3099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.583105578732083e-05,
+      "loss": 0.4342,
+      "step": 3100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.582817234579232e-05,
+      "loss": 0.3692,
+      "step": 3101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5825288068500502e-05,
+      "loss": 0.4108,
+      "step": 3102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.582240295566799e-05,
+      "loss": 0.4185,
+      "step": 3103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5819517007517477e-05,
+      "loss": 0.3684,
+      "step": 3104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5816630224271703e-05,
+      "loss": 0.37,
+      "step": 3105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5813742606153485e-05,
+      "loss": 0.4412,
+      "step": 3106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5810854153385708e-05,
+      "loss": 0.429,
+      "step": 3107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5807964866191307e-05,
+      "loss": 0.4517,
+      "step": 3108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5805074744793292e-05,
+      "loss": 0.3665,
+      "step": 3109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.580218378941473e-05,
+      "loss": 0.3906,
+      "step": 3110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5799292000278762e-05,
+      "loss": 0.394,
+      "step": 3111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5796399377608592e-05,
+      "loss": 0.3825,
+      "step": 3112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5793505921627476e-05,
+      "loss": 0.4633,
+      "step": 3113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.579061163255875e-05,
+      "loss": 0.3995,
+      "step": 3114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5787716510625804e-05,
+      "loss": 0.4088,
+      "step": 3115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5784820556052094e-05,
+      "loss": 0.388,
+      "step": 3116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5781923769061143e-05,
+      "loss": 0.3643,
+      "step": 3117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.577902614987654e-05,
+      "loss": 0.3589,
+      "step": 3118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.577612769872193e-05,
+      "loss": 0.4028,
+      "step": 3119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5773228415821033e-05,
+      "loss": 0.4045,
+      "step": 3120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.577032830139763e-05,
+      "loss": 0.4329,
+      "step": 3121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.576742735567555e-05,
+      "loss": 0.4327,
+      "step": 3122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5764525578878713e-05,
+      "loss": 0.3874,
+      "step": 3123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5761622971231082e-05,
+      "loss": 0.3644,
+      "step": 3124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.57587195329567e-05,
+      "loss": 0.3729,
+      "step": 3125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5755815264279663e-05,
+      "loss": 0.429,
+      "step": 3126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5752910165424138e-05,
+      "loss": 0.3206,
+      "step": 3127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.575000423661434e-05,
+      "loss": 0.387,
+      "step": 3128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5747097478074582e-05,
+      "loss": 0.4313,
+      "step": 3129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5744189890029197e-05,
+      "loss": 0.4409,
+      "step": 3130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5741281472702618e-05,
+      "loss": 0.3712,
+      "step": 3131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5738372226319327e-05,
+      "loss": 0.3665,
+      "step": 3132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.573546215110387e-05,
+      "loss": 0.4166,
+      "step": 3133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.573255124728086e-05,
+      "loss": 0.4214,
+      "step": 3134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5729639515074974e-05,
+      "loss": 0.3771,
+      "step": 3135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5726726954710947e-05,
+      "loss": 0.44,
+      "step": 3136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5723813566413583e-05,
+      "loss": 0.3863,
+      "step": 3137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5720899350407752e-05,
+      "loss": 0.3643,
+      "step": 3138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.571798430691839e-05,
+      "loss": 0.3924,
+      "step": 3139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.571506843617048e-05,
+      "loss": 0.4,
+      "step": 3140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5712151738389093e-05,
+      "loss": 0.3667,
+      "step": 3141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5709234213799343e-05,
+      "loss": 0.396,
+      "step": 3142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5706315862626423e-05,
+      "loss": 0.3744,
+      "step": 3143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.570339668509558e-05,
+      "loss": 0.3718,
+      "step": 3144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5700476681432128e-05,
+      "loss": 0.3435,
+      "step": 3145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5697555851861448e-05,
+      "loss": 0.3965,
+      "step": 3146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5694634196608977e-05,
+      "loss": 0.3996,
+      "step": 3147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5691711715900226e-05,
+      "loss": 0.4005,
+      "step": 3148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.568878840996076e-05,
+      "loss": 0.3744,
+      "step": 3149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.568586427901622e-05,
+      "loss": 0.4231,
+      "step": 3150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5682939323292284e-05,
+      "loss": 0.355,
+      "step": 3151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5680013543014734e-05,
+      "loss": 0.3875,
+      "step": 3152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5677086938409378e-05,
+      "loss": 0.4237,
+      "step": 3153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5674159509702113e-05,
+      "loss": 0.3633,
+      "step": 3154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5671231257118885e-05,
+      "loss": 0.4029,
+      "step": 3155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5668302180885714e-05,
+      "loss": 0.3702,
+      "step": 3156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.566537228122867e-05,
+      "loss": 0.4408,
+      "step": 3157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5662441558373903e-05,
+      "loss": 0.4101,
+      "step": 3158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5659510012547612e-05,
+      "loss": 0.4562,
+      "step": 3159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.565657764397607e-05,
+      "loss": 0.3718,
+      "step": 3160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5653644452885608e-05,
+      "loss": 0.4074,
+      "step": 3161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5650710439502624e-05,
+      "loss": 0.402,
+      "step": 3162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5647775604053578e-05,
+      "loss": 0.4431,
+      "step": 3163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.564483994676498e-05,
+      "loss": 0.3442,
+      "step": 3164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5641903467863433e-05,
+      "loss": 0.3946,
+      "step": 3165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.563896616757558e-05,
+      "loss": 0.3624,
+      "step": 3166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5636028046128133e-05,
+      "loss": 0.3894,
+      "step": 3167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.563308910374787e-05,
+      "loss": 0.3873,
+      "step": 3168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.563014934066163e-05,
+      "loss": 0.3862,
+      "step": 3169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5627208757096316e-05,
+      "loss": 0.4023,
+      "step": 3170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5624267353278897e-05,
+      "loss": 0.4096,
+      "step": 3171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5621325129436395e-05,
+      "loss": 0.4121,
+      "step": 3172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5618382085795912e-05,
+      "loss": 0.4213,
+      "step": 3173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5615438222584605e-05,
+      "loss": 0.4401,
+      "step": 3174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.561249354002968e-05,
+      "loss": 0.4128,
+      "step": 3175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.560954803835844e-05,
+      "loss": 0.4107,
+      "step": 3176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5606601717798212e-05,
+      "loss": 0.4399,
+      "step": 3177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.560365457857642e-05,
+      "loss": 0.3549,
+      "step": 3178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5600706620920528e-05,
+      "loss": 0.4509,
+      "step": 3179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.559775784505807e-05,
+      "loss": 0.3618,
+      "step": 3180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5594808251216656e-05,
+      "loss": 0.3418,
+      "step": 3181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5591857839623934e-05,
+      "loss": 0.4007,
+      "step": 3182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5588906610507638e-05,
+      "loss": 0.4017,
+      "step": 3183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.558595456409555e-05,
+      "loss": 0.39,
+      "step": 3184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5583001700615528e-05,
+      "loss": 0.3777,
+      "step": 3185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5580048020295482e-05,
+      "loss": 0.4539,
+      "step": 3186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5577093523363396e-05,
+      "loss": 0.3842,
+      "step": 3187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5574138210047295e-05,
+      "loss": 0.3984,
+      "step": 3188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.55711820805753e-05,
+      "loss": 0.3953,
+      "step": 3189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.556822513517556e-05,
+      "loss": 0.3863,
+      "step": 3190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5565267374076323e-05,
+      "loss": 0.4035,
+      "step": 3191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.556230879750587e-05,
+      "loss": 0.4098,
+      "step": 3192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5559349405692554e-05,
+      "loss": 0.485,
+      "step": 3193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5556389198864804e-05,
+      "loss": 0.3961,
+      "step": 3194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.555342817725109e-05,
+      "loss": 0.3527,
+      "step": 3195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.555046634107996e-05,
+      "loss": 0.345,
+      "step": 3196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5547503690580023e-05,
+      "loss": 0.3605,
+      "step": 3197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5544540225979946e-05,
+      "loss": 0.3866,
+      "step": 3198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5541575947508464e-05,
+      "loss": 0.4602,
+      "step": 3199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5538610855394366e-05,
+      "loss": 0.3917,
+      "step": 3200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.553564494986652e-05,
+      "loss": 0.3718,
+      "step": 3201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5532678231153838e-05,
+      "loss": 0.3726,
+      "step": 3202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5529710699485308e-05,
+      "loss": 0.3684,
+      "step": 3203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5526742355089976e-05,
+      "loss": 0.374,
+      "step": 3204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5523773198196952e-05,
+      "loss": 0.4553,
+      "step": 3205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5520803229035408e-05,
+      "loss": 0.405,
+      "step": 3206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.551783244783458e-05,
+      "loss": 0.3845,
+      "step": 3207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5514860854823752e-05,
+      "loss": 0.4022,
+      "step": 3208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5511888450232305e-05,
+      "loss": 0.418,
+      "step": 3209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5508915234289645e-05,
+      "loss": 0.4276,
+      "step": 3210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5505941207225267e-05,
+      "loss": 0.3723,
+      "step": 3211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.550296636926871e-05,
+      "loss": 0.3909,
+      "step": 3212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5499990720649596e-05,
+      "loss": 0.4042,
+      "step": 3213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5497014261597586e-05,
+      "loss": 0.3867,
+      "step": 3214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.549403699234242e-05,
+      "loss": 0.417,
+      "step": 3215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.54910589131139e-05,
+      "loss": 0.3474,
+      "step": 3216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5488080024141885e-05,
+      "loss": 0.3433,
+      "step": 3217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5485100325656295e-05,
+      "loss": 0.4286,
+      "step": 3218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.548211981788711e-05,
+      "loss": 0.4179,
+      "step": 3219
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.547913850106439e-05,
+      "loss": 0.3978,
+      "step": 3220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.547615637541824e-05,
+      "loss": 0.3792,
+      "step": 3221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5473173441178837e-05,
+      "loss": 0.4075,
+      "step": 3222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5470189698576405e-05,
+      "loss": 0.4152,
+      "step": 3223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5467205147841255e-05,
+      "loss": 0.3415,
+      "step": 3224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5464219789203736e-05,
+      "loss": 0.375,
+      "step": 3225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5461233622894277e-05,
+      "loss": 0.3977,
+      "step": 3226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5458246649143356e-05,
+      "loss": 0.3449,
+      "step": 3227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5455258868181536e-05,
+      "loss": 0.3954,
+      "step": 3228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5452270280239405e-05,
+      "loss": 0.4469,
+      "step": 3229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.544928088554765e-05,
+      "loss": 0.4236,
+      "step": 3230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5446290684336997e-05,
+      "loss": 0.4448,
+      "step": 3231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.544329967683825e-05,
+      "loss": 0.4357,
+      "step": 3232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5440307863282258e-05,
+      "loss": 0.4535,
+      "step": 3233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5437315243899948e-05,
+      "loss": 0.3771,
+      "step": 3234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.54343218189223e-05,
+      "loss": 0.3784,
+      "step": 3235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.543132758858036e-05,
+      "loss": 0.3835,
+      "step": 3236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.542833255310524e-05,
+      "loss": 0.3972,
+      "step": 3237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5425336712728098e-05,
+      "loss": 0.3954,
+      "step": 3238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.542234006768017e-05,
+      "loss": 0.4227,
+      "step": 3239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5419342618192757e-05,
+      "loss": 0.4074,
+      "step": 3240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5416344364497203e-05,
+      "loss": 0.3765,
+      "step": 3241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5413345306824935e-05,
+      "loss": 0.3766,
+      "step": 3242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5410345445407426e-05,
+      "loss": 0.3676,
+      "step": 3243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5407344780476222e-05,
+      "loss": 0.4042,
+      "step": 3244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5404343312262925e-05,
+      "loss": 0.4459,
+      "step": 3245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5401341040999202e-05,
+      "loss": 0.3966,
+      "step": 3246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5398337966916785e-05,
+      "loss": 0.3915,
+      "step": 3247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.539533409024745e-05,
+      "loss": 0.3931,
+      "step": 3248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5392329411223064e-05,
+      "loss": 0.3815,
+      "step": 3249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5389323930075533e-05,
+      "loss": 0.4077,
+      "step": 3250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5386317647036838e-05,
+      "loss": 0.4228,
+      "step": 3251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5383310562339003e-05,
+      "loss": 0.364,
+      "step": 3252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.538030267621414e-05,
+      "loss": 0.3727,
+      "step": 3253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5377293988894407e-05,
+      "loss": 0.4107,
+      "step": 3254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5374284500612028e-05,
+      "loss": 0.3941,
+      "step": 3255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5371274211599284e-05,
+      "loss": 0.44,
+      "step": 3256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5368263122088525e-05,
+      "loss": 0.3665,
+      "step": 3257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5365251232312157e-05,
+      "loss": 0.3939,
+      "step": 3258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5362238542502658e-05,
+      "loss": 0.3788,
+      "step": 3259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5359225052892543e-05,
+      "loss": 0.4009,
+      "step": 3260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5356210763714423e-05,
+      "loss": 0.3506,
+      "step": 3261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5353195675200946e-05,
+      "loss": 0.4389,
+      "step": 3262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5350179787584827e-05,
+      "loss": 0.3899,
+      "step": 3263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5347163101098846e-05,
+      "loss": 0.347,
+      "step": 3264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.534414561597585e-05,
+      "loss": 0.3913,
+      "step": 3265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5341127332448733e-05,
+      "loss": 0.4298,
+      "step": 3266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.533810825075046e-05,
+      "loss": 0.4365,
+      "step": 3267
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5335088371114053e-05,
+      "loss": 0.3442,
+      "step": 3268
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5332067693772616e-05,
+      "loss": 0.387,
+      "step": 3269
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5329046218959274e-05,
+      "loss": 0.4086,
+      "step": 3270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5326023946907253e-05,
+      "loss": 0.4113,
+      "step": 3271
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5323000877849816e-05,
+      "loss": 0.4357,
+      "step": 3272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5319977012020304e-05,
+      "loss": 0.3102,
+      "step": 3273
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5316952349652104e-05,
+      "loss": 0.3954,
+      "step": 3274
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.531392689097867e-05,
+      "loss": 0.341,
+      "step": 3275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.531090063623353e-05,
+      "loss": 0.3744,
+      "step": 3276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5307873585650257e-05,
+      "loss": 0.3776,
+      "step": 3277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.530484573946249e-05,
+      "loss": 0.4237,
+      "step": 3278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.530181709790393e-05,
+      "loss": 0.391,
+      "step": 3279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.529878766120835e-05,
+      "loss": 0.3525,
+      "step": 3280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.529575742960956e-05,
+      "loss": 0.3497,
+      "step": 3281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5292726403341454e-05,
+      "loss": 0.3807,
+      "step": 3282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.528969458263798e-05,
+      "loss": 0.3739,
+      "step": 3283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5286661967733143e-05,
+      "loss": 0.3722,
+      "step": 3284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5283628558861015e-05,
+      "loss": 0.3607,
+      "step": 3285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5280594356255725e-05,
+      "loss": 0.418,
+      "step": 3286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5277559360151465e-05,
+      "loss": 0.3935,
+      "step": 3287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.52745235707825e-05,
+      "loss": 0.4409,
+      "step": 3288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5271486988383124e-05,
+      "loss": 0.3964,
+      "step": 3289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.526844961318773e-05,
+      "loss": 0.377,
+      "step": 3290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5265411445430748e-05,
+      "loss": 0.4098,
+      "step": 3291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.5262372485346678e-05,
+      "loss": 0.4543,
+      "step": 3292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.525933273317008e-05,
+      "loss": 0.3813,
+      "step": 3293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.525629218913557e-05,
+      "loss": 0.3988,
+      "step": 3294
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5253250853477842e-05,
+      "loss": 0.4207,
+      "step": 3295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5250208726431623e-05,
+      "loss": 0.4013,
+      "step": 3296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.524716580823173e-05,
+      "loss": 0.3996,
+      "step": 3297
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5244122099113017e-05,
+      "loss": 0.3572,
+      "step": 3298
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5241077599310425e-05,
+      "loss": 0.3693,
+      "step": 3299
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5238032309058926e-05,
+      "loss": 0.3813,
+      "step": 3300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.523498622859358e-05,
+      "loss": 0.3604,
+      "step": 3301
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5231939358149482e-05,
+      "loss": 0.3955,
+      "step": 3302
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5228891697961817e-05,
+      "loss": 0.3612,
+      "step": 3303
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5225843248265807e-05,
+      "loss": 0.3783,
+      "step": 3304
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5222794009296744e-05,
+      "loss": 0.3929,
+      "step": 3305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.521974398128999e-05,
+      "loss": 0.4232,
+      "step": 3306
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.521669316448095e-05,
+      "loss": 0.4394,
+      "step": 3307
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5213641559105097e-05,
+      "loss": 0.4151,
+      "step": 3308
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5210589165397978e-05,
+      "loss": 0.4182,
+      "step": 3309
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5207535983595175e-05,
+      "loss": 0.3967,
+      "step": 3310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5204482013932357e-05,
+      "loss": 0.3756,
+      "step": 3311
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5201427256645233e-05,
+      "loss": 0.3752,
+      "step": 3312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5198371711969586e-05,
+      "loss": 0.3607,
+      "step": 3313
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5195315380141265e-05,
+      "loss": 0.3971,
+      "step": 3314
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5192258261396154e-05,
+      "loss": 0.4151,
+      "step": 3315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5189200355970224e-05,
+      "loss": 0.3794,
+      "step": 3316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.518614166409949e-05,
+      "loss": 0.4693,
+      "step": 3317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5183082186020043e-05,
+      "loss": 0.3945,
+      "step": 3318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.518002192196802e-05,
+      "loss": 0.3931,
+      "step": 3319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5176960872179628e-05,
+      "loss": 0.4191,
+      "step": 3320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.517389903689113e-05,
+      "loss": 0.374,
+      "step": 3321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5170836416338857e-05,
+      "loss": 0.3886,
+      "step": 3322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5167773010759185e-05,
+      "loss": 0.4606,
+      "step": 3323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.516470882038856e-05,
+      "loss": 0.4417,
+      "step": 3324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5161643845463498e-05,
+      "loss": 0.3787,
+      "step": 3325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.515857808622056e-05,
+      "loss": 0.386,
+      "step": 3326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5155511542896382e-05,
+      "loss": 0.3919,
+      "step": 3327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.515244421572764e-05,
+      "loss": 0.4239,
+      "step": 3328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5149376104951097e-05,
+      "loss": 0.3676,
+      "step": 3329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.514630721080355e-05,
+      "loss": 0.4536,
+      "step": 3330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5143237533521874e-05,
+      "loss": 0.3715,
+      "step": 3331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5140167073343e-05,
+      "loss": 0.4099,
+      "step": 3332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5137095830503923e-05,
+      "loss": 0.3583,
+      "step": 3333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5134023805241683e-05,
+      "loss": 0.3911,
+      "step": 3334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5130950997793404e-05,
+      "loss": 0.4015,
+      "step": 3335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.512787740839625e-05,
+      "loss": 0.3638,
+      "step": 3336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.512480303728746e-05,
+      "loss": 0.4575,
+      "step": 3337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.512172788470432e-05,
+      "loss": 0.3772,
+      "step": 3338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5118651950884184e-05,
+      "loss": 0.3491,
+      "step": 3339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.511557523606447e-05,
+      "loss": 0.4069,
+      "step": 3340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.511249774048265e-05,
+      "loss": 0.3645,
+      "step": 3341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5109419464376255e-05,
+      "loss": 0.3503,
+      "step": 3342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5106340407982875e-05,
+      "loss": 0.3486,
+      "step": 3343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.510326057154018e-05,
+      "loss": 0.347,
+      "step": 3344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5100179955285865e-05,
+      "loss": 0.3713,
+      "step": 3345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.509709855945772e-05,
+      "loss": 0.3813,
+      "step": 3346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5094016384293577e-05,
+      "loss": 0.3902,
+      "step": 3347
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5090933430031323e-05,
+      "loss": 0.4244,
+      "step": 3348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5087849696908922e-05,
+      "loss": 0.3871,
+      "step": 3349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.5084765185164386e-05,
+      "loss": 0.3792,
+      "step": 3350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.508167989503579e-05,
+      "loss": 0.4626,
+      "step": 3351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5078593826761272e-05,
+      "loss": 0.3838,
+      "step": 3352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5075506980579022e-05,
+      "loss": 0.413,
+      "step": 3353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.50724193567273e-05,
+      "loss": 0.3863,
+      "step": 3354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.506933095544442e-05,
+      "loss": 0.3676,
+      "step": 3355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5066241776968763e-05,
+      "loss": 0.3666,
+      "step": 3356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5063151821538753e-05,
+      "loss": 0.4569,
+      "step": 3357
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5060061089392897e-05,
+      "loss": 0.4137,
+      "step": 3358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.505696958076975e-05,
+      "loss": 0.3956,
+      "step": 3359
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.505387729590792e-05,
+      "loss": 0.4201,
+      "step": 3360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5050784235046083e-05,
+      "loss": 0.4074,
+      "step": 3361
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5047690398422977e-05,
+      "loss": 0.3283,
+      "step": 3362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.50445957862774e-05,
+      "loss": 0.3717,
+      "step": 3363
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5041500398848205e-05,
+      "loss": 0.397,
+      "step": 3364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5038404236374307e-05,
+      "loss": 0.4086,
+      "step": 3365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.503530729909468e-05,
+      "loss": 0.3568,
+      "step": 3366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5032209587248358e-05,
+      "loss": 0.4716,
+      "step": 3367
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5029111101074437e-05,
+      "loss": 0.402,
+      "step": 3368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5026011840812067e-05,
+      "loss": 0.3727,
+      "step": 3369
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5022911806700464e-05,
+      "loss": 0.3254,
+      "step": 3370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5019810998978902e-05,
+      "loss": 0.3831,
+      "step": 3371
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.501670941788672e-05,
+      "loss": 0.3584,
+      "step": 3372
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5013607063663298e-05,
+      "loss": 0.3811,
+      "step": 3373
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.5010503936548103e-05,
+      "loss": 0.3839,
+      "step": 3374
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.500740003678063e-05,
+      "loss": 0.3842,
+      "step": 3375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.500429536460047e-05,
+      "loss": 0.4662,
+      "step": 3376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.500118992024724e-05,
+      "loss": 0.424,
+      "step": 3377
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.499808370396064e-05,
+      "loss": 0.3629,
+      "step": 3378
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.499497671598042e-05,
+      "loss": 0.393,
+      "step": 3379
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4991868956546385e-05,
+      "loss": 0.3944,
+      "step": 3380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4988760425898402e-05,
+      "loss": 0.3977,
+      "step": 3381
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.498565112427641e-05,
+      "loss": 0.3582,
+      "step": 3382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4982541051920395e-05,
+      "loss": 0.3658,
+      "step": 3383
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4979430209070404e-05,
+      "loss": 0.3692,
+      "step": 3384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4976318595966536e-05,
+      "loss": 0.3598,
+      "step": 3385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4973206212848974e-05,
+      "loss": 0.4344,
+      "step": 3386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4970093059957936e-05,
+      "loss": 0.3909,
+      "step": 3387
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4966979137533708e-05,
+      "loss": 0.3548,
+      "step": 3388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4963864445816637e-05,
+      "loss": 0.4091,
+      "step": 3389
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.496074898504713e-05,
+      "loss": 0.4224,
+      "step": 3390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4957632755465647e-05,
+      "loss": 0.3626,
+      "step": 3391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4954515757312714e-05,
+      "loss": 0.3902,
+      "step": 3392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4951397990828914e-05,
+      "loss": 0.3721,
+      "step": 3393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.494827945625489e-05,
+      "loss": 0.3853,
+      "step": 3394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4945160153831334e-05,
+      "loss": 0.3898,
+      "step": 3395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.494204008379902e-05,
+      "loss": 0.4127,
+      "step": 3396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4938919246398765e-05,
+      "loss": 0.3998,
+      "step": 3397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.493579764187144e-05,
+      "loss": 0.3729,
+      "step": 3398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4932675270457997e-05,
+      "loss": 0.3838,
+      "step": 3399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.492955213239942e-05,
+      "loss": 0.3758,
+      "step": 3400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4926428227936772e-05,
+      "loss": 0.4304,
+      "step": 3401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.492330355731117e-05,
+      "loss": 0.4273,
+      "step": 3402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4920178120763787e-05,
+      "loss": 0.4059,
+      "step": 3403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4917051918535855e-05,
+      "loss": 0.3947,
+      "step": 3404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.491392495086867e-05,
+      "loss": 0.4029,
+      "step": 3405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4910797218003587e-05,
+      "loss": 0.3965,
+      "step": 3406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.490766872018201e-05,
+      "loss": 0.3691,
+      "step": 3407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4904539457645422e-05,
+      "loss": 0.3658,
+      "step": 3408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.490140943063534e-05,
+      "loss": 0.4161,
+      "step": 3409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.489827863939336e-05,
+      "loss": 0.4136,
+      "step": 3410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.489514708416112e-05,
+      "loss": 0.3845,
+      "step": 3411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4892014765180342e-05,
+      "loss": 0.3752,
+      "step": 3412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4888881682692776e-05,
+      "loss": 0.358,
+      "step": 3413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.488574783694025e-05,
+      "loss": 0.4323,
+      "step": 3414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.488261322816466e-05,
+      "loss": 0.4378,
+      "step": 3415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.487947785660793e-05,
+      "loss": 0.4216,
+      "step": 3416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4876341722512073e-05,
+      "loss": 0.3665,
+      "step": 3417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.487320482611914e-05,
+      "loss": 0.3775,
+      "step": 3418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.487006716767126e-05,
+      "loss": 0.3631,
+      "step": 3419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4866928747410604e-05,
+      "loss": 0.4009,
+      "step": 3420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4863789565579407e-05,
+      "loss": 0.3264,
+      "step": 3421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4860649622419967e-05,
+      "loss": 0.3985,
+      "step": 3422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4857508918174644e-05,
+      "loss": 0.3997,
+      "step": 3423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4854367453085835e-05,
+      "loss": 0.4027,
+      "step": 3424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4851225227396022e-05,
+      "loss": 0.401,
+      "step": 3425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4848082241347735e-05,
+      "loss": 0.4521,
+      "step": 3426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4844938495183563e-05,
+      "loss": 0.3475,
+      "step": 3427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4841793989146147e-05,
+      "loss": 0.4369,
+      "step": 3428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.48386487234782e-05,
+      "loss": 0.3917,
+      "step": 3429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4835502698422486e-05,
+      "loss": 0.4052,
+      "step": 3430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4832355914221826e-05,
+      "loss": 0.4048,
+      "step": 3431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4829208371119098e-05,
+      "loss": 0.4023,
+      "step": 3432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.482606006935725e-05,
+      "loss": 0.4086,
+      "step": 3433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.482291100917928e-05,
+      "loss": 0.4198,
+      "step": 3434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4819761190828238e-05,
+      "loss": 0.3455,
+      "step": 3435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4816610614547245e-05,
+      "loss": 0.3673,
+      "step": 3436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.481345928057948e-05,
+      "loss": 0.3836,
+      "step": 3437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.481030718916817e-05,
+      "loss": 0.4422,
+      "step": 3438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.480715434055661e-05,
+      "loss": 0.4123,
+      "step": 3439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4804000734988146e-05,
+      "loss": 0.3932,
+      "step": 3440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4800846372706193e-05,
+      "loss": 0.412,
+      "step": 3441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.479769125395421e-05,
+      "loss": 0.4126,
+      "step": 3442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4794535378975727e-05,
+      "loss": 0.364,
+      "step": 3443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4791378748014322e-05,
+      "loss": 0.4118,
+      "step": 3444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.478822136131365e-05,
+      "loss": 0.4339,
+      "step": 3445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4785063219117398e-05,
+      "loss": 0.3341,
+      "step": 3446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4781904321669325e-05,
+      "loss": 0.3849,
+      "step": 3447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4778744669213255e-05,
+      "loss": 0.3436,
+      "step": 3448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.477558426199306e-05,
+      "loss": 0.3887,
+      "step": 3449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.477242310025267e-05,
+      "loss": 0.383,
+      "step": 3450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4769261184236083e-05,
+      "loss": 0.3388,
+      "step": 3451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4766098514187344e-05,
+      "loss": 0.4344,
+      "step": 3452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4762935090350567e-05,
+      "loss": 0.3949,
+      "step": 3453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.475977091296991e-05,
+      "loss": 0.4239,
+      "step": 3454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.47566059822896e-05,
+      "loss": 0.3683,
+      "step": 3455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4753440298553925e-05,
+      "loss": 0.3576,
+      "step": 3456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4750273862007218e-05,
+      "loss": 0.4399,
+      "step": 3457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.474710667289388e-05,
+      "loss": 0.4043,
+      "step": 3458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.474393873145837e-05,
+      "loss": 0.3412,
+      "step": 3459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4740770037945202e-05,
+      "loss": 0.442,
+      "step": 3460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.473760059259895e-05,
+      "loss": 0.3982,
+      "step": 3461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.473443039566424e-05,
+      "loss": 0.3686,
+      "step": 3462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4731259447385767e-05,
+      "loss": 0.4464,
+      "step": 3463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4728087748008276e-05,
+      "loss": 0.3374,
+      "step": 3464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4724915297776566e-05,
+      "loss": 0.3649,
+      "step": 3465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.472174209693551e-05,
+      "loss": 0.3975,
+      "step": 3466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4718568145730026e-05,
+      "loss": 0.3559,
+      "step": 3467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.471539344440509e-05,
+      "loss": 0.3954,
+      "step": 3468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4712217993205738e-05,
+      "loss": 0.3771,
+      "step": 3469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4709041792377064e-05,
+      "loss": 0.4348,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4705864842164224e-05,
+      "loss": 0.3591,
+      "step": 3471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.470268714281243e-05,
+      "loss": 0.3586,
+      "step": 3472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.469950869456694e-05,
+      "loss": 0.3628,
+      "step": 3473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.469632949767309e-05,
+      "loss": 0.3788,
+      "step": 3474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4693149552376263e-05,
+      "loss": 0.3842,
+      "step": 3475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.468996885892189e-05,
+      "loss": 0.4353,
+      "step": 3476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.468678741755548e-05,
+      "loss": 0.4408,
+      "step": 3477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4683605228522593e-05,
+      "loss": 0.3769,
+      "step": 3478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4680422292068834e-05,
+      "loss": 0.3481,
+      "step": 3479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.467723860843988e-05,
+      "loss": 0.3903,
+      "step": 3480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4674054177881455e-05,
+      "loss": 0.3947,
+      "step": 3481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4670869000639357e-05,
+      "loss": 0.4019,
+      "step": 3482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4667683076959423e-05,
+      "loss": 0.3801,
+      "step": 3483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4664496407087565e-05,
+      "loss": 0.3796,
+      "step": 3484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.466130899126973e-05,
+      "loss": 0.395,
+      "step": 3485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4658120829751945e-05,
+      "loss": 0.3956,
+      "step": 3486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.465493192278029e-05,
+      "loss": 0.3953,
+      "step": 3487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4651742270600882e-05,
+      "loss": 0.3648,
+      "step": 3488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4648551873459927e-05,
+      "loss": 0.5335,
+      "step": 3489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4645360731603665e-05,
+      "loss": 0.3941,
+      "step": 3490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4642168845278405e-05,
+      "loss": 0.4043,
+      "step": 3491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.463897621473051e-05,
+      "loss": 0.3447,
+      "step": 3492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4635782840206397e-05,
+      "loss": 0.4254,
+      "step": 3493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4632588721952553e-05,
+      "loss": 0.406,
+      "step": 3494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.46293938602155e-05,
+      "loss": 0.3778,
+      "step": 3495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4626198255241838e-05,
+      "loss": 0.3587,
+      "step": 3496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.462300190727822e-05,
+      "loss": 0.3776,
+      "step": 3497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.461980481657135e-05,
+      "loss": 0.401,
+      "step": 3498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4616606983367993e-05,
+      "loss": 0.3911,
+      "step": 3499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.461340840791497e-05,
+      "loss": 0.3945,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4610209090459163e-05,
+      "loss": 0.38,
+      "step": 3501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4607009031247497e-05,
+      "loss": 0.4508,
+      "step": 3502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4603808230526984e-05,
+      "loss": 0.3619,
+      "step": 3503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4600606688544666e-05,
+      "loss": 0.4014,
+      "step": 3504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4597404405547652e-05,
+      "loss": 0.4237,
+      "step": 3505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4594201381783105e-05,
+      "loss": 0.37,
+      "step": 3506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4590997617498254e-05,
+      "loss": 0.3425,
+      "step": 3507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.458779311294037e-05,
+      "loss": 0.3715,
+      "step": 3508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4584587868356798e-05,
+      "loss": 0.3845,
+      "step": 3509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.458138188399493e-05,
+      "loss": 0.4041,
+      "step": 3510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4578175160102213e-05,
+      "loss": 0.4106,
+      "step": 3511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4574967696926164e-05,
+      "loss": 0.3832,
+      "step": 3512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4571759494714338e-05,
+      "loss": 0.386,
+      "step": 3513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4568550553714362e-05,
+      "loss": 0.5094,
+      "step": 3514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.456534087417392e-05,
+      "loss": 0.3409,
+      "step": 3515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.456213045634074e-05,
+      "loss": 0.3913,
+      "step": 3516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4558919300462622e-05,
+      "loss": 0.3851,
+      "step": 3517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.455570740678741e-05,
+      "loss": 0.4182,
+      "step": 3518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4552494775563014e-05,
+      "loss": 0.3833,
+      "step": 3519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4549281407037407e-05,
+      "loss": 0.4003,
+      "step": 3520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4546067301458594e-05,
+      "loss": 0.3822,
+      "step": 3521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4542852459074667e-05,
+      "loss": 0.3989,
+      "step": 3522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4539636880133755e-05,
+      "loss": 0.3542,
+      "step": 3523
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.453642056488405e-05,
+      "loss": 0.4121,
+      "step": 3524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4533203513573795e-05,
+      "loss": 0.3346,
+      "step": 3525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.452998572645131e-05,
+      "loss": 0.3695,
+      "step": 3526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.452676720376494e-05,
+      "loss": 0.3612,
+      "step": 3527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4523547945763113e-05,
+      "loss": 0.4452,
+      "step": 3528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4520327952694307e-05,
+      "loss": 0.345,
+      "step": 3529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4517107224807052e-05,
+      "loss": 0.4102,
+      "step": 3530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4513885762349932e-05,
+      "loss": 0.3446,
+      "step": 3531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.45106635655716e-05,
+      "loss": 0.3504,
+      "step": 3532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4507440634720756e-05,
+      "loss": 0.4076,
+      "step": 3533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4504216970046155e-05,
+      "loss": 0.3714,
+      "step": 3534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4500992571796624e-05,
+      "loss": 0.4118,
+      "step": 3535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4497767440221025e-05,
+      "loss": 0.3825,
+      "step": 3536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.449454157556829e-05,
+      "loss": 0.4332,
+      "step": 3537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4491314978087405e-05,
+      "loss": 0.3478,
+      "step": 3538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.448808764802741e-05,
+      "loss": 0.4179,
+      "step": 3539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.448485958563741e-05,
+      "loss": 0.4072,
+      "step": 3540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.448163079116655e-05,
+      "loss": 0.389,
+      "step": 3541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4478401264864054e-05,
+      "loss": 0.4801,
+      "step": 3542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.447517100697918e-05,
+      "loss": 0.3937,
+      "step": 3543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.447194001776126e-05,
+      "loss": 0.4141,
+      "step": 3544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.446870829745967e-05,
+      "loss": 0.4023,
+      "step": 3545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4465475846323848e-05,
+      "loss": 0.3642,
+      "step": 3546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4462242664603293e-05,
+      "loss": 0.3831,
+      "step": 3547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4459008752547553e-05,
+      "loss": 0.3525,
+      "step": 3548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4455774110406228e-05,
+      "loss": 0.362,
+      "step": 3549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4452538738428992e-05,
+      "loss": 0.3927,
+      "step": 3550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4449302636865556e-05,
+      "loss": 0.3905,
+      "step": 3551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4446065805965704e-05,
+      "loss": 0.4221,
+      "step": 3552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4442828245979255e-05,
+      "loss": 0.3932,
+      "step": 3553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.443958995715611e-05,
+      "loss": 0.3627,
+      "step": 3554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.443635093974621e-05,
+      "loss": 0.4357,
+      "step": 3555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4433111193999552e-05,
+      "loss": 0.4321,
+      "step": 3556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4429870720166195e-05,
+      "loss": 0.3568,
+      "step": 3557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.442662951849626e-05,
+      "loss": 0.3827,
+      "step": 3558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4423387589239905e-05,
+      "loss": 0.3956,
+      "step": 3559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.442014493264736e-05,
+      "loss": 0.4043,
+      "step": 3560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.441690154896891e-05,
+      "loss": 0.3794,
+      "step": 3561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4413657438454888e-05,
+      "loss": 0.3855,
+      "step": 3562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4410412601355685e-05,
+      "loss": 0.3747,
+      "step": 3563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.440716703792176e-05,
+      "loss": 0.4169,
+      "step": 3564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.440392074840361e-05,
+      "loss": 0.3976,
+      "step": 3565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4400673733051807e-05,
+      "loss": 0.4115,
+      "step": 3566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.439742599211696e-05,
+      "loss": 0.3607,
+      "step": 3567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4394177525849748e-05,
+      "loss": 0.3663,
+      "step": 3568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4390928334500904e-05,
+      "loss": 0.4105,
+      "step": 3569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4387678418321207e-05,
+      "loss": 0.3917,
+      "step": 3570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.43844277775615e-05,
+      "loss": 0.3812,
+      "step": 3571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4381176412472686e-05,
+      "loss": 0.397,
+      "step": 3572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4377924323305715e-05,
+      "loss": 0.3401,
+      "step": 3573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.43746715103116e-05,
+      "loss": 0.3962,
+      "step": 3574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4371417973741404e-05,
+      "loss": 0.3816,
+      "step": 3575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.436816371384625e-05,
+      "loss": 0.4279,
+      "step": 3576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.436490873087731e-05,
+      "loss": 0.3839,
+      "step": 3577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4361653025085818e-05,
+      "loss": 0.554,
+      "step": 3578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4358396596723073e-05,
+      "loss": 0.3549,
+      "step": 3579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.435513944604041e-05,
+      "loss": 0.4175,
+      "step": 3580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4351881573289233e-05,
+      "loss": 0.3511,
+      "step": 3581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4348622978720993e-05,
+      "loss": 0.3784,
+      "step": 3582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4345363662587214e-05,
+      "loss": 0.4028,
+      "step": 3583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4342103625139453e-05,
+      "loss": 0.3771,
+      "step": 3584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.4338842866629332e-05,
+      "loss": 0.3527,
+      "step": 3585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.433558138730854e-05,
+      "loss": 0.3931,
+      "step": 3586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.43323191874288e-05,
+      "loss": 0.3492,
+      "step": 3587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.432905626724191e-05,
+      "loss": 0.4542,
+      "step": 3588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4325792626999715e-05,
+      "loss": 0.341,
+      "step": 3589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.432252826695411e-05,
+      "loss": 0.3829,
+      "step": 3590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4319263187357062e-05,
+      "loss": 0.41,
+      "step": 3591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4315997388460573e-05,
+      "loss": 0.366,
+      "step": 3592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4312730870516716e-05,
+      "loss": 0.386,
+      "step": 3593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4309463633777617e-05,
+      "loss": 0.3712,
+      "step": 3594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4306195678495448e-05,
+      "loss": 0.36,
+      "step": 3595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4302927004922452e-05,
+      "loss": 0.3579,
+      "step": 3596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4299657613310912e-05,
+      "loss": 0.3882,
+      "step": 3597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4296387503913176e-05,
+      "loss": 0.4218,
+      "step": 3598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4293116676981647e-05,
+      "loss": 0.3989,
+      "step": 3599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.428984513276877e-05,
+      "loss": 0.3229,
+      "step": 3600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4286572871527072e-05,
+      "loss": 0.4219,
+      "step": 3601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.428329989350911e-05,
+      "loss": 0.3652,
+      "step": 3602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4280026198967506e-05,
+      "loss": 0.4438,
+      "step": 3603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4276751788154945e-05,
+      "loss": 0.4057,
+      "step": 3604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.427347666132415e-05,
+      "loss": 0.3395,
+      "step": 3605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.427020081872791e-05,
+      "loss": 0.4372,
+      "step": 3606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.426692426061908e-05,
+      "loss": 0.3655,
+      "step": 3607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.426364698725054e-05,
+      "loss": 0.3693,
+      "step": 3608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.426036899887526e-05,
+      "loss": 0.3922,
+      "step": 3609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.425709029574624e-05,
+      "loss": 0.4276,
+      "step": 3610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4253810878116548e-05,
+      "loss": 0.3595,
+      "step": 3611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4250530746239296e-05,
+      "loss": 0.3641,
+      "step": 3612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4247249900367666e-05,
+      "loss": 0.4209,
+      "step": 3613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4243968340754883e-05,
+      "loss": 0.37,
+      "step": 3614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4240686067654234e-05,
+      "loss": 0.4138,
+      "step": 3615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4237403081319057e-05,
+      "loss": 0.4387,
+      "step": 3616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4234119382002746e-05,
+      "loss": 0.4124,
+      "step": 3617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.423083496995875e-05,
+      "loss": 0.34,
+      "step": 3618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4227549845440578e-05,
+      "loss": 0.4175,
+      "step": 3619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4224264008701786e-05,
+      "loss": 0.3728,
+      "step": 3620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4220977459995988e-05,
+      "loss": 0.4084,
+      "step": 3621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4217690199576856e-05,
+      "loss": 0.3714,
+      "step": 3622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.421440222769811e-05,
+      "loss": 0.4012,
+      "step": 3623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4211113544613533e-05,
+      "loss": 0.3855,
+      "step": 3624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4207824150576964e-05,
+      "loss": 0.3754,
+      "step": 3625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.420453404584228e-05,
+      "loss": 0.3665,
+      "step": 3626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4201243230663432e-05,
+      "loss": 0.4257,
+      "step": 3627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.419795170529442e-05,
+      "loss": 0.4162,
+      "step": 3628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4194659469989294e-05,
+      "loss": 0.378,
+      "step": 3629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4191366525002165e-05,
+      "loss": 0.4109,
+      "step": 3630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4188072870587195e-05,
+      "loss": 0.3703,
+      "step": 3631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4184778506998603e-05,
+      "loss": 0.3784,
+      "step": 3632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4181483434490662e-05,
+      "loss": 0.3982,
+      "step": 3633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4178187653317698e-05,
+      "loss": 0.3652,
+      "step": 3634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4174891163734094e-05,
+      "loss": 0.3615,
+      "step": 3635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4171593965994285e-05,
+      "loss": 0.4304,
+      "step": 3636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4168296060352764e-05,
+      "loss": 0.3867,
+      "step": 3637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4164997447064076e-05,
+      "loss": 0.3419,
+      "step": 3638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4161698126382822e-05,
+      "loss": 0.4197,
+      "step": 3639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4158398098563658e-05,
+      "loss": 0.3339,
+      "step": 3640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4155097363861292e-05,
+      "loss": 0.4258,
+      "step": 3641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4151795922530496e-05,
+      "loss": 0.4508,
+      "step": 3642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4148493774826077e-05,
+      "loss": 0.3954,
+      "step": 3643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4145190921002918e-05,
+      "loss": 0.3943,
+      "step": 3644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.414188736131594e-05,
+      "loss": 0.3798,
+      "step": 3645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.413858309602013e-05,
+      "loss": 0.3463,
+      "step": 3646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.413527812537052e-05,
+      "loss": 0.3939,
+      "step": 3647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.413197244962221e-05,
+      "loss": 0.4103,
+      "step": 3648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4128666069030338e-05,
+      "loss": 0.4281,
+      "step": 3649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4125358983850106e-05,
+      "loss": 0.3355,
+      "step": 3650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4122051194336778e-05,
+      "loss": 0.4272,
+      "step": 3651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4118742700745646e-05,
+      "loss": 0.3827,
+      "step": 3652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4115433503332083e-05,
+      "loss": 0.4002,
+      "step": 3653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.411212360235151e-05,
+      "loss": 0.4079,
+      "step": 3654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.410881299805939e-05,
+      "loss": 0.3437,
+      "step": 3655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4105501690711256e-05,
+      "loss": 0.399,
+      "step": 3656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.410218968056268e-05,
+      "loss": 0.3854,
+      "step": 3657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4098876967869307e-05,
+      "loss": 0.3686,
+      "step": 3658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.409556355288682e-05,
+      "loss": 0.3673,
+      "step": 3659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.409224943587097e-05,
+      "loss": 0.3714,
+      "step": 3660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.408893461707754e-05,
+      "loss": 0.3758,
+      "step": 3661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4085619096762395e-05,
+      "loss": 0.3641,
+      "step": 3662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.408230287518143e-05,
+      "loss": 0.3724,
+      "step": 3663
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4078985952590613e-05,
+      "loss": 0.4497,
+      "step": 3664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.407566832924595e-05,
+      "loss": 0.4103,
+      "step": 3665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.407235000540352e-05,
+      "loss": 0.3934,
+      "step": 3666
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4069030981319438e-05,
+      "loss": 0.4164,
+      "step": 3667
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.406571125724988e-05,
+      "loss": 0.3915,
+      "step": 3668
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4062390833451073e-05,
+      "loss": 0.4397,
+      "step": 3669
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.405906971017931e-05,
+      "loss": 0.397,
+      "step": 3670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.405574788769092e-05,
+      "loss": 0.3774,
+      "step": 3671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4052425366242303e-05,
+      "loss": 0.3968,
+      "step": 3672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4049102146089895e-05,
+      "loss": 0.3722,
+      "step": 3673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4045778227490206e-05,
+      "loss": 0.3898,
+      "step": 3674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4042453610699788e-05,
+      "loss": 0.401,
+      "step": 3675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4039128295975245e-05,
+      "loss": 0.4366,
+      "step": 3676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.403580228357324e-05,
+      "loss": 0.4007,
+      "step": 3677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4032475573750485e-05,
+      "loss": 0.3882,
+      "step": 3678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4029148166763757e-05,
+      "loss": 0.4006,
+      "step": 3679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4025820062869876e-05,
+      "loss": 0.4214,
+      "step": 3680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.402249126232572e-05,
+      "loss": 0.4178,
+      "step": 3681
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.401916176538821e-05,
+      "loss": 0.3629,
+      "step": 3682
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.401583157231434e-05,
+      "loss": 0.3956,
+      "step": 3683
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4012500683361144e-05,
+      "loss": 0.3476,
+      "step": 3684
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.400916909878572e-05,
+      "loss": 0.4461,
+      "step": 3685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.4005836818845214e-05,
+      "loss": 0.4082,
+      "step": 3686
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.400250384379681e-05,
+      "loss": 0.4007,
+      "step": 3687
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3999170173897775e-05,
+      "loss": 0.3463,
+      "step": 3688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.399583580940542e-05,
+      "loss": 0.3802,
+      "step": 3689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3992500750577085e-05,
+      "loss": 0.3644,
+      "step": 3690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3989164997670204e-05,
+      "loss": 0.3528,
+      "step": 3691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3985828550942236e-05,
+      "loss": 0.3409,
+      "step": 3692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.39824914106507e-05,
+      "loss": 0.3943,
+      "step": 3693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.397915357705317e-05,
+      "loss": 0.4005,
+      "step": 3694
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3975815050407276e-05,
+      "loss": 0.4065,
+      "step": 3695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3972475830970705e-05,
+      "loss": 0.3929,
+      "step": 3696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.396913591900118e-05,
+      "loss": 0.3972,
+      "step": 3697
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3965795314756498e-05,
+      "loss": 0.3897,
+      "step": 3698
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3962454018494497e-05,
+      "loss": 0.3517,
+      "step": 3699
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3959112030473076e-05,
+      "loss": 0.353,
+      "step": 3700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3955769350950182e-05,
+      "loss": 0.3617,
+      "step": 3701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.395242598018381e-05,
+      "loss": 0.3967,
+      "step": 3702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3949081918432023e-05,
+      "loss": 0.3837,
+      "step": 3703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.394573716595293e-05,
+      "loss": 0.3657,
+      "step": 3704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.394239172300469e-05,
+      "loss": 0.3786,
+      "step": 3705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3939045589845513e-05,
+      "loss": 0.3364,
+      "step": 3706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3935698766733678e-05,
+      "loss": 0.379,
+      "step": 3707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3932351253927505e-05,
+      "loss": 0.4024,
+      "step": 3708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.392900305168536e-05,
+      "loss": 0.3597,
+      "step": 3709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.392565416026568e-05,
+      "loss": 0.3999,
+      "step": 3710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.392230457992694e-05,
+      "loss": 0.381,
+      "step": 3711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.391895431092768e-05,
+      "loss": 0.4401,
+      "step": 3712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3915603353526483e-05,
+      "loss": 0.4083,
+      "step": 3713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3912251707981997e-05,
+      "loss": 0.3865,
+      "step": 3714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3908899374552912e-05,
+      "loss": 0.3598,
+      "step": 3715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3905546353497968e-05,
+      "loss": 0.3721,
+      "step": 3716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.390219264507597e-05,
+      "loss": 0.4148,
+      "step": 3717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.389883824954578e-05,
+      "loss": 0.3825,
+      "step": 3718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3895483167166294e-05,
+      "loss": 0.38,
+      "step": 3719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3892127398196472e-05,
+      "loss": 0.3536,
+      "step": 3720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.388877094289533e-05,
+      "loss": 0.3763,
+      "step": 3721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.388541380152193e-05,
+      "loss": 0.3523,
+      "step": 3722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.388205597433539e-05,
+      "loss": 0.4115,
+      "step": 3723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3878697461594885e-05,
+      "loss": 0.3572,
+      "step": 3724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3875338263559633e-05,
+      "loss": 0.3205,
+      "step": 3725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.387197838048892e-05,
+      "loss": 0.3543,
+      "step": 3726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.386861781264206e-05,
+      "loss": 0.4095,
+      "step": 3727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3865256560278454e-05,
+      "loss": 0.3997,
+      "step": 3728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3861894623657524e-05,
+      "loss": 0.4228,
+      "step": 3729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3858532003038768e-05,
+      "loss": 0.3771,
+      "step": 3730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3855168698681714e-05,
+      "loss": 0.4225,
+      "step": 3731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.385180471084597e-05,
+      "loss": 0.3648,
+      "step": 3732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3848440039791178e-05,
+      "loss": 0.3746,
+      "step": 3733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3845074685777033e-05,
+      "loss": 0.3825,
+      "step": 3734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.384170864906329e-05,
+      "loss": 0.3594,
+      "step": 3735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.383834192990975e-05,
+      "loss": 0.3688,
+      "step": 3736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.383497452857628e-05,
+      "loss": 0.3925,
+      "step": 3737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.383160644532278e-05,
+      "loss": 0.4096,
+      "step": 3738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3828237680409215e-05,
+      "loss": 0.3595,
+      "step": 3739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3824868234095606e-05,
+      "loss": 0.339,
+      "step": 3740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3821498106642014e-05,
+      "loss": 0.3666,
+      "step": 3741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3818127298308564e-05,
+      "loss": 0.3799,
+      "step": 3742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3814755809355426e-05,
+      "loss": 0.3917,
+      "step": 3743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3811383640042822e-05,
+      "loss": 0.376,
+      "step": 3744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3808010790631043e-05,
+      "loss": 0.5044,
+      "step": 3745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3804637261380402e-05,
+      "loss": 0.4073,
+      "step": 3746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3801263052551297e-05,
+      "loss": 0.3615,
+      "step": 3747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3797888164404157e-05,
+      "loss": 0.3539,
+      "step": 3748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.379451259719947e-05,
+      "loss": 0.3965,
+      "step": 3749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3791136351197775e-05,
+      "loss": 0.3722,
+      "step": 3750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.378775942665967e-05,
+      "loss": 0.3738,
+      "step": 3751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3784381823845793e-05,
+      "loss": 0.3996,
+      "step": 3752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3781003543016847e-05,
+      "loss": 0.3949,
+      "step": 3753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3777624584433577e-05,
+      "loss": 0.3802,
+      "step": 3754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3774244948356787e-05,
+      "loss": 0.3626,
+      "step": 3755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3770864635047338e-05,
+      "loss": 0.3962,
+      "step": 3756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3767483644766124e-05,
+      "loss": 0.4055,
+      "step": 3757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3764101977774113e-05,
+      "loss": 0.3975,
+      "step": 3758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.376071963433231e-05,
+      "loss": 0.4244,
+      "step": 3759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.375733661470179e-05,
+      "loss": 0.3835,
+      "step": 3760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.375395291914366e-05,
+      "loss": 0.3676,
+      "step": 3761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.375056854791908e-05,
+      "loss": 0.4405,
+      "step": 3762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3747183501289286e-05,
+      "loss": 0.3799,
+      "step": 3763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.374379777951554e-05,
+      "loss": 0.3938,
+      "step": 3764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.374041138285916e-05,
+      "loss": 0.4752,
+      "step": 3765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.373702431158154e-05,
+      "loss": 0.4361,
+      "step": 3766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3733636565944096e-05,
+      "loss": 0.367,
+      "step": 3767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3730248146208305e-05,
+      "loss": 0.4142,
+      "step": 3768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.372685905263571e-05,
+      "loss": 0.373,
+      "step": 3769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.372346928548789e-05,
+      "loss": 0.3809,
+      "step": 3770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3720078845026482e-05,
+      "loss": 0.4538,
+      "step": 3771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.371668773151317e-05,
+      "loss": 0.3868,
+      "step": 3772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.37132959452097e-05,
+      "loss": 0.4056,
+      "step": 3773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3709903486377865e-05,
+      "loss": 0.4113,
+      "step": 3774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3706510355279504e-05,
+      "loss": 0.4027,
+      "step": 3775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3703116552176517e-05,
+      "loss": 0.4059,
+      "step": 3776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3699722077330845e-05,
+      "loss": 0.3956,
+      "step": 3777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.36963269310045e-05,
+      "loss": 0.3853,
+      "step": 3778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.369293111345952e-05,
+      "loss": 0.4056,
+      "step": 3779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3689534624958016e-05,
+      "loss": 0.3908,
+      "step": 3780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.368613746576214e-05,
+      "loss": 0.4319,
+      "step": 3781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.368273963613411e-05,
+      "loss": 0.4122,
+      "step": 3782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3679341136336164e-05,
+      "loss": 0.3597,
+      "step": 3783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.367594196663063e-05,
+      "loss": 0.4233,
+      "step": 3784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3672542127279855e-05,
+      "loss": 0.4228,
+      "step": 3785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3669141618546273e-05,
+      "loss": 0.4188,
+      "step": 3786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.366574044069233e-05,
+      "loss": 0.3897,
+      "step": 3787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.366233859398055e-05,
+      "loss": 0.4123,
+      "step": 3788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.36589360786735e-05,
+      "loss": 0.3912,
+      "step": 3789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3655532895033815e-05,
+      "loss": 0.3592,
+      "step": 3790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.365212904332414e-05,
+      "loss": 0.4191,
+      "step": 3791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3648724523807216e-05,
+      "loss": 0.3343,
+      "step": 3792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.364531933674582e-05,
+      "loss": 0.3444,
+      "step": 3793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3641913482402767e-05,
+      "loss": 0.3925,
+      "step": 3794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.363850696104095e-05,
+      "loss": 0.3949,
+      "step": 3795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.363509977292328e-05,
+      "loss": 0.3797,
+      "step": 3796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3631691918312753e-05,
+      "loss": 0.3833,
+      "step": 3797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.362828339747239e-05,
+      "loss": 0.3759,
+      "step": 3798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3624874210665287e-05,
+      "loss": 0.387,
+      "step": 3799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3621464358154572e-05,
+      "loss": 0.3395,
+      "step": 3800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3618053840203434e-05,
+      "loss": 0.3665,
+      "step": 3801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3614642657075106e-05,
+      "loss": 0.3871,
+      "step": 3802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.361123080903288e-05,
+      "loss": 0.3735,
+      "step": 3803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.36078182963401e-05,
+      "loss": 0.3668,
+      "step": 3804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3604405119260153e-05,
+      "loss": 0.3845,
+      "step": 3805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3600991278056487e-05,
+      "loss": 0.3611,
+      "step": 3806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.359757677299259e-05,
+      "loss": 0.3791,
+      "step": 3807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3594161604332016e-05,
+      "loss": 0.3797,
+      "step": 3808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3590745772338354e-05,
+      "loss": 0.382,
+      "step": 3809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3587329277275252e-05,
+      "loss": 0.351,
+      "step": 3810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.358391211940642e-05,
+      "loss": 0.3463,
+      "step": 3811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3580494298995603e-05,
+      "loss": 0.3794,
+      "step": 3812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3577075816306593e-05,
+      "loss": 0.4177,
+      "step": 3813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3573656671603253e-05,
+      "loss": 0.3782,
+      "step": 3814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3570236865149483e-05,
+      "loss": 0.3575,
+      "step": 3815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3566816397209246e-05,
+      "loss": 0.3546,
+      "step": 3816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3563395268046536e-05,
+      "loss": 0.3861,
+      "step": 3817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3559973477925412e-05,
+      "loss": 0.3938,
+      "step": 3818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3556551027109987e-05,
+      "loss": 0.4048,
+      "step": 3819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.355312791586442e-05,
+      "loss": 0.3881,
+      "step": 3820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.354970414445292e-05,
+      "loss": 0.3848,
+      "step": 3821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3546279713139743e-05,
+      "loss": 0.3391,
+      "step": 3822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.354285462218921e-05,
+      "loss": 0.3464,
+      "step": 3823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3539428871865673e-05,
+      "loss": 0.3778,
+      "step": 3824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3536002462433553e-05,
+      "loss": 0.364,
+      "step": 3825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.353257539415731e-05,
+      "loss": 0.3743,
+      "step": 3826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3529147667301467e-05,
+      "loss": 0.3801,
+      "step": 3827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.352571928213058e-05,
+      "loss": 0.3168,
+      "step": 3828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3522290238909274e-05,
+      "loss": 0.3769,
+      "step": 3829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3518860537902212e-05,
+      "loss": 0.3569,
+      "step": 3830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3515430179374118e-05,
+      "loss": 0.3842,
+      "step": 3831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3511999163589755e-05,
+      "loss": 0.4438,
+      "step": 3832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3508567490813943e-05,
+      "loss": 0.3216,
+      "step": 3833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.350513516131156e-05,
+      "loss": 0.3401,
+      "step": 3834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3501702175347524e-05,
+      "loss": 0.3614,
+      "step": 3835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3498268533186804e-05,
+      "loss": 0.3885,
+      "step": 3836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3494834235094422e-05,
+      "loss": 0.3645,
+      "step": 3837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.349139928133546e-05,
+      "loss": 0.3787,
+      "step": 3838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.348796367217503e-05,
+      "loss": 0.3619,
+      "step": 3839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3484527407878317e-05,
+      "loss": 0.3797,
+      "step": 3840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.348109048871054e-05,
+      "loss": 0.3957,
+      "step": 3841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3477652914936983e-05,
+      "loss": 0.3688,
+      "step": 3842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3474214686822955e-05,
+      "loss": 0.3793,
+      "step": 3843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.347077580463385e-05,
+      "loss": 0.4019,
+      "step": 3844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3467336268635086e-05,
+      "loss": 0.3915,
+      "step": 3845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.346389607909215e-05,
+      "loss": 0.3623,
+      "step": 3846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3460455236270557e-05,
+      "loss": 0.3783,
+      "step": 3847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3457013740435895e-05,
+      "loss": 0.4322,
+      "step": 3848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3453571591853788e-05,
+      "loss": 0.36,
+      "step": 3849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3450128790789923e-05,
+      "loss": 0.3498,
+      "step": 3850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3446685337510018e-05,
+      "loss": 0.4276,
+      "step": 3851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.344324123227986e-05,
+      "loss": 0.3981,
+      "step": 3852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3439796475365283e-05,
+      "loss": 0.3697,
+      "step": 3853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.343635106703216e-05,
+      "loss": 0.3817,
+      "step": 3854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.343290500754642e-05,
+      "loss": 0.3435,
+      "step": 3855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3429458297174054e-05,
+      "loss": 0.3609,
+      "step": 3856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.342601093618109e-05,
+      "loss": 0.3695,
+      "step": 3857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3422562924833606e-05,
+      "loss": 0.4237,
+      "step": 3858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.341911426339774e-05,
+      "loss": 0.3803,
+      "step": 3859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3415664952139665e-05,
+      "loss": 0.3709,
+      "step": 3860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3412214991325624e-05,
+      "loss": 0.3745,
+      "step": 3861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.340876438122189e-05,
+      "loss": 0.3854,
+      "step": 3862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.34053131220948e-05,
+      "loss": 0.3873,
+      "step": 3863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3401861214210733e-05,
+      "loss": 0.3844,
+      "step": 3864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3398408657836126e-05,
+      "loss": 0.3996,
+      "step": 3865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.339495545323746e-05,
+      "loss": 0.3582,
+      "step": 3866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3391501600681266e-05,
+      "loss": 0.4275,
+      "step": 3867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.338804710043413e-05,
+      "loss": 0.4208,
+      "step": 3868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3384591952762685e-05,
+      "loss": 0.397,
+      "step": 3869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3381136157933603e-05,
+      "loss": 0.4163,
+      "step": 3870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3377679716213632e-05,
+      "loss": 0.34,
+      "step": 3871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.337422262786955e-05,
+      "loss": 0.3574,
+      "step": 3872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.337076489316818e-05,
+      "loss": 0.4127,
+      "step": 3873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3367306512376415e-05,
+      "loss": 0.4006,
+      "step": 3874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3363847485761184e-05,
+      "loss": 0.4018,
+      "step": 3875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3360387813589466e-05,
+      "loss": 0.4384,
+      "step": 3876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.33569274961283e-05,
+      "loss": 0.3582,
+      "step": 3877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.335346653364476e-05,
+      "loss": 0.376,
+      "step": 3878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.335000492640598e-05,
+      "loss": 0.3567,
+      "step": 3879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3346542674679143e-05,
+      "loss": 0.4786,
+      "step": 3880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.334307977873148e-05,
+      "loss": 0.3949,
+      "step": 3881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3339616238830267e-05,
+      "loss": 0.3771,
+      "step": 3882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.333615205524284e-05,
+      "loss": 0.44,
+      "step": 3883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3332687228236577e-05,
+      "loss": 0.3458,
+      "step": 3884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.332922175807891e-05,
+      "loss": 0.4105,
+      "step": 3885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.332575564503731e-05,
+      "loss": 0.3958,
+      "step": 3886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3322288889379315e-05,
+      "loss": 0.4078,
+      "step": 3887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3318821491372505e-05,
+      "loss": 0.3583,
+      "step": 3888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.33153534512845e-05,
+      "loss": 0.4116,
+      "step": 3889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3311884769382976e-05,
+      "loss": 0.4118,
+      "step": 3890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.330841544593567e-05,
+      "loss": 0.4189,
+      "step": 3891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.330494548121036e-05,
+      "loss": 0.401,
+      "step": 3892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3301474875474863e-05,
+      "loss": 0.4058,
+      "step": 3893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3298003628997056e-05,
+      "loss": 0.3839,
+      "step": 3894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.329453174204487e-05,
+      "loss": 0.3866,
+      "step": 3895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3291059214886274e-05,
+      "loss": 0.3533,
+      "step": 3896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3287586047789296e-05,
+      "loss": 0.4097,
+      "step": 3897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.328411224102201e-05,
+      "loss": 0.3775,
+      "step": 3898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3280637794852536e-05,
+      "loss": 0.376,
+      "step": 3899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3277162709549048e-05,
+      "loss": 0.4003,
+      "step": 3900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3273686985379766e-05,
+      "loss": 0.3881,
+      "step": 3901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.327021062261296e-05,
+      "loss": 0.3645,
+      "step": 3902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3266733621516953e-05,
+      "loss": 0.3913,
+      "step": 3903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3263255982360113e-05,
+      "loss": 0.4307,
+      "step": 3904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3259777705410855e-05,
+      "loss": 0.3892,
+      "step": 3905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3256298790937656e-05,
+      "loss": 0.3689,
+      "step": 3906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3252819239209024e-05,
+      "loss": 0.4083,
+      "step": 3907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3249339050493526e-05,
+      "loss": 0.3381,
+      "step": 3908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3245858225059784e-05,
+      "loss": 0.3605,
+      "step": 3909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.324237676317646e-05,
+      "loss": 0.3849,
+      "step": 3910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.323889466511226e-05,
+      "loss": 0.3581,
+      "step": 3911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.323541193113596e-05,
+      "loss": 0.3764,
+      "step": 3912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3231928561516366e-05,
+      "loss": 0.3606,
+      "step": 3913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3228444556522334e-05,
+      "loss": 0.4198,
+      "step": 3914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3224959916422777e-05,
+      "loss": 0.4066,
+      "step": 3915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3221474641486654e-05,
+      "loss": 0.4027,
+      "step": 3916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3217988731982984e-05,
+      "loss": 0.3267,
+      "step": 3917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3214502188180802e-05,
+      "loss": 0.3701,
+      "step": 3918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.321101501034923e-05,
+      "loss": 0.4223,
+      "step": 3919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3207527198757418e-05,
+      "loss": 0.4267,
+      "step": 3920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3204038753674576e-05,
+      "loss": 0.3912,
+      "step": 3921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3200549675369943e-05,
+      "loss": 0.3588,
+      "step": 3922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3197059964112836e-05,
+      "loss": 0.4039,
+      "step": 3923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3193569620172594e-05,
+      "loss": 0.3945,
+      "step": 3924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3190078643818623e-05,
+      "loss": 0.3288,
+      "step": 3925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3186587035320366e-05,
+      "loss": 0.3577,
+      "step": 3926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3183094794947325e-05,
+      "loss": 0.3878,
+      "step": 3927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.317960192296904e-05,
+      "loss": 0.3512,
+      "step": 3928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.317610841965511e-05,
+      "loss": 0.4309,
+      "step": 3929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3172614285275175e-05,
+      "loss": 0.3925,
+      "step": 3930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3169119520098928e-05,
+      "loss": 0.4295,
+      "step": 3931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3165624124396112e-05,
+      "loss": 0.3603,
+      "step": 3932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3162128098436514e-05,
+      "loss": 0.3631,
+      "step": 3933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3158631442489968e-05,
+      "loss": 0.3677,
+      "step": 3934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3155134156826366e-05,
+      "loss": 0.3351,
+      "step": 3935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3151636241715643e-05,
+      "loss": 0.4335,
+      "step": 3936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.314813769742778e-05,
+      "loss": 0.3524,
+      "step": 3937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3144638524232803e-05,
+      "loss": 0.3802,
+      "step": 3938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.314113872240081e-05,
+      "loss": 0.4031,
+      "step": 3939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3137638292201915e-05,
+      "loss": 0.3878,
+      "step": 3940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3134137233906302e-05,
+      "loss": 0.4063,
+      "step": 3941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.313063554778419e-05,
+      "loss": 0.3469,
+      "step": 3942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3127133234105867e-05,
+      "loss": 0.3559,
+      "step": 3943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.312363029314165e-05,
+      "loss": 0.3536,
+      "step": 3944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3120126725161905e-05,
+      "loss": 0.3483,
+      "step": 3945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.311662253043705e-05,
+      "loss": 0.3786,
+      "step": 3946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.311311770923757e-05,
+      "loss": 0.3919,
+      "step": 3947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3109612261833967e-05,
+      "loss": 0.3802,
+      "step": 3948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.310610618849681e-05,
+      "loss": 0.4035,
+      "step": 3949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.310259948949671e-05,
+      "loss": 0.3692,
+      "step": 3950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3099092165104328e-05,
+      "loss": 0.416,
+      "step": 3951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3095584215590386e-05,
+      "loss": 0.419,
+      "step": 3952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3092075641225625e-05,
+      "loss": 0.4051,
+      "step": 3953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3088566442280867e-05,
+      "loss": 0.3708,
+      "step": 3954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.308505661902695e-05,
+      "loss": 0.42,
+      "step": 3955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.308154617173479e-05,
+      "loss": 0.3807,
+      "step": 3956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3078035100675335e-05,
+      "loss": 0.4066,
+      "step": 3957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3074523406119583e-05,
+      "loss": 0.356,
+      "step": 3958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3071011088338577e-05,
+      "loss": 0.3863,
+      "step": 3959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3067498147603416e-05,
+      "loss": 0.397,
+      "step": 3960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3063984584185245e-05,
+      "loss": 0.376,
+      "step": 3961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3060470398355253e-05,
+      "loss": 0.4067,
+      "step": 3962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.305695559038468e-05,
+      "loss": 0.3648,
+      "step": 3963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3053440160544813e-05,
+      "loss": 0.3947,
+      "step": 3964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.304992410910699e-05,
+      "loss": 0.3966,
+      "step": 3965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.304640743634259e-05,
+      "loss": 0.371,
+      "step": 3966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.304289014252305e-05,
+      "loss": 0.3796,
+      "step": 3967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.303937222791984e-05,
+      "loss": 0.3756,
+      "step": 3968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3035853692804503e-05,
+      "loss": 0.3502,
+      "step": 3969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3032334537448598e-05,
+      "loss": 0.39,
+      "step": 3970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3028814762123756e-05,
+      "loss": 0.3746,
+      "step": 3971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3025294367101648e-05,
+      "loss": 0.3219,
+      "step": 3972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3021773352653992e-05,
+      "loss": 0.4229,
+      "step": 3973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3018251719052553e-05,
+      "loss": 0.4009,
+      "step": 3974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3014729466569143e-05,
+      "loss": 0.3034,
+      "step": 3975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3011206595475628e-05,
+      "loss": 0.3661,
+      "step": 3976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.300768310604392e-05,
+      "loss": 0.3824,
+      "step": 3977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3004158998545973e-05,
+      "loss": 0.3968,
+      "step": 3978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3000634273253792e-05,
+      "loss": 0.3727,
+      "step": 3979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2997108930439427e-05,
+      "loss": 0.3699,
+      "step": 3980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2993582970374985e-05,
+      "loss": 0.3639,
+      "step": 3981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2990056393332615e-05,
+      "loss": 0.3579,
+      "step": 3982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2986529199584507e-05,
+      "loss": 0.3569,
+      "step": 3983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.298300138940291e-05,
+      "loss": 0.4266,
+      "step": 3984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.297947296306011e-05,
+      "loss": 0.3227,
+      "step": 3985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2975943920828445e-05,
+      "loss": 0.3116,
+      "step": 3986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.29724142629803e-05,
+      "loss": 0.3562,
+      "step": 3987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2968883989788122e-05,
+      "loss": 0.359,
+      "step": 3988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2965353101524376e-05,
+      "loss": 0.3448,
+      "step": 3989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.29618215984616e-05,
+      "loss": 0.4429,
+      "step": 3990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2958289480872368e-05,
+      "loss": 0.425,
+      "step": 3991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.29547567490293e-05,
+      "loss": 0.4315,
+      "step": 3992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2951223403205072e-05,
+      "loss": 0.3953,
+      "step": 3993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2947689443672396e-05,
+      "loss": 0.4364,
+      "step": 3994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2944154870704046e-05,
+      "loss": 0.4039,
+      "step": 3995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2940619684572828e-05,
+      "loss": 0.4167,
+      "step": 3996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2937083885551606e-05,
+      "loss": 0.4145,
+      "step": 3997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2933547473913286e-05,
+      "loss": 0.3345,
+      "step": 3998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2930010449930826e-05,
+      "loss": 0.3402,
+      "step": 3999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2926472813877223e-05,
+      "loss": 0.4103,
+      "step": 4000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292293456602553e-05,
+      "loss": 0.3761,
+      "step": 4001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2919395706648843e-05,
+      "loss": 0.3808,
+      "step": 4002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2915856236020308e-05,
+      "loss": 0.35,
+      "step": 4003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2912316154413108e-05,
+      "loss": 0.3483,
+      "step": 4004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.290877546210049e-05,
+      "loss": 0.427,
+      "step": 4005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2905234159355736e-05,
+      "loss": 0.3423,
+      "step": 4006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.290169224645218e-05,
+      "loss": 0.4061,
+      "step": 4007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.28981497236632e-05,
+      "loss": 0.3589,
+      "step": 4008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.289460659126222e-05,
+      "loss": 0.393,
+      "step": 4009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.289106284952272e-05,
+      "loss": 0.3585,
+      "step": 4010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2887518498718217e-05,
+      "loss": 0.3604,
+      "step": 4011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2883973539122277e-05,
+      "loss": 0.326,
+      "step": 4012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2880427971008517e-05,
+      "loss": 0.3391,
+      "step": 4013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.28768817946506e-05,
+      "loss": 0.3554,
+      "step": 4014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2873335010322237e-05,
+      "loss": 0.3711,
+      "step": 4015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.286978761829717e-05,
+      "loss": 0.4426,
+      "step": 4016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2866239618849215e-05,
+      "loss": 0.384,
+      "step": 4017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.286269101225222e-05,
+      "loss": 0.3995,
+      "step": 4018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2859141798780077e-05,
+      "loss": 0.3866,
+      "step": 4019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.285559197870673e-05,
+      "loss": 0.3683,
+      "step": 4020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2852041552306173e-05,
+      "loss": 0.374,
+      "step": 4021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2848490519852434e-05,
+      "loss": 0.3734,
+      "step": 4022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2844938881619606e-05,
+      "loss": 0.4241,
+      "step": 4023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2841386637881806e-05,
+      "loss": 0.4315,
+      "step": 4024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2837833788913226e-05,
+      "loss": 0.3416,
+      "step": 4025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2834280334988088e-05,
+      "loss": 0.3859,
+      "step": 4026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.283072627638065e-05,
+      "loss": 0.3469,
+      "step": 4027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2827171613365236e-05,
+      "loss": 0.3756,
+      "step": 4028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2823616346216216e-05,
+      "loss": 0.3818,
+      "step": 4029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2820060475207993e-05,
+      "loss": 0.4133,
+      "step": 4030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2816504000615022e-05,
+      "loss": 0.3923,
+      "step": 4031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2812946922711816e-05,
+      "loss": 0.4645,
+      "step": 4032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2809389241772912e-05,
+      "loss": 0.3882,
+      "step": 4033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.280583095807292e-05,
+      "loss": 0.335,
+      "step": 4034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.280227207188647e-05,
+      "loss": 0.3678,
+      "step": 4035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.279871258348826e-05,
+      "loss": 0.3474,
+      "step": 4036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2795152493153028e-05,
+      "loss": 0.4165,
+      "step": 4037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2791591801155548e-05,
+      "loss": 0.339,
+      "step": 4038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278803050777066e-05,
+      "loss": 0.3163,
+      "step": 4039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2784468613273227e-05,
+      "loss": 0.3396,
+      "step": 4040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2780906117938185e-05,
+      "loss": 0.4296,
+      "step": 4041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.277734302204049e-05,
+      "loss": 0.3662,
+      "step": 4042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.277377932585516e-05,
+      "loss": 0.4194,
+      "step": 4043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2770215029657258e-05,
+      "loss": 0.3731,
+      "step": 4044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2766650133721884e-05,
+      "loss": 0.3776,
+      "step": 4045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2763084638324204e-05,
+      "loss": 0.4524,
+      "step": 4046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.275951854373941e-05,
+      "loss": 0.3982,
+      "step": 4047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.275595185024275e-05,
+      "loss": 0.4141,
+      "step": 4048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2752384558109516e-05,
+      "loss": 0.367,
+      "step": 4049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274881666761504e-05,
+      "loss": 0.3936,
+      "step": 4050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2745248179034716e-05,
+      "loss": 0.3667,
+      "step": 4051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2741679092643973e-05,
+      "loss": 0.3644,
+      "step": 4052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2738109408718286e-05,
+      "loss": 0.4223,
+      "step": 4053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.273453912753317e-05,
+      "loss": 0.3916,
+      "step": 4054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2730968249364215e-05,
+      "loss": 0.403,
+      "step": 4055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2727396774487016e-05,
+      "loss": 0.3482,
+      "step": 4056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2723824703177242e-05,
+      "loss": 0.3622,
+      "step": 4057
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.27202520357106e-05,
+      "loss": 0.3675,
+      "step": 4058
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2716678772362848e-05,
+      "loss": 0.3579,
+      "step": 4059
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2713104913409778e-05,
+      "loss": 0.3135,
+      "step": 4060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.270953045912723e-05,
+      "loss": 0.3316,
+      "step": 4061
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2705955409791114e-05,
+      "loss": 0.3866,
+      "step": 4062
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.270237976567735e-05,
+      "loss": 0.3665,
+      "step": 4063
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.269880352706193e-05,
+      "loss": 0.3292,
+      "step": 4064
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.269522669422088e-05,
+      "loss": 0.4119,
+      "step": 4065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2691649267430272e-05,
+      "loss": 0.4031,
+      "step": 4066
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2688071246966233e-05,
+      "loss": 0.4512,
+      "step": 4067
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2684492633104925e-05,
+      "loss": 0.3654,
+      "step": 4068
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.268091342612256e-05,
+      "loss": 0.3509,
+      "step": 4069
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.26773336262954e-05,
+      "loss": 0.3698,
+      "step": 4070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2673753233899747e-05,
+      "loss": 0.4253,
+      "step": 4071
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.267017224921195e-05,
+      "loss": 0.3776,
+      "step": 4072
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2666590672508402e-05,
+      "loss": 0.4662,
+      "step": 4073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2663008504065548e-05,
+      "loss": 0.4393,
+      "step": 4074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.265942574415987e-05,
+      "loss": 0.3992,
+      "step": 4075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2655842393067908e-05,
+      "loss": 0.352,
+      "step": 4076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2652258451066232e-05,
+      "loss": 0.3487,
+      "step": 4077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.264867391843147e-05,
+      "loss": 0.3588,
+      "step": 4078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2645088795440284e-05,
+      "loss": 0.3773,
+      "step": 4079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2641503082369402e-05,
+      "loss": 0.3502,
+      "step": 4080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.263791677949557e-05,
+      "loss": 0.3817,
+      "step": 4081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2634329887095607e-05,
+      "loss": 0.4012,
+      "step": 4082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2630742405446352e-05,
+      "loss": 0.3685,
+      "step": 4083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2627154334824714e-05,
+      "loss": 0.3867,
+      "step": 4084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.262356567550762e-05,
+      "loss": 0.3197,
+      "step": 4085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.261997642777207e-05,
+      "loss": 0.3924,
+      "step": 4086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.261638659189509e-05,
+      "loss": 0.4361,
+      "step": 4087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2612796168153767e-05,
+      "loss": 0.3848,
+      "step": 4088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.260920515682522e-05,
+      "loss": 0.3734,
+      "step": 4089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2605613558186613e-05,
+      "loss": 0.3854,
+      "step": 4090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.260202137251517e-05,
+      "loss": 0.4342,
+      "step": 4091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2598428600088135e-05,
+      "loss": 0.3323,
+      "step": 4092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.259483524118284e-05,
+      "loss": 0.3693,
+      "step": 4093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2591241296076606e-05,
+      "loss": 0.3902,
+      "step": 4094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2587646765046846e-05,
+      "loss": 0.3715,
+      "step": 4095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2584051648370997e-05,
+      "loss": 0.4214,
+      "step": 4096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2580455946326553e-05,
+      "loss": 0.3798,
+      "step": 4097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2576859659191026e-05,
+      "loss": 0.4169,
+      "step": 4098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2573262787242005e-05,
+      "loss": 0.3731,
+      "step": 4099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2569665330757116e-05,
+      "loss": 0.3749,
+      "step": 4100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2566067290014015e-05,
+      "loss": 0.3787,
+      "step": 4101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2562468665290416e-05,
+      "loss": 0.3859,
+      "step": 4102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.255886945686408e-05,
+      "loss": 0.3863,
+      "step": 4103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2555269665012807e-05,
+      "loss": 0.319,
+      "step": 4104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2551669290014443e-05,
+      "loss": 0.3659,
+      "step": 4105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2548068332146888e-05,
+      "loss": 0.3645,
+      "step": 4106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2544466791688066e-05,
+      "loss": 0.3883,
+      "step": 4107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2540864668915963e-05,
+      "loss": 0.3764,
+      "step": 4108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.253726196410861e-05,
+      "loss": 0.4004,
+      "step": 4109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2533658677544073e-05,
+      "loss": 0.3851,
+      "step": 4110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2530054809500474e-05,
+      "loss": 0.3876,
+      "step": 4111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2526450360255972e-05,
+      "loss": 0.333,
+      "step": 4112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2522845330088774e-05,
+      "loss": 0.3529,
+      "step": 4113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.251923971927713e-05,
+      "loss": 0.3514,
+      "step": 4114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.251563352809934e-05,
+      "loss": 0.3731,
+      "step": 4115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2512026756833735e-05,
+      "loss": 0.3313,
+      "step": 4116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.250841940575871e-05,
+      "loss": 0.418,
+      "step": 4117
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2504811475152695e-05,
+      "loss": 0.4016,
+      "step": 4118
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.250120296529416e-05,
+      "loss": 0.3643,
+      "step": 4119
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2497593876461623e-05,
+      "loss": 0.3732,
+      "step": 4120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2493984208933658e-05,
+      "loss": 0.3392,
+      "step": 4121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2490373962988868e-05,
+      "loss": 0.3381,
+      "step": 4122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.248676313890591e-05,
+      "loss": 0.3259,
+      "step": 4123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2483151736963473e-05,
+      "loss": 0.3838,
+      "step": 4124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.247953975744031e-05,
+      "loss": 0.3796,
+      "step": 4125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.247592720061521e-05,
+      "loss": 0.3709,
+      "step": 4126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2472314066767e-05,
+      "loss": 0.3671,
+      "step": 4127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2468700356174553e-05,
+      "loss": 0.4213,
+      "step": 4128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2465086069116795e-05,
+      "loss": 0.2958,
+      "step": 4129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2461471205872695e-05,
+      "loss": 0.3711,
+      "step": 4130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2457855766721258e-05,
+      "loss": 0.3874,
+      "step": 4131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2454239751941538e-05,
+      "loss": 0.3955,
+      "step": 4132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.245062316181264e-05,
+      "loss": 0.4268,
+      "step": 4133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.24470059966137e-05,
+      "loss": 0.3809,
+      "step": 4134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2443388256623915e-05,
+      "loss": 0.4192,
+      "step": 4135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2439769942122506e-05,
+      "loss": 0.3817,
+      "step": 4136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2436151053388754e-05,
+      "loss": 0.3689,
+      "step": 4137
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2432531590701984e-05,
+      "loss": 0.4175,
+      "step": 4138
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2428911554341557e-05,
+      "loss": 0.3436,
+      "step": 4139
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2425290944586883e-05,
+      "loss": 0.3731,
+      "step": 4140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2421669761717418e-05,
+      "loss": 0.3513,
+      "step": 4141
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2418048006012662e-05,
+      "loss": 0.3202,
+      "step": 4142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.241442567775215e-05,
+      "loss": 0.3247,
+      "step": 4143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2410802777215468e-05,
+      "loss": 0.4039,
+      "step": 4144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2407179304682254e-05,
+      "loss": 0.3787,
+      "step": 4145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.240355526043218e-05,
+      "loss": 0.3996,
+      "step": 4146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.239993064474496e-05,
+      "loss": 0.3734,
+      "step": 4147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.239630545790036e-05,
+      "loss": 0.4096,
+      "step": 4148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2392679700178192e-05,
+      "loss": 0.423,
+      "step": 4149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.23890533718583e-05,
+      "loss": 0.3658,
+      "step": 4150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2385426473220586e-05,
+      "loss": 0.3633,
+      "step": 4151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2381799004544976e-05,
+      "loss": 0.3543,
+      "step": 4152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2378170966111468e-05,
+      "loss": 0.4001,
+      "step": 4153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2374542358200078e-05,
+      "loss": 0.3751,
+      "step": 4154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.237091318109089e-05,
+      "loss": 0.3595,
+      "step": 4155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2367283435064005e-05,
+      "loss": 0.3828,
+      "step": 4156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2363653120399593e-05,
+      "loss": 0.3718,
+      "step": 4157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2360022237377842e-05,
+      "loss": 0.3944,
+      "step": 4158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2356390786279015e-05,
+      "loss": 0.3701,
+      "step": 4159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.23527587673834e-05,
+      "loss": 0.359,
+      "step": 4160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.234912618097132e-05,
+      "loss": 0.4105,
+      "step": 4161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2345493027323157e-05,
+      "loss": 0.3688,
+      "step": 4162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2341859306719343e-05,
+      "loss": 0.3641,
+      "step": 4163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2338225019440335e-05,
+      "loss": 0.3615,
+      "step": 4164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2334590165766642e-05,
+      "loss": 0.4049,
+      "step": 4165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2330954745978818e-05,
+      "loss": 0.3842,
+      "step": 4166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2327318760357466e-05,
+      "loss": 0.4047,
+      "step": 4167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2323682209183218e-05,
+      "loss": 0.4178,
+      "step": 4168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.232004509273676e-05,
+      "loss": 0.3782,
+      "step": 4169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2316407411298824e-05,
+      "loss": 0.3441,
+      "step": 4170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.231276916515018e-05,
+      "loss": 0.3954,
+      "step": 4171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.230913035457164e-05,
+      "loss": 0.4023,
+      "step": 4172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2305490979844062e-05,
+      "loss": 0.3933,
+      "step": 4173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.230185104124835e-05,
+      "loss": 0.4136,
+      "step": 4174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2298210539065452e-05,
+      "loss": 0.4362,
+      "step": 4175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2294569473576355e-05,
+      "loss": 0.4029,
+      "step": 4176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.229092784506209e-05,
+      "loss": 0.4235,
+      "step": 4177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.228728565380373e-05,
+      "loss": 0.3747,
+      "step": 4178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2283642900082407e-05,
+      "loss": 0.37,
+      "step": 4179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2279999584179275e-05,
+      "loss": 0.3782,
+      "step": 4180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2276355706375533e-05,
+      "loss": 0.3496,
+      "step": 4181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2272711266952443e-05,
+      "loss": 0.3898,
+      "step": 4182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.22690662661913e-05,
+      "loss": 0.3675,
+      "step": 4183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2265420704373426e-05,
+      "loss": 0.474,
+      "step": 4184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2261774581780205e-05,
+      "loss": 0.3829,
+      "step": 4185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.225812789869307e-05,
+      "loss": 0.4304,
+      "step": 4186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.225448065539348e-05,
+      "loss": 0.3862,
+      "step": 4187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2250832852162943e-05,
+      "loss": 0.387,
+      "step": 4188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.224718448928301e-05,
+      "loss": 0.3733,
+      "step": 4189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.224353556703529e-05,
+      "loss": 0.4023,
+      "step": 4190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2239886085701403e-05,
+      "loss": 0.3822,
+      "step": 4191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2236236045563045e-05,
+      "loss": 0.3945,
+      "step": 4192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2232585446901935e-05,
+      "loss": 0.4143,
+      "step": 4193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2228934289999842e-05,
+      "loss": 0.4189,
+      "step": 4194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.222528257513858e-05,
+      "loss": 0.3646,
+      "step": 4195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2221630302600006e-05,
+      "loss": 0.4238,
+      "step": 4196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2217977472666004e-05,
+      "loss": 0.427,
+      "step": 4197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.221432408561853e-05,
+      "loss": 0.3349,
+      "step": 4198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.221067014173956e-05,
+      "loss": 0.351,
+      "step": 4199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.220701564131112e-05,
+      "loss": 0.3568,
+      "step": 4200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2203360584615288e-05,
+      "loss": 0.4098,
+      "step": 4201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2199704971934167e-05,
+      "loss": 0.4006,
+      "step": 4202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.219604880354991e-05,
+      "loss": 0.3952,
+      "step": 4203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.219239207974473e-05,
+      "loss": 0.3986,
+      "step": 4204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2188734800800855e-05,
+      "loss": 0.3666,
+      "step": 4205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.218507696700057e-05,
+      "loss": 0.3973,
+      "step": 4206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.218141857862621e-05,
+      "loss": 0.41,
+      "step": 4207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2177759635960137e-05,
+      "loss": 0.3497,
+      "step": 4208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2174100139284768e-05,
+      "loss": 0.375,
+      "step": 4209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2170440088882548e-05,
+      "loss": 0.3744,
+      "step": 4210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2166779485035987e-05,
+      "loss": 0.3743,
+      "step": 4211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2163118328027624e-05,
+      "loss": 0.3935,
+      "step": 4212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.215945661814004e-05,
+      "loss": 0.3746,
+      "step": 4213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.215579435565586e-05,
+      "loss": 0.4229,
+      "step": 4214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2152131540857746e-05,
+      "loss": 0.2934,
+      "step": 4215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.214846817402842e-05,
+      "loss": 0.3602,
+      "step": 4216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2144804255450642e-05,
+      "loss": 0.3702,
+      "step": 4217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.214113978540719e-05,
+      "loss": 0.3507,
+      "step": 4218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.213747476418091e-05,
+      "loss": 0.3854,
+      "step": 4219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.213380919205469e-05,
+      "loss": 0.3827,
+      "step": 4220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2130143069311454e-05,
+      "loss": 0.3495,
+      "step": 4221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2126476396234152e-05,
+      "loss": 0.3985,
+      "step": 4222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.212280917310581e-05,
+      "loss": 0.3833,
+      "step": 4223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.211914140020948e-05,
+      "loss": 0.4005,
+      "step": 4224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2115473077828248e-05,
+      "loss": 0.4171,
+      "step": 4225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.211180420624525e-05,
+      "loss": 0.3921,
+      "step": 4226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2108134785743668e-05,
+      "loss": 0.3485,
+      "step": 4227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2104464816606724e-05,
+      "loss": 0.3547,
+      "step": 4228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2100794299117685e-05,
+      "loss": 0.4039,
+      "step": 4229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2097123233559844e-05,
+      "loss": 0.3979,
+      "step": 4230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.209345162021656e-05,
+      "loss": 0.3934,
+      "step": 4231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.208977945937122e-05,
+      "loss": 0.3925,
+      "step": 4232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.208610675130726e-05,
+      "loss": 0.3817,
+      "step": 4233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2082433496308148e-05,
+      "loss": 0.4161,
+      "step": 4234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.207875969465741e-05,
+      "loss": 0.3561,
+      "step": 4235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2075085346638598e-05,
+      "loss": 0.3777,
+      "step": 4236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2071410452535315e-05,
+      "loss": 0.3532,
+      "step": 4237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2067735012631204e-05,
+      "loss": 0.472,
+      "step": 4238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2064059027209957e-05,
+      "loss": 0.3959,
+      "step": 4239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.206038249655529e-05,
+      "loss": 0.3438,
+      "step": 4240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2056705420950982e-05,
+      "loss": 0.3883,
+      "step": 4241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2053027800680846e-05,
+      "loss": 0.3768,
+      "step": 4242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2049349636028732e-05,
+      "loss": 0.326,
+      "step": 4243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2045670927278534e-05,
+      "loss": 0.3333,
+      "step": 4244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2041991674714193e-05,
+      "loss": 0.4253,
+      "step": 4245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2038311878619696e-05,
+      "loss": 0.3647,
+      "step": 4246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2034631539279053e-05,
+      "loss": 0.3891,
+      "step": 4247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2030950656976328e-05,
+      "loss": 0.3496,
+      "step": 4248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2027269231995633e-05,
+      "loss": 0.3433,
+      "step": 4249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2023587264621117e-05,
+      "loss": 0.3508,
+      "step": 4250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2019904755136967e-05,
+      "loss": 0.3611,
+      "step": 4251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.201622170382741e-05,
+      "loss": 0.3688,
+      "step": 4252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2012538110976726e-05,
+      "loss": 0.4083,
+      "step": 4253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2008853976869234e-05,
+      "loss": 0.3748,
+      "step": 4254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2005169301789273e-05,
+      "loss": 0.4525,
+      "step": 4255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2001484086021257e-05,
+      "loss": 0.4203,
+      "step": 4256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.199779832984962e-05,
+      "loss": 0.3469,
+      "step": 4257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1994112033558845e-05,
+      "loss": 0.3892,
+      "step": 4258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1990425197433463e-05,
+      "loss": 0.3713,
+      "step": 4259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1986737821758028e-05,
+      "loss": 0.3421,
+      "step": 4260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.198304990681715e-05,
+      "loss": 0.3961,
+      "step": 4261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1979361452895478e-05,
+      "loss": 0.3721,
+      "step": 4262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.197567246027771e-05,
+      "loss": 0.3614,
+      "step": 4263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1971982929248567e-05,
+      "loss": 0.3611,
+      "step": 4264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.196829286009283e-05,
+      "loss": 0.3714,
+      "step": 4265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1964602253095303e-05,
+      "loss": 0.317,
+      "step": 4266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1960911108540857e-05,
+      "loss": 0.3116,
+      "step": 4267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1957219426714378e-05,
+      "loss": 0.3566,
+      "step": 4268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1953527207900815e-05,
+      "loss": 0.3843,
+      "step": 4269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.194983445238514e-05,
+      "loss": 0.3495,
+      "step": 4270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1946141160452376e-05,
+      "loss": 0.3787,
+      "step": 4271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1942447332387594e-05,
+      "loss": 0.3815,
+      "step": 4272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1938752968475892e-05,
+      "loss": 0.3645,
+      "step": 4273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1935058069002422e-05,
+      "loss": 0.4177,
+      "step": 4274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1931362634252366e-05,
+      "loss": 0.371,
+      "step": 4275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1927666664510958e-05,
+      "loss": 0.409,
+      "step": 4276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1923970160063467e-05,
+      "loss": 0.4107,
+      "step": 4277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19202731211952e-05,
+      "loss": 0.3301,
+      "step": 4278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1916575548191512e-05,
+      "loss": 0.3433,
+      "step": 4279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.19128774413378e-05,
+      "loss": 0.3883,
+      "step": 4280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1909178800919498e-05,
+      "loss": 0.3677,
+      "step": 4281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.190547962722208e-05,
+      "loss": 0.4168,
+      "step": 4282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.190177992053107e-05,
+      "loss": 0.4052,
+      "step": 4283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1898079681132023e-05,
+      "loss": 0.389,
+      "step": 4284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1894378909310535e-05,
+      "loss": 0.3885,
+      "step": 4285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1890677605352246e-05,
+      "loss": 0.3932,
+      "step": 4286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1886975769542847e-05,
+      "loss": 0.4242,
+      "step": 4287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.188327340216806e-05,
+      "loss": 0.3806,
+      "step": 4288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.187957050351364e-05,
+      "loss": 0.3605,
+      "step": 4289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1875867073865398e-05,
+      "loss": 0.3739,
+      "step": 4290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1872163113509177e-05,
+      "loss": 0.3746,
+      "step": 4291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1868458622730868e-05,
+      "loss": 0.3525,
+      "step": 4292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.18647536018164e-05,
+      "loss": 0.3659,
+      "step": 4293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1861048051051734e-05,
+      "loss": 0.3109,
+      "step": 4294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1857341970722892e-05,
+      "loss": 0.3366,
+      "step": 4295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1853635361115918e-05,
+      "loss": 0.3646,
+      "step": 4296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1849928222516896e-05,
+      "loss": 0.3692,
+      "step": 4297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.184622055521197e-05,
+      "loss": 0.3382,
+      "step": 4298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.184251235948731e-05,
+      "loss": 0.3499,
+      "step": 4299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.183880363562913e-05,
+      "loss": 0.3946,
+      "step": 4300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1835094383923678e-05,
+      "loss": 0.383,
+      "step": 4301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1831384604657256e-05,
+      "loss": 0.3976,
+      "step": 4302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.18276742981162e-05,
+      "loss": 0.3229,
+      "step": 4303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.182396346458689e-05,
+      "loss": 0.3616,
+      "step": 4304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1820252104355737e-05,
+      "loss": 0.4123,
+      "step": 4305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1816540217709202e-05,
+      "loss": 0.3901,
+      "step": 4306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.181282780493378e-05,
+      "loss": 0.4237,
+      "step": 4307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1809114866316017e-05,
+      "loss": 0.3672,
+      "step": 4308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1805401402142487e-05,
+      "loss": 0.3757,
+      "step": 4309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.180168741269982e-05,
+      "loss": 0.3158,
+      "step": 4310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1797972898274667e-05,
+      "loss": 0.3507,
+      "step": 4311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1794257859153736e-05,
+      "loss": 0.4338,
+      "step": 4312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.179054229562376e-05,
+      "loss": 0.397,
+      "step": 4313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1786826207971533e-05,
+      "loss": 0.3853,
+      "step": 4314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1783109596483874e-05,
+      "loss": 0.3599,
+      "step": 4315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1779392461447647e-05,
+      "loss": 0.3869,
+      "step": 4316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.177567480314975e-05,
+      "loss": 0.4156,
+      "step": 4317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1771956621877133e-05,
+      "loss": 0.3809,
+      "step": 4318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1768237917916778e-05,
+      "loss": 0.4321,
+      "step": 4319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1764518691555717e-05,
+      "loss": 0.3543,
+      "step": 4320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.176079894308101e-05,
+      "loss": 0.3908,
+      "step": 4321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1757078672779756e-05,
+      "loss": 0.333,
+      "step": 4322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1753357880939114e-05,
+      "loss": 0.4175,
+      "step": 4323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1749636567846257e-05,
+      "loss": 0.3737,
+      "step": 4324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1745914733788426e-05,
+      "loss": 0.3824,
+      "step": 4325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1742192379052876e-05,
+      "loss": 0.3796,
+      "step": 4326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1738469503926914e-05,
+      "loss": 0.3562,
+      "step": 4327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1734746108697895e-05,
+      "loss": 0.3717,
+      "step": 4328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1731022193653205e-05,
+      "loss": 0.3598,
+      "step": 4329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.172729775908026e-05,
+      "loss": 0.3712,
+      "step": 4330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.172357280526654e-05,
+      "loss": 0.3599,
+      "step": 4331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1719847332499545e-05,
+      "loss": 0.3558,
+      "step": 4332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1716121341066826e-05,
+      "loss": 0.3811,
+      "step": 4333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1712394831255972e-05,
+      "loss": 0.3373,
+      "step": 4334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1708667803354603e-05,
+      "loss": 0.3311,
+      "step": 4335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1704940257650393e-05,
+      "loss": 0.3313,
+      "step": 4336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.170121219443105e-05,
+      "loss": 0.3437,
+      "step": 4337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1697483613984317e-05,
+      "loss": 0.4047,
+      "step": 4338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1693754516597988e-05,
+      "loss": 0.3738,
+      "step": 4339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1690024902559883e-05,
+      "loss": 0.377,
+      "step": 4340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.1686294772157874e-05,
+      "loss": 0.3509,
+      "step": 4341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.168256412567986e-05,
+      "loss": 0.3309,
+      "step": 4342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1678832963413798e-05,
+      "loss": 0.3382,
+      "step": 4343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.167510128564767e-05,
+      "loss": 0.3663,
+      "step": 4344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1671369092669507e-05,
+      "loss": 0.3271,
+      "step": 4345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1667636384767367e-05,
+      "loss": 0.3479,
+      "step": 4346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1663903162229357e-05,
+      "loss": 0.3409,
+      "step": 4347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1660169425343627e-05,
+      "loss": 0.3882,
+      "step": 4348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1656435174398362e-05,
+      "loss": 0.407,
+      "step": 4349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1652700409681786e-05,
+      "loss": 0.4201,
+      "step": 4350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1648965131482166e-05,
+      "loss": 0.3639,
+      "step": 4351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1645229340087794e-05,
+      "loss": 0.3778,
+      "step": 4352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1641493035787034e-05,
+      "loss": 0.3743,
+      "step": 4353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1637756218868253e-05,
+      "loss": 0.3864,
+      "step": 4354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1634018889619886e-05,
+      "loss": 0.3873,
+      "step": 4355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1630281048330385e-05,
+      "loss": 0.389,
+      "step": 4356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1626542695288257e-05,
+      "loss": 0.352,
+      "step": 4357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1622803830782045e-05,
+      "loss": 0.3471,
+      "step": 4358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.161906445510033e-05,
+      "loss": 0.3499,
+      "step": 4359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1615324568531725e-05,
+      "loss": 0.3565,
+      "step": 4360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.16115841713649e-05,
+      "loss": 0.3448,
+      "step": 4361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.160784326388855e-05,
+      "loss": 0.371,
+      "step": 4362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1604101846391416e-05,
+      "loss": 0.4029,
+      "step": 4363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.160035991916227e-05,
+      "loss": 0.3648,
+      "step": 4364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1596617482489935e-05,
+      "loss": 0.3968,
+      "step": 4365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1592874536663274e-05,
+      "loss": 0.3786,
+      "step": 4366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1589131081971167e-05,
+      "loss": 0.2691,
+      "step": 4367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.158538711870256e-05,
+      "loss": 0.3287,
+      "step": 4368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1581642647146428e-05,
+      "loss": 0.38,
+      "step": 4369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1577897667591783e-05,
+      "loss": 0.3178,
+      "step": 4370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1574152180327676e-05,
+      "loss": 0.338,
+      "step": 4371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1570406185643197e-05,
+      "loss": 0.3461,
+      "step": 4372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1566659683827485e-05,
+      "loss": 0.383,
+      "step": 4373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.156291267516971e-05,
+      "loss": 0.3716,
+      "step": 4374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1559165159959075e-05,
+      "loss": 0.4041,
+      "step": 4375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1555417138484832e-05,
+      "loss": 0.3157,
+      "step": 4376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1551668611036268e-05,
+      "loss": 0.3629,
+      "step": 4377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.154791957790272e-05,
+      "loss": 0.3409,
+      "step": 4378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1544170039373538e-05,
+      "loss": 0.3768,
+      "step": 4379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.154041999573813e-05,
+      "loss": 0.3247,
+      "step": 4380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.153666944728595e-05,
+      "loss": 0.3679,
+      "step": 4381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.153291839430648e-05,
+      "loss": 0.4578,
+      "step": 4382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1529166837089225e-05,
+      "loss": 0.3735,
+      "step": 4383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1525414775923763e-05,
+      "loss": 0.3783,
+      "step": 4384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.152166221109969e-05,
+      "loss": 0.408,
+      "step": 4385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.151790914290664e-05,
+      "loss": 0.388,
+      "step": 4386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1514155571634298e-05,
+      "loss": 0.353,
+      "step": 4387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1510401497572367e-05,
+      "loss": 0.3247,
+      "step": 4388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1506646921010617e-05,
+      "loss": 0.3093,
+      "step": 4389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1502891842238835e-05,
+      "loss": 0.3799,
+      "step": 4390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.149913626154685e-05,
+      "loss": 0.4424,
+      "step": 4391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.149538017922454e-05,
+      "loss": 0.3688,
+      "step": 4392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1491623595561814e-05,
+      "loss": 0.3726,
+      "step": 4393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.148786651084861e-05,
+      "loss": 0.4223,
+      "step": 4394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1484108925374935e-05,
+      "loss": 0.3878,
+      "step": 4395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.14803508394308e-05,
+      "loss": 0.3427,
+      "step": 4396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1476592253306275e-05,
+      "loss": 0.4331,
+      "step": 4397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.147283316729146e-05,
+      "loss": 0.4113,
+      "step": 4398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.14690735816765e-05,
+      "loss": 0.3464,
+      "step": 4399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1465313496751577e-05,
+      "loss": 0.3848,
+      "step": 4400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.1461552912806902e-05,
+      "loss": 0.4159,
+      "step": 4401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.145779183013274e-05,
+      "loss": 0.3453,
+      "step": 4402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.145403024901939e-05,
+      "loss": 0.3964,
+      "step": 4403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1450268169757176e-05,
+      "loss": 0.3862,
+      "step": 4404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1446505592636477e-05,
+      "loss": 0.3864,
+      "step": 4405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1442742517947703e-05,
+      "loss": 0.3542,
+      "step": 4406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.143897894598131e-05,
+      "loss": 0.3585,
+      "step": 4407
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.143521487702778e-05,
+      "loss": 0.414,
+      "step": 4408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1431450311377632e-05,
+      "loss": 0.411,
+      "step": 4409
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1427685249321445e-05,
+      "loss": 0.381,
+      "step": 4410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1423919691149817e-05,
+      "loss": 0.3575,
+      "step": 4411
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1420153637153386e-05,
+      "loss": 0.3726,
+      "step": 4412
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1416387087622833e-05,
+      "loss": 0.3402,
+      "step": 4413
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1412620042848876e-05,
+      "loss": 0.3823,
+      "step": 4414
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1408852503122276e-05,
+      "loss": 0.3825,
+      "step": 4415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1405084468733822e-05,
+      "loss": 0.3434,
+      "step": 4416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1401315939974348e-05,
+      "loss": 0.3552,
+      "step": 4417
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1397546917134724e-05,
+      "loss": 0.4261,
+      "step": 4418
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1393777400505857e-05,
+      "loss": 0.3677,
+      "step": 4419
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1390007390378704e-05,
+      "loss": 0.5113,
+      "step": 4420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1386236887044236e-05,
+      "loss": 0.3741,
+      "step": 4421
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1382465890793483e-05,
+      "loss": 0.3771,
+      "step": 4422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1378694401917512e-05,
+      "loss": 0.373,
+      "step": 4423
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1374922420707415e-05,
+      "loss": 0.375,
+      "step": 4424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1371149947454324e-05,
+      "loss": 0.346,
+      "step": 4425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1367376982449425e-05,
+      "loss": 0.3499,
+      "step": 4426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.136360352598392e-05,
+      "loss": 0.3389,
+      "step": 4427
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1359829578349076e-05,
+      "loss": 0.3506,
+      "step": 4428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.135605513983617e-05,
+      "loss": 0.3577,
+      "step": 4429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1352280210736526e-05,
+      "loss": 0.3615,
+      "step": 4430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.134850479134152e-05,
+      "loss": 0.3832,
+      "step": 4431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1344728881942542e-05,
+      "loss": 0.3741,
+      "step": 4432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1340952482831043e-05,
+      "loss": 0.3684,
+      "step": 4433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.13371755942985e-05,
+      "loss": 0.3903,
+      "step": 4434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.133339821663642e-05,
+      "loss": 0.397,
+      "step": 4435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1329620350136367e-05,
+      "loss": 0.3739,
+      "step": 4436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1325841995089926e-05,
+      "loss": 0.3577,
+      "step": 4437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.132206315178873e-05,
+      "loss": 0.3823,
+      "step": 4438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.131828382052444e-05,
+      "loss": 0.3912,
+      "step": 4439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.131450400158877e-05,
+      "loss": 0.3487,
+      "step": 4440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1310723695273458e-05,
+      "loss": 0.3349,
+      "step": 4441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1306942901870273e-05,
+      "loss": 0.4073,
+      "step": 4442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.130316162167105e-05,
+      "loss": 0.3688,
+      "step": 4443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1299379854967634e-05,
+      "loss": 0.374,
+      "step": 4444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1295597602051922e-05,
+      "loss": 0.388,
+      "step": 4445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1291814863215834e-05,
+      "loss": 0.3658,
+      "step": 4446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1288031638751345e-05,
+      "loss": 0.3754,
+      "step": 4447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1284247928950463e-05,
+      "loss": 0.3995,
+      "step": 4448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.128046373410523e-05,
+      "loss": 0.3777,
+      "step": 4449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.127667905450772e-05,
+      "loss": 0.3442,
+      "step": 4450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.127289389045005e-05,
+      "loss": 0.3292,
+      "step": 4451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1269108242224384e-05,
+      "loss": 0.3747,
+      "step": 4452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1265322110122904e-05,
+      "loss": 0.3799,
+      "step": 4453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1261535494437843e-05,
+      "loss": 0.3745,
+      "step": 4454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.125774839546147e-05,
+      "loss": 0.4174,
+      "step": 4455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1253960813486086e-05,
+      "loss": 0.3871,
+      "step": 4456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1250172748804036e-05,
+      "loss": 0.3975,
+      "step": 4457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1246384201707693e-05,
+      "loss": 0.3838,
+      "step": 4458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.1242595172489476e-05,
+      "loss": 0.3993,
+      "step": 4459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.123880566144184e-05,
+      "loss": 0.3368,
+      "step": 4460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1235015668857273e-05,
+      "loss": 0.3985,
+      "step": 4461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.12312251950283e-05,
+      "loss": 0.3951,
+      "step": 4462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1227434240247484e-05,
+      "loss": 0.3538,
+      "step": 4463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1223642804807436e-05,
+      "loss": 0.3687,
+      "step": 4464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.121985088900079e-05,
+      "loss": 0.3927,
+      "step": 4465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1216058493120217e-05,
+      "loss": 0.4219,
+      "step": 4466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1212265617458437e-05,
+      "loss": 0.3569,
+      "step": 4467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1208472262308196e-05,
+      "loss": 0.3774,
+      "step": 4468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1204678427962286e-05,
+      "loss": 0.3727,
+      "step": 4469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1200884114713518e-05,
+      "loss": 0.3463,
+      "step": 4470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1197089322854767e-05,
+      "loss": 0.3717,
+      "step": 4471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1193294052678923e-05,
+      "loss": 0.4134,
+      "step": 4472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1189498304478924e-05,
+      "loss": 0.3922,
+      "step": 4473
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1185702078547742e-05,
+      "loss": 0.4231,
+      "step": 4474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1181905375178382e-05,
+      "loss": 0.3635,
+      "step": 4475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.117810819466389e-05,
+      "loss": 0.3664,
+      "step": 4476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1174310537297355e-05,
+      "loss": 0.4814,
+      "step": 4477
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1170512403371885e-05,
+      "loss": 0.3765,
+      "step": 4478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1166713793180643e-05,
+      "loss": 0.3474,
+      "step": 4479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1162914707016817e-05,
+      "loss": 0.384,
+      "step": 4480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1159115145173644e-05,
+      "loss": 0.3634,
+      "step": 4481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1155315107944386e-05,
+      "loss": 0.3561,
+      "step": 4482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.115151459562234e-05,
+      "loss": 0.3673,
+      "step": 4483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1147713608500846e-05,
+      "loss": 0.3381,
+      "step": 4484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1143912146873293e-05,
+      "loss": 0.3766,
+      "step": 4485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1140110211033084e-05,
+      "loss": 0.3578,
+      "step": 4486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1136307801273665e-05,
+      "loss": 0.3781,
+      "step": 4487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1132504917888523e-05,
+      "loss": 0.4124,
+      "step": 4488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.112870156117119e-05,
+      "loss": 0.3699,
+      "step": 4489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.112489773141521e-05,
+      "loss": 0.3665,
+      "step": 4490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1121093428914192e-05,
+      "loss": 0.3412,
+      "step": 4491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.111728865396176e-05,
+      "loss": 0.4058,
+      "step": 4492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1113483406851582e-05,
+      "loss": 0.3214,
+      "step": 4493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.110967768787737e-05,
+      "loss": 0.3765,
+      "step": 4494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.110587149733286e-05,
+      "loss": 0.3846,
+      "step": 4495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.110206483551182e-05,
+      "loss": 0.3724,
+      "step": 4496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1098257702708084e-05,
+      "loss": 0.3614,
+      "step": 4497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1094450099215493e-05,
+      "loss": 0.3566,
+      "step": 4498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1090642025327924e-05,
+      "loss": 0.3701,
+      "step": 4499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1086833481339313e-05,
+      "loss": 0.4128,
+      "step": 4500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1083024467543617e-05,
+      "loss": 0.3855,
+      "step": 4501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.107921498423483e-05,
+      "loss": 0.3524,
+      "step": 4502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1075405031706983e-05,
+      "loss": 0.3785,
+      "step": 4503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.107159461025414e-05,
+      "loss": 0.4495,
+      "step": 4504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1067783720170412e-05,
+      "loss": 0.3529,
+      "step": 4505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1063972361749937e-05,
+      "loss": 0.4042,
+      "step": 4506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.106016053528689e-05,
+      "loss": 0.3297,
+      "step": 4507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1056348241075482e-05,
+      "loss": 0.3902,
+      "step": 4508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1052535479409973e-05,
+      "loss": 0.3737,
+      "step": 4509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1048722250584632e-05,
+      "loss": 0.3403,
+      "step": 4510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1044908554893792e-05,
+      "loss": 0.3869,
+      "step": 4511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1041094392631804e-05,
+      "loss": 0.3223,
+      "step": 4512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.103727976409306e-05,
+      "loss": 0.3613,
+      "step": 4513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.103346466957199e-05,
+      "loss": 0.3248,
+      "step": 4514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1029649109363056e-05,
+      "loss": 0.3389,
+      "step": 4515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.102583308376077e-05,
+      "loss": 0.3443,
+      "step": 4516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.1022016593059656e-05,
+      "loss": 0.3255,
+      "step": 4517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.101819963755429e-05,
+      "loss": 0.3814,
+      "step": 4518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1014382217539285e-05,
+      "loss": 0.3915,
+      "step": 4519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1010564333309287e-05,
+      "loss": 0.3757,
+      "step": 4520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1006745985158965e-05,
+      "loss": 0.3863,
+      "step": 4521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1002927173383042e-05,
+      "loss": 0.3793,
+      "step": 4522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.099910789827627e-05,
+      "loss": 0.3966,
+      "step": 4523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.099528816013344e-05,
+      "loss": 0.3488,
+      "step": 4524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0991467959249367e-05,
+      "loss": 0.3943,
+      "step": 4525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0987647295918913e-05,
+      "loss": 0.3498,
+      "step": 4526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.098382617043698e-05,
+      "loss": 0.3676,
+      "step": 4527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0980004583098487e-05,
+      "loss": 0.3799,
+      "step": 4528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0976182534198406e-05,
+      "loss": 0.3891,
+      "step": 4529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0972360024031737e-05,
+      "loss": 0.3684,
+      "step": 4530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.096853705289352e-05,
+      "loss": 0.3596,
+      "step": 4531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0964713621078823e-05,
+      "loss": 0.3445,
+      "step": 4532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0960889728882755e-05,
+      "loss": 0.348,
+      "step": 4533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0957065376600464e-05,
+      "loss": 0.4227,
+      "step": 4534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0953240564527135e-05,
+      "loss": 0.3161,
+      "step": 4535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0949415292957963e-05,
+      "loss": 0.3393,
+      "step": 4536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0945589562188213e-05,
+      "loss": 0.3751,
+      "step": 4537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.094176337251317e-05,
+      "loss": 0.3411,
+      "step": 4538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0937936724228156e-05,
+      "loss": 0.3885,
+      "step": 4539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0934109617628525e-05,
+      "loss": 0.3274,
+      "step": 4540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0930282053009666e-05,
+      "loss": 0.379,
+      "step": 4541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0926454030667012e-05,
+      "loss": 0.3904,
+      "step": 4542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.092262555089602e-05,
+      "loss": 0.3775,
+      "step": 4543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0918796613992194e-05,
+      "loss": 0.3421,
+      "step": 4544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0914967220251062e-05,
+      "loss": 0.3234,
+      "step": 4545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0911137369968198e-05,
+      "loss": 0.414,
+      "step": 4546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0907307063439198e-05,
+      "loss": 0.3393,
+      "step": 4547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.090347630095971e-05,
+      "loss": 0.348,
+      "step": 4548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.08996450828254e-05,
+      "loss": 0.3609,
+      "step": 4549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0895813409331985e-05,
+      "loss": 0.358,
+      "step": 4550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.08919812807752e-05,
+      "loss": 0.3406,
+      "step": 4551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0888148697450834e-05,
+      "loss": 0.3558,
+      "step": 4552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0884315659654697e-05,
+      "loss": 0.3903,
+      "step": 4553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.088048216768264e-05,
+      "loss": 0.3715,
+      "step": 4554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0876648221830547e-05,
+      "loss": 0.3983,
+      "step": 4555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0872813822394338e-05,
+      "loss": 0.3476,
+      "step": 4556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0868978969669968e-05,
+      "loss": 0.3385,
+      "step": 4557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.086514366395343e-05,
+      "loss": 0.3293,
+      "step": 4558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0861307905540737e-05,
+      "loss": 0.3827,
+      "step": 4559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0857471694727965e-05,
+      "loss": 0.3374,
+      "step": 4560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0853635031811202e-05,
+      "loss": 0.4192,
+      "step": 4561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0849797917086573e-05,
+      "loss": 0.3573,
+      "step": 4562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0845960350850246e-05,
+      "loss": 0.4413,
+      "step": 4563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.084212233339842e-05,
+      "loss": 0.3704,
+      "step": 4564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.083828386502734e-05,
+      "loss": 0.3566,
+      "step": 4565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0834444946033252e-05,
+      "loss": 0.3962,
+      "step": 4566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0830605576712472e-05,
+      "loss": 0.3423,
+      "step": 4567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.082676575736135e-05,
+      "loss": 0.3495,
+      "step": 4568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.082292548827624e-05,
+      "loss": 0.3483,
+      "step": 4569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.081908476975356e-05,
+      "loss": 0.3459,
+      "step": 4570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0815243602089746e-05,
+      "loss": 0.3713,
+      "step": 4571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0811401985581288e-05,
+      "loss": 0.3789,
+      "step": 4572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0807559920524685e-05,
+      "loss": 0.3637,
+      "step": 4573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.080371740721649e-05,
+      "loss": 0.3457,
+      "step": 4574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0799874445953278e-05,
+      "loss": 0.4584,
+      "step": 4575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0796031037031674e-05,
+      "loss": 0.35,
+      "step": 4576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.079218718074832e-05,
+      "loss": 0.4041,
+      "step": 4577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.078834287739991e-05,
+      "loss": 0.3723,
+      "step": 4578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0784498127283148e-05,
+      "loss": 0.3245,
+      "step": 4579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0780652930694806e-05,
+      "loss": 0.3326,
+      "step": 4580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0776807287931662e-05,
+      "loss": 0.4222,
+      "step": 4581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.077296119929053e-05,
+      "loss": 0.3709,
+      "step": 4582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0769114665068287e-05,
+      "loss": 0.3682,
+      "step": 4583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0765267685561815e-05,
+      "loss": 0.4102,
+      "step": 4584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.076142026106804e-05,
+      "loss": 0.3844,
+      "step": 4585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0757572391883916e-05,
+      "loss": 0.3706,
+      "step": 4586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.075372407830645e-05,
+      "loss": 0.3617,
+      "step": 4587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.074987532063266e-05,
+      "loss": 0.424,
+      "step": 4588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0746026119159615e-05,
+      "loss": 0.3547,
+      "step": 4589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0742176474184406e-05,
+      "loss": 0.3434,
+      "step": 4590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0738326386004175e-05,
+      "loss": 0.3613,
+      "step": 4591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.073447585491608e-05,
+      "loss": 0.3127,
+      "step": 4592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0730624881217323e-05,
+      "loss": 0.3448,
+      "step": 4593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0726773465205137e-05,
+      "loss": 0.3811,
+      "step": 4594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0722921607176793e-05,
+      "loss": 0.3779,
+      "step": 4595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.071906930742959e-05,
+      "loss": 0.4166,
+      "step": 4596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.071521656626087e-05,
+      "loss": 0.4011,
+      "step": 4597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0711363383968e-05,
+      "loss": 0.4024,
+      "step": 4598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0707509760848377e-05,
+      "loss": 0.4265,
+      "step": 4599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.070365569719945e-05,
+      "loss": 0.4154,
+      "step": 4600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0699801193318693e-05,
+      "loss": 0.4097,
+      "step": 4601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0695946249503606e-05,
+      "loss": 0.3699,
+      "step": 4602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.069209086605173e-05,
+      "loss": 0.3797,
+      "step": 4603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.068823504326064e-05,
+      "loss": 0.3542,
+      "step": 4604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.068437878142795e-05,
+      "loss": 0.3696,
+      "step": 4605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0680522080851297e-05,
+      "loss": 0.3303,
+      "step": 4606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.067666494182836e-05,
+      "loss": 0.383,
+      "step": 4607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0672807364656844e-05,
+      "loss": 0.3462,
+      "step": 4608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.06689493496345e-05,
+      "loss": 0.363,
+      "step": 4609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0665090897059102e-05,
+      "loss": 0.4171,
+      "step": 4610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0661232007228456e-05,
+      "loss": 0.3155,
+      "step": 4611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.065737268044041e-05,
+      "loss": 0.358,
+      "step": 4612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0653512916992855e-05,
+      "loss": 0.3552,
+      "step": 4613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.064965271718369e-05,
+      "loss": 0.3784,
+      "step": 4614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0645792081310863e-05,
+      "loss": 0.3609,
+      "step": 4615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0641931009672355e-05,
+      "loss": 0.3846,
+      "step": 4616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0638069502566186e-05,
+      "loss": 0.3477,
+      "step": 4617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0634207560290396e-05,
+      "loss": 0.3923,
+      "step": 4618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0630345183143063e-05,
+      "loss": 0.412,
+      "step": 4619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.06264823714223e-05,
+      "loss": 0.3604,
+      "step": 4620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0622619125426274e-05,
+      "loss": 0.3839,
+      "step": 4621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0618755445453146e-05,
+      "loss": 0.3941,
+      "step": 4622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.061489133180114e-05,
+      "loss": 0.3878,
+      "step": 4623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0611026784768498e-05,
+      "loss": 0.4047,
+      "step": 4624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0607161804653512e-05,
+      "loss": 0.3559,
+      "step": 4625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.060329639175449e-05,
+      "loss": 0.3763,
+      "step": 4626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.059943054636978e-05,
+      "loss": 0.3267,
+      "step": 4627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0595564268797765e-05,
+      "loss": 0.4079,
+      "step": 4628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.059169755933686e-05,
+      "loss": 0.3795,
+      "step": 4629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.058783041828552e-05,
+      "loss": 0.3712,
+      "step": 4630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0583962845942214e-05,
+      "loss": 0.3647,
+      "step": 4631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0580094842605472e-05,
+      "loss": 0.3219,
+      "step": 4632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0576226408573832e-05,
+      "loss": 0.3679,
+      "step": 4633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.057235754414588e-05,
+      "loss": 0.3728,
+      "step": 4634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0568488249620232e-05,
+      "loss": 0.4482,
+      "step": 4635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056461852529553e-05,
+      "loss": 0.358,
+      "step": 4636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0560748371470466e-05,
+      "loss": 0.3533,
+      "step": 4637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0556877788443746e-05,
+      "loss": 0.3476,
+      "step": 4638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0553006776514118e-05,
+      "loss": 0.3641,
+      "step": 4639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.054913533598037e-05,
+      "loss": 0.3876,
+      "step": 4640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0545263467141306e-05,
+      "loss": 0.3313,
+      "step": 4641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0541391170295786e-05,
+      "loss": 0.3712,
+      "step": 4642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0537518445742673e-05,
+      "loss": 0.3823,
+      "step": 4643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0533645293780895e-05,
+      "loss": 0.3492,
+      "step": 4644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0529771714709384e-05,
+      "loss": 0.352,
+      "step": 4645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052589770882713e-05,
+      "loss": 0.3796,
+      "step": 4646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0522023276433147e-05,
+      "loss": 0.3868,
+      "step": 4647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051814841782647e-05,
+      "loss": 0.3469,
+      "step": 4648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0514273133306178e-05,
+      "loss": 0.4195,
+      "step": 4649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0510397423171387e-05,
+      "loss": 0.3663,
+      "step": 4650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0506521287721244e-05,
+      "loss": 0.3606,
+      "step": 4651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0502644727254912e-05,
+      "loss": 0.3797,
+      "step": 4652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0498767742071603e-05,
+      "loss": 0.3387,
+      "step": 4653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049489033247057e-05,
+      "loss": 0.3522,
+      "step": 4654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0491012498751082e-05,
+      "loss": 0.4104,
+      "step": 4655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048713424121244e-05,
+      "loss": 0.3929,
+      "step": 4656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0483255560153992e-05,
+      "loss": 0.3429,
+      "step": 4657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047937645587511e-05,
+      "loss": 0.3648,
+      "step": 4658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0475496928675193e-05,
+      "loss": 0.4362,
+      "step": 4659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0471616978853686e-05,
+      "loss": 0.4093,
+      "step": 4660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0467736606710048e-05,
+      "loss": 0.3822,
+      "step": 4661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04638558125438e-05,
+      "loss": 0.3859,
+      "step": 4662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045997459665447e-05,
+      "loss": 0.3695,
+      "step": 4663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0456092959341615e-05,
+      "loss": 0.3933,
+      "step": 4664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0452210900904854e-05,
+      "loss": 0.4236,
+      "step": 4665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.044832842164381e-05,
+      "loss": 0.3975,
+      "step": 4666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0444445521858154e-05,
+      "loss": 0.3362,
+      "step": 4667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0440562201847577e-05,
+      "loss": 0.3881,
+      "step": 4668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0436678461911817e-05,
+      "loss": 0.3461,
+      "step": 4669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0432794302350635e-05,
+      "loss": 0.3271,
+      "step": 4670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0428909723463825e-05,
+      "loss": 0.3817,
+      "step": 4671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0425024725551215e-05,
+      "loss": 0.3588,
+      "step": 4672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0421139308912672e-05,
+      "loss": 0.3787,
+      "step": 4673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0417253473848076e-05,
+      "loss": 0.3478,
+      "step": 4674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041336722065736e-05,
+      "loss": 0.3028,
+      "step": 4675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0409480549640486e-05,
+      "loss": 0.3968,
+      "step": 4676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040559346109743e-05,
+      "loss": 0.3487,
+      "step": 4677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0401705955328222e-05,
+      "loss": 0.4176,
+      "step": 4678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0397818032632916e-05,
+      "loss": 0.3591,
+      "step": 4679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03939296933116e-05,
+      "loss": 0.4328,
+      "step": 4680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0390040937664384e-05,
+      "loss": 0.3484,
+      "step": 4681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0386151765991427e-05,
+      "loss": 0.3306,
+      "step": 4682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0382262178592905e-05,
+      "loss": 0.3607,
+      "step": 4683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0378372175769038e-05,
+      "loss": 0.364,
+      "step": 4684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0374481757820067e-05,
+      "loss": 0.3898,
+      "step": 4685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0370590925046275e-05,
+      "loss": 0.3428,
+      "step": 4686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0366699677747968e-05,
+      "loss": 0.3888,
+      "step": 4687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03628080162255e-05,
+      "loss": 0.3888,
+      "step": 4688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035891594077923e-05,
+      "loss": 0.3467,
+      "step": 4689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0355023451709573e-05,
+      "loss": 0.3562,
+      "step": 4690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0351130549316964e-05,
+      "loss": 0.3092,
+      "step": 4691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0347237233901882e-05,
+      "loss": 0.3422,
+      "step": 4692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0343343505764816e-05,
+      "loss": 0.3699,
+      "step": 4693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.033944936520631e-05,
+      "loss": 0.3983,
+      "step": 4694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.033555481252693e-05,
+      "loss": 0.3734,
+      "step": 4695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0331659848027267e-05,
+      "loss": 0.3556,
+      "step": 4696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0327764472007958e-05,
+      "loss": 0.3729,
+      "step": 4697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0323868684769657e-05,
+      "loss": 0.3659,
+      "step": 4698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0319972486613058e-05,
+      "loss": 0.3602,
+      "step": 4699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0316075877838897e-05,
+      "loss": 0.3708,
+      "step": 4700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0312178858747916e-05,
+      "loss": 0.3285,
+      "step": 4701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.030828142964091e-05,
+      "loss": 0.343,
+      "step": 4702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.03043835908187e-05,
+      "loss": 0.3609,
+      "step": 4703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.030048534258213e-05,
+      "loss": 0.3641,
+      "step": 4704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0296586685232093e-05,
+      "loss": 0.3988,
+      "step": 4705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0292687619069494e-05,
+      "loss": 0.347,
+      "step": 4706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0288788144395287e-05,
+      "loss": 0.3745,
+      "step": 4707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.028488826151045e-05,
+      "loss": 0.3924,
+      "step": 4708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0280987970715982e-05,
+      "loss": 0.3635,
+      "step": 4709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.027708727231293e-05,
+      "loss": 0.4135,
+      "step": 4710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.027318616660237e-05,
+      "loss": 0.3304,
+      "step": 4711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.02692846538854e-05,
+      "loss": 0.3708,
+      "step": 4712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0265382734463158e-05,
+      "loss": 0.4122,
+      "step": 4713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.026148040863681e-05,
+      "loss": 0.3963,
+      "step": 4714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0257577676707552e-05,
+      "loss": 0.3442,
+      "step": 4715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.025367453897661e-05,
+      "loss": 0.3502,
+      "step": 4716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0249770995745256e-05,
+      "loss": 0.3641,
+      "step": 4717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0245867047314768e-05,
+      "loss": 0.3977,
+      "step": 4718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0241962693986477e-05,
+      "loss": 0.3299,
+      "step": 4719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0238057936061732e-05,
+      "loss": 0.3469,
+      "step": 4720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0234152773841927e-05,
+      "loss": 0.3421,
+      "step": 4721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.023024720762847e-05,
+      "loss": 0.4139,
+      "step": 4722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0226341237722814e-05,
+      "loss": 0.3715,
+      "step": 4723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0222434864426434e-05,
+      "loss": 0.3345,
+      "step": 4724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.021852808804084e-05,
+      "loss": 0.3644,
+      "step": 4725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.021462090886758e-05,
+      "loss": 0.3802,
+      "step": 4726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0210713327208223e-05,
+      "loss": 0.4211,
+      "step": 4727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0206805343364363e-05,
+      "loss": 0.346,
+      "step": 4728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.020289695763765e-05,
+      "loss": 0.3312,
+      "step": 4729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.019898817032974e-05,
+      "loss": 0.3766,
+      "step": 4730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.019507898174233e-05,
+      "loss": 0.3946,
+      "step": 4731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.019116939217715e-05,
+      "loss": 0.3119,
+      "step": 4732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0187259401935955e-05,
+      "loss": 0.3892,
+      "step": 4733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.018334901132054e-05,
+      "loss": 0.3852,
+      "step": 4734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.017943822063272e-05,
+      "loss": 0.3713,
+      "step": 4735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0175527030174343e-05,
+      "loss": 0.3566,
+      "step": 4736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.01716154402473e-05,
+      "loss": 0.315,
+      "step": 4737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0167703451153504e-05,
+      "loss": 0.364,
+      "step": 4738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0163791063194886e-05,
+      "loss": 0.3987,
+      "step": 4739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0159878276673425e-05,
+      "loss": 0.406,
+      "step": 4740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.015596509189114e-05,
+      "loss": 0.3876,
+      "step": 4741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0152051509150053e-05,
+      "loss": 0.3894,
+      "step": 4742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0148137528752232e-05,
+      "loss": 0.3668,
+      "step": 4743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0144223150999776e-05,
+      "loss": 0.3525,
+      "step": 4744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0140308376194812e-05,
+      "loss": 0.3299,
+      "step": 4745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0136393204639508e-05,
+      "loss": 0.3642,
+      "step": 4746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0132477636636033e-05,
+      "loss": 0.4097,
+      "step": 4747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0128561672486625e-05,
+      "loss": 0.3879,
+      "step": 4748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0124645312493528e-05,
+      "loss": 0.3482,
+      "step": 4749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0120728556959022e-05,
+      "loss": 0.3563,
+      "step": 4750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0116811406185418e-05,
+      "loss": 0.3995,
+      "step": 4751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.011289386047506e-05,
+      "loss": 0.3577,
+      "step": 4752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.010897592013032e-05,
+      "loss": 0.3353,
+      "step": 4753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.01050575854536e-05,
+      "loss": 0.4029,
+      "step": 4754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.010113885674733e-05,
+      "loss": 0.3584,
+      "step": 4755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0097219734313987e-05,
+      "loss": 0.3172,
+      "step": 4756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0093300218456048e-05,
+      "loss": 0.4207,
+      "step": 4757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.008938030947605e-05,
+      "loss": 0.3516,
+      "step": 4758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0085460007676544e-05,
+      "loss": 0.3216,
+      "step": 4759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.008153931336011e-05,
+      "loss": 0.3282,
+      "step": 4760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.007761822682937e-05,
+      "loss": 0.3345,
+      "step": 4761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.007369674838697e-05,
+      "loss": 0.3786,
+      "step": 4762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0069774878335583e-05,
+      "loss": 0.3652,
+      "step": 4763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0065852616977914e-05,
+      "loss": 0.3437,
+      "step": 4764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0061929964616697e-05,
+      "loss": 0.403,
+      "step": 4765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.005800692155471e-05,
+      "loss": 0.3373,
+      "step": 4766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0054083488094742e-05,
+      "loss": 0.3737,
+      "step": 4767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0050159664539615e-05,
+      "loss": 0.3749,
+      "step": 4768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0046235451192195e-05,
+      "loss": 0.3654,
+      "step": 4769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0042310848355366e-05,
+      "loss": 0.3767,
+      "step": 4770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0038385856332048e-05,
+      "loss": 0.3417,
+      "step": 4771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.003446047542518e-05,
+      "loss": 0.3563,
+      "step": 4772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0030534705937743e-05,
+      "loss": 0.3787,
+      "step": 4773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.002660854817275e-05,
+      "loss": 0.4015,
+      "step": 4774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0022682002433236e-05,
+      "loss": 0.3564,
+      "step": 4775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0018755069022254e-05,
+      "loss": 0.3561,
+      "step": 4776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.001482774824292e-05,
+      "loss": 0.3507,
+      "step": 4777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0010900040398354e-05,
+      "loss": 0.4091,
+      "step": 4778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.000697194579171e-05,
+      "loss": 0.351,
+      "step": 4779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.000304346472618e-05,
+      "loss": 0.4354,
+      "step": 4780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999114597504974e-05,
+      "loss": 0.3893,
+      "step": 4781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9995185344431345e-05,
+      "loss": 0.4045,
+      "step": 4782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9991255705808567e-05,
+      "loss": 0.3561,
+      "step": 4783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.998732568193994e-05,
+      "loss": 0.4228,
+      "step": 4784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9983395273128805e-05,
+      "loss": 0.3551,
+      "step": 4785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.997946447967853e-05,
+      "loss": 0.3347,
+      "step": 4786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.997553330189251e-05,
+      "loss": 0.3407,
+      "step": 4787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9971601740074156e-05,
+      "loss": 0.4144,
+      "step": 4788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.996766979452694e-05,
+      "loss": 0.4145,
+      "step": 4789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9963737465554334e-05,
+      "loss": 0.3516,
+      "step": 4790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9959804753459858e-05,
+      "loss": 0.4258,
+      "step": 4791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9955871658547053e-05,
+      "loss": 0.3749,
+      "step": 4792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.995193818111949e-05,
+      "loss": 0.3637,
+      "step": 4793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.994800432148077e-05,
+      "loss": 0.3609,
+      "step": 4794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9944070079934535e-05,
+      "loss": 0.358,
+      "step": 4795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9940135456784433e-05,
+      "loss": 0.3728,
+      "step": 4796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.993620045233416e-05,
+      "loss": 0.3838,
+      "step": 4797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9932265066887434e-05,
+      "loss": 0.3579,
+      "step": 4798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9928329300748006e-05,
+      "loss": 0.3876,
+      "step": 4799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9924393154219657e-05,
+      "loss": 0.3673,
+      "step": 4800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99204566276062e-05,
+      "loss": 0.343,
+      "step": 4801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.991651972121145e-05,
+      "loss": 0.4036,
+      "step": 4802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99125824353393e-05,
+      "loss": 0.3334,
+      "step": 4803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.990864477029364e-05,
+      "loss": 0.3421,
+      "step": 4804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9904706726378382e-05,
+      "loss": 0.3992,
+      "step": 4805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9900768303897488e-05,
+      "loss": 0.4533,
+      "step": 4806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.989682950315495e-05,
+      "loss": 0.4619,
+      "step": 4807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9892890324454774e-05,
+      "loss": 0.317,
+      "step": 4808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9888950768100998e-05,
+      "loss": 0.3334,
+      "step": 4809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9885010834397697e-05,
+      "loss": 0.3853,
+      "step": 4810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9881070523648975e-05,
+      "loss": 0.3334,
+      "step": 4811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.987712983615896e-05,
+      "loss": 0.3683,
+      "step": 4812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9873188772231805e-05,
+      "loss": 0.325,
+      "step": 4813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9869247332171703e-05,
+      "loss": 0.3442,
+      "step": 4814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.986530551628287e-05,
+      "loss": 0.4017,
+      "step": 4815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9861363324869558e-05,
+      "loss": 0.3912,
+      "step": 4816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.985742075823603e-05,
+      "loss": 0.3245,
+      "step": 4817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.985347781668659e-05,
+      "loss": 0.3242,
+      "step": 4818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.984953450052558e-05,
+      "loss": 0.3395,
+      "step": 4819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9845590810057358e-05,
+      "loss": 0.3316,
+      "step": 4820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9841646745586314e-05,
+      "loss": 0.3796,
+      "step": 4821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.983770230741686e-05,
+      "loss": 0.3749,
+      "step": 4822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.983375749585346e-05,
+      "loss": 0.3637,
+      "step": 4823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9829812311200573e-05,
+      "loss": 0.3586,
+      "step": 4824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9825866753762717e-05,
+      "loss": 0.385,
+      "step": 4825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.982192082384442e-05,
+      "loss": 0.3398,
+      "step": 4826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9817974521750254e-05,
+      "loss": 0.3941,
+      "step": 4827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9814027847784802e-05,
+      "loss": 0.3865,
+      "step": 4828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9810080802252684e-05,
+      "loss": 0.4046,
+      "step": 4829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.980613338545856e-05,
+      "loss": 0.3384,
+      "step": 4830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9802185597707102e-05,
+      "loss": 0.3863,
+      "step": 4831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9798237439303007e-05,
+      "loss": 0.357,
+      "step": 4832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9794288910551028e-05,
+      "loss": 0.3851,
+      "step": 4833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9790340011755915e-05,
+      "loss": 0.3494,
+      "step": 4834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9786390743222463e-05,
+      "loss": 0.3325,
+      "step": 4835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9782441105255503e-05,
+      "loss": 0.3649,
+      "step": 4836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9778491098159873e-05,
+      "loss": 0.3299,
+      "step": 4837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9774540722240455e-05,
+      "loss": 0.4024,
+      "step": 4838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9770589977802153e-05,
+      "loss": 0.362,
+      "step": 4839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9766638865149903e-05,
+      "loss": 0.3463,
+      "step": 4840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9762687384588673e-05,
+      "loss": 0.3862,
+      "step": 4841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.975873553642345e-05,
+      "loss": 0.4001,
+      "step": 4842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9754783320959255e-05,
+      "loss": 0.3718,
+      "step": 4843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9750830738501138e-05,
+      "loss": 0.3263,
+      "step": 4844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9746877789354174e-05,
+      "loss": 0.4041,
+      "step": 4845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9742924473823464e-05,
+      "loss": 0.3612,
+      "step": 4846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9738970792214147e-05,
+      "loss": 0.3808,
+      "step": 4847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9735016744831382e-05,
+      "loss": 0.3677,
+      "step": 4848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9731062331980365e-05,
+      "loss": 0.3722,
+      "step": 4849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.972710755396631e-05,
+      "loss": 0.3597,
+      "step": 4850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9723152411094454e-05,
+      "loss": 0.3388,
+      "step": 4851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.971919690367008e-05,
+      "loss": 0.3672,
+      "step": 4852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.97152410319985e-05,
+      "loss": 0.3661,
+      "step": 4853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.971128479638503e-05,
+      "loss": 0.3375,
+      "step": 4854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9707328197135026e-05,
+      "loss": 0.3624,
+      "step": 4855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9703371234553885e-05,
+      "loss": 0.3689,
+      "step": 4856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9699413908947026e-05,
+      "loss": 0.3765,
+      "step": 4857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.969545622061988e-05,
+      "loss": 0.3566,
+      "step": 4858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.969149816987792e-05,
+      "loss": 0.3427,
+      "step": 4859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9687539757026648e-05,
+      "loss": 0.3569,
+      "step": 4860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.968358098237159e-05,
+      "loss": 0.3649,
+      "step": 4861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9679621846218306e-05,
+      "loss": 0.4167,
+      "step": 4862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9675662348872367e-05,
+      "loss": 0.4437,
+      "step": 4863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.967170249063939e-05,
+      "loss": 0.385,
+      "step": 4864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.966774227182502e-05,
+      "loss": 0.3937,
+      "step": 4865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.966378169273491e-05,
+      "loss": 0.3815,
+      "step": 4866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9659820753674757e-05,
+      "loss": 0.339,
+      "step": 4867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9655859454950294e-05,
+      "loss": 0.344,
+      "step": 4868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9651897796867254e-05,
+      "loss": 0.3518,
+      "step": 4869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9647935779731427e-05,
+      "loss": 0.4014,
+      "step": 4870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9643973403848612e-05,
+      "loss": 0.3438,
+      "step": 4871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9640010669524646e-05,
+      "loss": 0.3752,
+      "step": 4872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9636047577065385e-05,
+      "loss": 0.3537,
+      "step": 4873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9632084126776716e-05,
+      "loss": 0.3815,
+      "step": 4874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9628120318964555e-05,
+      "loss": 0.3558,
+      "step": 4875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9624156153934852e-05,
+      "loss": 0.4086,
+      "step": 4876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.962019163199357e-05,
+      "loss": 0.386,
+      "step": 4877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.961622675344671e-05,
+      "loss": 0.3539,
+      "step": 4878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.96122615186003e-05,
+      "loss": 0.3679,
+      "step": 4879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9608295927760388e-05,
+      "loss": 0.4097,
+      "step": 4880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9604329981233057e-05,
+      "loss": 0.398,
+      "step": 4881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9600363679324423e-05,
+      "loss": 0.3257,
+      "step": 4882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9596397022340612e-05,
+      "loss": 0.3293,
+      "step": 4883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9592430010587784e-05,
+      "loss": 0.3714,
+      "step": 4884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.958846264437214e-05,
+      "loss": 0.3474,
+      "step": 4885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9584494923999893e-05,
+      "loss": 0.3604,
+      "step": 4886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9580526849777288e-05,
+      "loss": 0.3574,
+      "step": 4887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9576558422010595e-05,
+      "loss": 0.3566,
+      "step": 4888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9572589641006118e-05,
+      "loss": 0.4206,
+      "step": 4889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9568620507070184e-05,
+      "loss": 0.3518,
+      "step": 4890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9564651020509146e-05,
+      "loss": 0.3352,
+      "step": 4891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9560681181629382e-05,
+      "loss": 0.3687,
+      "step": 4892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.95567109907373e-05,
+      "loss": 0.3562,
+      "step": 4893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9552740448139345e-05,
+      "loss": 0.3723,
+      "step": 4894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9548769554141975e-05,
+      "loss": 0.3676,
+      "step": 4895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9544798309051678e-05,
+      "loss": 0.3882,
+      "step": 4896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.954082671317497e-05,
+      "loss": 0.3746,
+      "step": 4897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9536854766818402e-05,
+      "loss": 0.3688,
+      "step": 4898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.953288247028854e-05,
+      "loss": 0.4189,
+      "step": 4899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9528909823891983e-05,
+      "loss": 0.3593,
+      "step": 4900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9524936827935355e-05,
+      "loss": 0.3253,
+      "step": 4901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9520963482725317e-05,
+      "loss": 0.3583,
+      "step": 4902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9516989788568537e-05,
+      "loss": 0.3768,
+      "step": 4903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9513015745771726e-05,
+      "loss": 0.385,
+      "step": 4904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9509041354641617e-05,
+      "loss": 0.3558,
+      "step": 4905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9505066615484972e-05,
+      "loss": 0.3542,
+      "step": 4906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9501091528608577e-05,
+      "loss": 0.3823,
+      "step": 4907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.949711609431924e-05,
+      "loss": 0.3753,
+      "step": 4908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9493140312923807e-05,
+      "loss": 0.329,
+      "step": 4909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9489164184729146e-05,
+      "loss": 0.4016,
+      "step": 4910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9485187710042148e-05,
+      "loss": 0.3907,
+      "step": 4911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9481210889169736e-05,
+      "loss": 0.3544,
+      "step": 4912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.947723372241886e-05,
+      "loss": 0.3732,
+      "step": 4913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9473256210096485e-05,
+      "loss": 0.3462,
+      "step": 4914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9469278352509624e-05,
+      "loss": 0.3263,
+      "step": 4915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9465300149965294e-05,
+      "loss": 0.4027,
+      "step": 4916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9461321602770557e-05,
+      "loss": 0.3422,
+      "step": 4917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9457342711232487e-05,
+      "loss": 0.39,
+      "step": 4918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.94533634756582e-05,
+      "loss": 0.4079,
+      "step": 4919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.944938389635482e-05,
+      "loss": 0.3526,
+      "step": 4920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9445403973629516e-05,
+      "loss": 0.4049,
+      "step": 4921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9441423707789468e-05,
+      "loss": 0.3939,
+      "step": 4922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9437443099141895e-05,
+      "loss": 0.3359,
+      "step": 4923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.943346214799404e-05,
+      "loss": 0.3395,
+      "step": 4924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9429480854653153e-05,
+      "loss": 0.3727,
+      "step": 4925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9425499219426544e-05,
+      "loss": 0.3618,
+      "step": 4926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.942151724262152e-05,
+      "loss": 0.4225,
+      "step": 4927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9417534924545445e-05,
+      "loss": 0.3596,
+      "step": 4928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9413552265505668e-05,
+      "loss": 0.3311,
+      "step": 4929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9409569265809598e-05,
+      "loss": 0.3694,
+      "step": 4930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.940558592576466e-05,
+      "loss": 0.3509,
+      "step": 4931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9401602245678308e-05,
+      "loss": 0.3582,
+      "step": 4932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9397618225858013e-05,
+      "loss": 0.3964,
+      "step": 4933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9393633866611274e-05,
+      "loss": 0.3883,
+      "step": 4934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.938964916824563e-05,
+      "loss": 0.4262,
+      "step": 4935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9385664131068636e-05,
+      "loss": 0.3528,
+      "step": 4936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9381678755387866e-05,
+      "loss": 0.3379,
+      "step": 4937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9377693041510933e-05,
+      "loss": 0.2993,
+      "step": 4938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9373706989745473e-05,
+      "loss": 0.3753,
+      "step": 4939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9369720600399142e-05,
+      "loss": 0.3772,
+      "step": 4940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.936573387377963e-05,
+      "loss": 0.3506,
+      "step": 4941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9361746810194638e-05,
+      "loss": 0.3903,
+      "step": 4942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9357759409951923e-05,
+      "loss": 0.3826,
+      "step": 4943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9353771673359234e-05,
+      "loss": 0.3586,
+      "step": 4944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.934978360072436e-05,
+      "loss": 0.3893,
+      "step": 4945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.934579519235513e-05,
+      "loss": 0.3811,
+      "step": 4946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.934180644855938e-05,
+      "loss": 0.3911,
+      "step": 4947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.933781736964497e-05,
+      "loss": 0.3105,
+      "step": 4948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9333827955919807e-05,
+      "loss": 0.3707,
+      "step": 4949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.93298382076918e-05,
+      "loss": 0.3328,
+      "step": 4950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.93258481252689e-05,
+      "loss": 0.3917,
+      "step": 4951
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9321857708959073e-05,
+      "loss": 0.3093,
+      "step": 4952
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.931786695907032e-05,
+      "loss": 0.3785,
+      "step": 4953
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.931387587591067e-05,
+      "loss": 0.4129,
+      "step": 4954
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9309884459788155e-05,
+      "loss": 0.3647,
+      "step": 4955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.930589271101086e-05,
+      "loss": 0.3102,
+      "step": 4956
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9301900629886885e-05,
+      "loss": 0.36,
+      "step": 4957
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9297908216724353e-05,
+      "loss": 0.3319,
+      "step": 4958
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9293915471831415e-05,
+      "loss": 0.3954,
+      "step": 4959
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9289922395516248e-05,
+      "loss": 0.3328,
+      "step": 4960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9285928988087055e-05,
+      "loss": 0.3726,
+      "step": 4961
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9281935249852065e-05,
+      "loss": 0.3469,
+      "step": 4962
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9277941181119523e-05,
+      "loss": 0.3516,
+      "step": 4963
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.927394678219772e-05,
+      "loss": 0.3304,
+      "step": 4964
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9269952053394953e-05,
+      "loss": 0.3414,
+      "step": 4965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9265956995019552e-05,
+      "loss": 0.3713,
+      "step": 4966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9261961607379875e-05,
+      "loss": 0.3248,
+      "step": 4967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9257965890784295e-05,
+      "loss": 0.3573,
+      "step": 4968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.925396984554123e-05,
+      "loss": 0.4054,
+      "step": 4969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9249973471959108e-05,
+      "loss": 0.3808,
+      "step": 4970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9245976770346378e-05,
+      "loss": 0.4003,
+      "step": 4971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9241979741011526e-05,
+      "loss": 0.3757,
+      "step": 4972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9237982384263065e-05,
+      "loss": 0.3999,
+      "step": 4973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9233984700409523e-05,
+      "loss": 0.324,
+      "step": 4974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.922998668975945e-05,
+      "loss": 0.3884,
+      "step": 4975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9225988352621445e-05,
+      "loss": 0.4055,
+      "step": 4976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.922198968930411e-05,
+      "loss": 0.3411,
+      "step": 4977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9217990700116073e-05,
+      "loss": 0.3488,
+      "step": 4978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9213991385365993e-05,
+      "loss": 0.3382,
+      "step": 4979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9209991745362562e-05,
+      "loss": 0.3602,
+      "step": 4980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9205991780414478e-05,
+      "loss": 0.3417,
+      "step": 4981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.920199149083049e-05,
+      "loss": 0.3829,
+      "step": 4982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9197990876919337e-05,
+      "loss": 0.3817,
+      "step": 4983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9193989938989818e-05,
+      "loss": 0.381,
+      "step": 4984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.918998867735074e-05,
+      "loss": 0.3388,
+      "step": 4985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9185987092310935e-05,
+      "loss": 0.3907,
+      "step": 4986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9181985184179257e-05,
+      "loss": 0.3833,
+      "step": 4987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9177982953264596e-05,
+      "loss": 0.3671,
+      "step": 4988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.917398039987586e-05,
+      "loss": 0.3755,
+      "step": 4989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9169977524321985e-05,
+      "loss": 0.4034,
+      "step": 4990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9165974326911924e-05,
+      "loss": 0.3423,
+      "step": 4991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.916197080795466e-05,
+      "loss": 0.3407,
+      "step": 4992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9157966967759207e-05,
+      "loss": 0.3899,
+      "step": 4993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9153962806634598e-05,
+      "loss": 0.4409,
+      "step": 4994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9149958324889888e-05,
+      "loss": 0.4026,
+      "step": 4995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9145953522834154e-05,
+      "loss": 0.3616,
+      "step": 4996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.914194840077651e-05,
+      "loss": 0.3525,
+      "step": 4997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.913794295902609e-05,
+      "loss": 0.3604,
+      "step": 4998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9133937197892044e-05,
+      "loss": 0.4298,
+      "step": 4999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.912993111768356e-05,
+      "loss": 0.3375,
+      "step": 5000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9125924718709832e-05,
+      "loss": 0.3443,
+      "step": 5001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9121918001280107e-05,
+      "loss": 0.3476,
+      "step": 5002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9117910965703635e-05,
+      "loss": 0.3743,
+      "step": 5003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9113903612289687e-05,
+      "loss": 0.369,
+      "step": 5004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9109895941347573e-05,
+      "loss": 0.3957,
+      "step": 5005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9105887953186624e-05,
+      "loss": 0.324,
+      "step": 5006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9101879648116194e-05,
+      "loss": 0.3598,
+      "step": 5007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.909787102644565e-05,
+      "loss": 0.3684,
+      "step": 5008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9093862088484402e-05,
+      "loss": 0.4025,
+      "step": 5009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9089852834541885e-05,
+      "loss": 0.4344,
+      "step": 5010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9085843264927532e-05,
+      "loss": 0.3523,
+      "step": 5011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.908183337995083e-05,
+      "loss": 0.3747,
+      "step": 5012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9077823179921272e-05,
+      "loss": 0.3665,
+      "step": 5013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9073812665148393e-05,
+      "loss": 0.3991,
+      "step": 5014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.906980183594173e-05,
+      "loss": 0.3602,
+      "step": 5015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9065790692610857e-05,
+      "loss": 0.3548,
+      "step": 5016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9061779235465375e-05,
+      "loss": 0.3156,
+      "step": 5017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9057767464814902e-05,
+      "loss": 0.3781,
+      "step": 5018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9053755380969088e-05,
+      "loss": 0.4159,
+      "step": 5019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9049742984237594e-05,
+      "loss": 0.3855,
+      "step": 5020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9045730274930115e-05,
+      "loss": 0.3196,
+      "step": 5021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9041717253356378e-05,
+      "loss": 0.3328,
+      "step": 5022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9037703919826116e-05,
+      "loss": 0.3504,
+      "step": 5023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9033690274649095e-05,
+      "loss": 0.3628,
+      "step": 5024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.902967631813511e-05,
+      "loss": 0.3859,
+      "step": 5025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9025662050593965e-05,
+      "loss": 0.3757,
+      "step": 5026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9021647472335512e-05,
+      "loss": 0.3419,
+      "step": 5027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9017632583669597e-05,
+      "loss": 0.3655,
+      "step": 5028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9013617384906117e-05,
+      "loss": 0.3309,
+      "step": 5029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9009601876354976e-05,
+      "loss": 0.3214,
+      "step": 5030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9005586058326118e-05,
+      "loss": 0.3931,
+      "step": 5031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9001569931129485e-05,
+      "loss": 0.3517,
+      "step": 5032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.899755349507507e-05,
+      "loss": 0.3714,
+      "step": 5033
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8993536750472867e-05,
+      "loss": 0.3186,
+      "step": 5034
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.898951969763292e-05,
+      "loss": 0.3719,
+      "step": 5035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8985502336865273e-05,
+      "loss": 0.4389,
+      "step": 5036
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8981484668480005e-05,
+      "loss": 0.3428,
+      "step": 5037
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897746669278721e-05,
+      "loss": 0.3722,
+      "step": 5038
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973448410097024e-05,
+      "loss": 0.4223,
+      "step": 5039
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8969429820719582e-05,
+      "loss": 0.3168,
+      "step": 5040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896541092496506e-05,
+      "loss": 0.4262,
+      "step": 5041
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896139172314366e-05,
+      "loss": 0.3732,
+      "step": 5042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.895737221556559e-05,
+      "loss": 0.3564,
+      "step": 5043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.89533524025411e-05,
+      "loss": 0.3939,
+      "step": 5044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8949332284380452e-05,
+      "loss": 0.3127,
+      "step": 5045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945311861393933e-05,
+      "loss": 0.3144,
+      "step": 5046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941291133891867e-05,
+      "loss": 0.3393,
+      "step": 5047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893727010218458e-05,
+      "loss": 0.4052,
+      "step": 5048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933248766582434e-05,
+      "loss": 0.4112,
+      "step": 5049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892922712739581e-05,
+      "loss": 0.4316,
+      "step": 5050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925205184935123e-05,
+      "loss": 0.354,
+      "step": 5051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921182939510797e-05,
+      "loss": 0.3362,
+      "step": 5052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8917160391433288e-05,
+      "loss": 0.3884,
+      "step": 5053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8913137541013067e-05,
+      "loss": 0.3495,
+      "step": 5054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8909114388560644e-05,
+      "loss": 0.3518,
+      "step": 5055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8905090934386536e-05,
+      "loss": 0.3827,
+      "step": 5056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8901067178801294e-05,
+      "loss": 0.3644,
+      "step": 5057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.889704312211548e-05,
+      "loss": 0.333,
+      "step": 5058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8893018764639697e-05,
+      "loss": 0.3518,
+      "step": 5059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.888899410668456e-05,
+      "loss": 0.3492,
+      "step": 5060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8884969148560704e-05,
+      "loss": 0.4136,
+      "step": 5061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.888094389057879e-05,
+      "loss": 0.3855,
+      "step": 5062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8876918333049518e-05,
+      "loss": 0.349,
+      "step": 5063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8872892476283582e-05,
+      "loss": 0.3609,
+      "step": 5064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.886886632059172e-05,
+      "loss": 0.3628,
+      "step": 5065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.886483986628469e-05,
+      "loss": 0.386,
+      "step": 5066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8860813113673268e-05,
+      "loss": 0.3203,
+      "step": 5067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8856786063068258e-05,
+      "loss": 0.4017,
+      "step": 5068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8852758714780477e-05,
+      "loss": 0.3999,
+      "step": 5069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8848731069120785e-05,
+      "loss": 0.3742,
+      "step": 5070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.884470312640004e-05,
+      "loss": 0.3549,
+      "step": 5071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8840674886929144e-05,
+      "loss": 0.3444,
+      "step": 5072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8836646351019003e-05,
+      "loss": 0.3548,
+      "step": 5073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.883261751898057e-05,
+      "loss": 0.363,
+      "step": 5074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8828588391124794e-05,
+      "loss": 0.3682,
+      "step": 5075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.882455896776267e-05,
+      "loss": 0.3615,
+      "step": 5076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8820529249205197e-05,
+      "loss": 0.3925,
+      "step": 5077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8816499235763414e-05,
+      "loss": 0.3764,
+      "step": 5078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8812468927748366e-05,
+      "loss": 0.3903,
+      "step": 5079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8808438325471134e-05,
+      "loss": 0.3988,
+      "step": 5080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8804407429242818e-05,
+      "loss": 0.3623,
+      "step": 5081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8800376239374538e-05,
+      "loss": 0.4104,
+      "step": 5082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8796344756177427e-05,
+      "loss": 0.342,
+      "step": 5083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.879231297996267e-05,
+      "loss": 0.3477,
+      "step": 5084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.878828091104145e-05,
+      "loss": 0.3889,
+      "step": 5085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8784248549724972e-05,
+      "loss": 0.3153,
+      "step": 5086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8780215896324472e-05,
+      "loss": 0.3446,
+      "step": 5087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.877618295115121e-05,
+      "loss": 0.3766,
+      "step": 5088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8772149714516472e-05,
+      "loss": 0.3277,
+      "step": 5089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.876811618673155e-05,
+      "loss": 0.3093,
+      "step": 5090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8764082368107766e-05,
+      "loss": 0.382,
+      "step": 5091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8760048258956472e-05,
+      "loss": 0.342,
+      "step": 5092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.875601385958905e-05,
+      "loss": 0.3331,
+      "step": 5093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8751979170316867e-05,
+      "loss": 0.3615,
+      "step": 5094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8747944191451355e-05,
+      "loss": 0.3015,
+      "step": 5095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8743908923303945e-05,
+      "loss": 0.3716,
+      "step": 5096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8739873366186095e-05,
+      "loss": 0.3981,
+      "step": 5097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8735837520409286e-05,
+      "loss": 0.3216,
+      "step": 5098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.873180138628502e-05,
+      "loss": 0.4478,
+      "step": 5099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.872776496412483e-05,
+      "loss": 0.3486,
+      "step": 5100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8723728254240254e-05,
+      "loss": 0.3579,
+      "step": 5101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.871969125694287e-05,
+      "loss": 0.3753,
+      "step": 5102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8715653972544262e-05,
+      "loss": 0.3272,
+      "step": 5103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8711616401356055e-05,
+      "loss": 0.3247,
+      "step": 5104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8707578543689873e-05,
+      "loss": 0.3323,
+      "step": 5105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8703540399857387e-05,
+      "loss": 0.3567,
+      "step": 5106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.869950197017027e-05,
+      "loss": 0.3272,
+      "step": 5107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8695463254940227e-05,
+      "loss": 0.3859,
+      "step": 5108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.869142425447898e-05,
+      "loss": 0.3955,
+      "step": 5109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8687384969098282e-05,
+      "loss": 0.3485,
+      "step": 5110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.86833453991099e-05,
+      "loss": 0.3821,
+      "step": 5111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8679305544825618e-05,
+      "loss": 0.32,
+      "step": 5112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8675265406557253e-05,
+      "loss": 0.3359,
+      "step": 5113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8671224984616645e-05,
+      "loss": 0.3827,
+      "step": 5114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8667184279315643e-05,
+      "loss": 0.3357,
+      "step": 5115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.866314329096613e-05,
+      "loss": 0.3746,
+      "step": 5116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.865910201988e-05,
+      "loss": 0.3694,
+      "step": 5117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8655060466369186e-05,
+      "loss": 0.3385,
+      "step": 5118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8651018630745623e-05,
+      "loss": 0.3388,
+      "step": 5119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8646976513321276e-05,
+      "loss": 0.3908,
+      "step": 5120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.864293411440814e-05,
+      "loss": 0.3702,
+      "step": 5121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8638891434318218e-05,
+      "loss": 0.3229,
+      "step": 5122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8634848473363544e-05,
+      "loss": 0.3766,
+      "step": 5123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.863080523185617e-05,
+      "loss": 0.3643,
+      "step": 5124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8626761710108163e-05,
+      "loss": 0.3218,
+      "step": 5125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.862271790843163e-05,
+      "loss": 0.3505,
+      "step": 5126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8618673827138686e-05,
+      "loss": 0.3602,
+      "step": 5127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8614629466541464e-05,
+      "loss": 0.3199,
+      "step": 5128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8610584826952124e-05,
+      "loss": 0.375,
+      "step": 5129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8606539908682857e-05,
+      "loss": 0.3902,
+      "step": 5130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8602494712045864e-05,
+      "loss": 0.359,
+      "step": 5131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8598449237353366e-05,
+      "loss": 0.351,
+      "step": 5132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.859440348491761e-05,
+      "loss": 0.3323,
+      "step": 5133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8590357455050863e-05,
+      "loss": 0.2983,
+      "step": 5134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8586311148065424e-05,
+      "loss": 0.3117,
+      "step": 5135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8582264564273588e-05,
+      "loss": 0.4049,
+      "step": 5136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.85782177039877e-05,
+      "loss": 0.3404,
+      "step": 5137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.857417056752011e-05,
+      "loss": 0.3257,
+      "step": 5138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8570123155183192e-05,
+      "loss": 0.3312,
+      "step": 5139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8566075467289337e-05,
+      "loss": 0.3521,
+      "step": 5140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.856202750415097e-05,
+      "loss": 0.3898,
+      "step": 5141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.855797926608053e-05,
+      "loss": 0.4031,
+      "step": 5142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.855393075339047e-05,
+      "loss": 0.3406,
+      "step": 5143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8549881966393277e-05,
+      "loss": 0.3277,
+      "step": 5144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.854583290540145e-05,
+      "loss": 0.367,
+      "step": 5145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8541783570727513e-05,
+      "loss": 0.36,
+      "step": 5146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8537733962684014e-05,
+      "loss": 0.3446,
+      "step": 5147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.853368408158351e-05,
+      "loss": 0.3815,
+      "step": 5148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.85296339277386e-05,
+      "loss": 0.3892,
+      "step": 5149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8525583501461876e-05,
+      "loss": 0.3262,
+      "step": 5150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8521532803065984e-05,
+      "loss": 0.4365,
+      "step": 5151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8517481832863558e-05,
+      "loss": 0.3643,
+      "step": 5152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8513430591167282e-05,
+      "loss": 0.3696,
+      "step": 5153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8509379078289834e-05,
+      "loss": 0.3717,
+      "step": 5154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.850532729454394e-05,
+      "loss": 0.4235,
+      "step": 5155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.850127524024233e-05,
+      "loss": 0.371,
+      "step": 5156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8497222915697748e-05,
+      "loss": 0.3348,
+      "step": 5157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.849317032122298e-05,
+      "loss": 0.3749,
+      "step": 5158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8489117457130826e-05,
+      "loss": 0.355,
+      "step": 5159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.848506432373409e-05,
+      "loss": 0.3843,
+      "step": 5160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8481010921345613e-05,
+      "loss": 0.3707,
+      "step": 5161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.847695725027826e-05,
+      "loss": 0.3397,
+      "step": 5162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.847290331084491e-05,
+      "loss": 0.3444,
+      "step": 5163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8468849103358456e-05,
+      "loss": 0.3411,
+      "step": 5164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.846479462813182e-05,
+      "loss": 0.3813,
+      "step": 5165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8460739885477944e-05,
+      "loss": 0.4363,
+      "step": 5166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.84566848757098e-05,
+      "loss": 0.3286,
+      "step": 5167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8452629599140352e-05,
+      "loss": 0.3339,
+      "step": 5168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8448574056082618e-05,
+      "loss": 0.4074,
+      "step": 5169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8444518246849608e-05,
+      "loss": 0.386,
+      "step": 5170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8440462171754385e-05,
+      "loss": 0.3599,
+      "step": 5171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.843640583111e-05,
+      "loss": 0.3905,
+      "step": 5172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.843234922522954e-05,
+      "loss": 0.39,
+      "step": 5173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.842829235442611e-05,
+      "loss": 0.3227,
+      "step": 5174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.842423521901284e-05,
+      "loss": 0.3707,
+      "step": 5175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8420177819302878e-05,
+      "loss": 0.3695,
+      "step": 5176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8416120155609384e-05,
+      "loss": 0.363,
+      "step": 5177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.841206222824555e-05,
+      "loss": 0.3298,
+      "step": 5178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8408004037524584e-05,
+      "loss": 0.35,
+      "step": 5179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8403945583759713e-05,
+      "loss": 0.347,
+      "step": 5180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8399886867264178e-05,
+      "loss": 0.3775,
+      "step": 5181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.839582788835126e-05,
+      "loss": 0.3647,
+      "step": 5182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8391768647334245e-05,
+      "loss": 0.389,
+      "step": 5183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.838770914452644e-05,
+      "loss": 0.3514,
+      "step": 5184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.838364938024117e-05,
+      "loss": 0.4229,
+      "step": 5185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.837958935479179e-05,
+      "loss": 0.3404,
+      "step": 5186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.837552906849167e-05,
+      "loss": 0.3685,
+      "step": 5187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8371468521654196e-05,
+      "loss": 0.3761,
+      "step": 5188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8367407714592775e-05,
+      "loss": 0.3666,
+      "step": 5189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.836334664762085e-05,
+      "loss": 0.3618,
+      "step": 5190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8359285321051855e-05,
+      "loss": 0.4156,
+      "step": 5191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.835522373519927e-05,
+      "loss": 0.3487,
+      "step": 5192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8351161890376587e-05,
+      "loss": 0.3053,
+      "step": 5193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8347099786897308e-05,
+      "loss": 0.3404,
+      "step": 5194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8343037425074965e-05,
+      "loss": 0.301,
+      "step": 5195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8338974805223114e-05,
+      "loss": 0.3028,
+      "step": 5196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8334911927655315e-05,
+      "loss": 0.3418,
+      "step": 5197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.833084879268517e-05,
+      "loss": 0.3667,
+      "step": 5198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8326785400626277e-05,
+      "loss": 0.411,
+      "step": 5199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8322721751792272e-05,
+      "loss": 0.3393,
+      "step": 5200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831865784649681e-05,
+      "loss": 0.3339,
+      "step": 5201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831459368505354e-05,
+      "loss": 0.3904,
+      "step": 5202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8310529267776167e-05,
+      "loss": 0.353,
+      "step": 5203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8306464594978405e-05,
+      "loss": 0.3699,
+      "step": 5204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.830239966697397e-05,
+      "loss": 0.3881,
+      "step": 5205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8298334484076606e-05,
+      "loss": 0.4153,
+      "step": 5206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8294269046600094e-05,
+      "loss": 0.3799,
+      "step": 5207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8290203354858214e-05,
+      "loss": 0.3394,
+      "step": 5208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.828613740916478e-05,
+      "loss": 0.3316,
+      "step": 5209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8282071209833608e-05,
+      "loss": 0.3278,
+      "step": 5210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.827800475717855e-05,
+      "loss": 0.3222,
+      "step": 5211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.827393805151347e-05,
+      "loss": 0.3662,
+      "step": 5212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8269871093152262e-05,
+      "loss": 0.2907,
+      "step": 5213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8265803882408813e-05,
+      "loss": 0.3771,
+      "step": 5214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.826173641959706e-05,
+      "loss": 0.3586,
+      "step": 5215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.825766870503095e-05,
+      "loss": 0.3288,
+      "step": 5216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8253600739024436e-05,
+      "loss": 0.404,
+      "step": 5217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8249532521891503e-05,
+      "loss": 0.3378,
+      "step": 5218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.824546405394615e-05,
+      "loss": 0.3545,
+      "step": 5219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8241395335502416e-05,
+      "loss": 0.354,
+      "step": 5220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.823732636687432e-05,
+      "loss": 0.3338,
+      "step": 5221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.823325714837593e-05,
+      "loss": 0.3257,
+      "step": 5222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8229187680321325e-05,
+      "loss": 0.3478,
+      "step": 5223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8225117963024608e-05,
+      "loss": 0.3816,
+      "step": 5224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.822104799679989e-05,
+      "loss": 0.383,
+      "step": 5225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.821697778196131e-05,
+      "loss": 0.4015,
+      "step": 5226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8212907318823024e-05,
+      "loss": 0.3779,
+      "step": 5227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.820883660769921e-05,
+      "loss": 0.3539,
+      "step": 5228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8204765648904057e-05,
+      "loss": 0.3945,
+      "step": 5229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8200694442751782e-05,
+      "loss": 0.335,
+      "step": 5230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8196622989556623e-05,
+      "loss": 0.3625,
+      "step": 5231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8192551289632818e-05,
+      "loss": 0.3926,
+      "step": 5232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.818847934329465e-05,
+      "loss": 0.3416,
+      "step": 5233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8184407150856406e-05,
+      "loss": 0.345,
+      "step": 5234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.818033471263239e-05,
+      "loss": 0.3768,
+      "step": 5235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8176262028936933e-05,
+      "loss": 0.3615,
+      "step": 5236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.817218910008438e-05,
+      "loss": 0.3463,
+      "step": 5237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8168115926389103e-05,
+      "loss": 0.3555,
+      "step": 5238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.816404250816548e-05,
+      "loss": 0.3605,
+      "step": 5239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8159968845727915e-05,
+      "loss": 0.3619,
+      "step": 5240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8155894939390828e-05,
+      "loss": 0.351,
+      "step": 5241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8151820789468665e-05,
+      "loss": 0.388,
+      "step": 5242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8147746396275885e-05,
+      "loss": 0.3513,
+      "step": 5243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.814367176012696e-05,
+      "loss": 0.344,
+      "step": 5244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8139596881336397e-05,
+      "loss": 0.3481,
+      "step": 5245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8135521760218714e-05,
+      "loss": 0.4015,
+      "step": 5246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8131446397088432e-05,
+      "loss": 0.3811,
+      "step": 5247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.812737079226011e-05,
+      "loss": 0.3487,
+      "step": 5248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8123294946048327e-05,
+      "loss": 0.3513,
+      "step": 5249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8119218858767667e-05,
+      "loss": 0.4018,
+      "step": 5250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8115142530732742e-05,
+      "loss": 0.3432,
+      "step": 5251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.811106596225818e-05,
+      "loss": 0.3703,
+      "step": 5252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8106989153658622e-05,
+      "loss": 0.3567,
+      "step": 5253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8102912105248745e-05,
+      "loss": 0.3466,
+      "step": 5254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8098834817343223e-05,
+      "loss": 0.3539,
+      "step": 5255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8094757290256755e-05,
+      "loss": 0.362,
+      "step": 5256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8090679524304072e-05,
+      "loss": 0.3512,
+      "step": 5257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8086601519799905e-05,
+      "loss": 0.4036,
+      "step": 5258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8082523277059017e-05,
+      "loss": 0.3961,
+      "step": 5259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8078444796396176e-05,
+      "loss": 0.383,
+      "step": 5260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8074366078126183e-05,
+      "loss": 0.3585,
+      "step": 5261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8070287122563848e-05,
+      "loss": 0.3639,
+      "step": 5262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8066207930024e-05,
+      "loss": 0.3279,
+      "step": 5263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8062128500821492e-05,
+      "loss": 0.3099,
+      "step": 5264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.805804883527119e-05,
+      "loss": 0.3861,
+      "step": 5265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8053968933687978e-05,
+      "loss": 0.3733,
+      "step": 5266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.804988879638676e-05,
+      "loss": 0.3979,
+      "step": 5267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.804580842368246e-05,
+      "loss": 0.3704,
+      "step": 5268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8041727815890013e-05,
+      "loss": 0.3511,
+      "step": 5269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.803764697332438e-05,
+      "loss": 0.3814,
+      "step": 5270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.803356589630054e-05,
+      "loss": 0.3833,
+      "step": 5271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.802948458513349e-05,
+      "loss": 0.3807,
+      "step": 5272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8025403040138235e-05,
+      "loss": 0.3511,
+      "step": 5273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8021321261629806e-05,
+      "loss": 0.3721,
+      "step": 5274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.801723924992326e-05,
+      "loss": 0.3486,
+      "step": 5275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8013157005333653e-05,
+      "loss": 0.3256,
+      "step": 5276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8009074528176074e-05,
+      "loss": 0.3573,
+      "step": 5277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8004991818765625e-05,
+      "loss": 0.3409,
+      "step": 5278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8000908877417433e-05,
+      "loss": 0.3767,
+      "step": 5279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.799682570444663e-05,
+      "loss": 0.3578,
+      "step": 5280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.799274230016837e-05,
+      "loss": 0.3484,
+      "step": 5281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.798865866489783e-05,
+      "loss": 0.3364,
+      "step": 5282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7984574798950207e-05,
+      "loss": 0.41,
+      "step": 5283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7980490702640702e-05,
+      "loss": 0.3676,
+      "step": 5284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7976406376284547e-05,
+      "loss": 0.3876,
+      "step": 5285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7972321820196983e-05,
+      "loss": 0.3805,
+      "step": 5286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7968237034693283e-05,
+      "loss": 0.3658,
+      "step": 5287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.796415202008872e-05,
+      "loss": 0.3498,
+      "step": 5288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7960066776698592e-05,
+      "loss": 0.3449,
+      "step": 5289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7955981304838217e-05,
+      "loss": 0.3856,
+      "step": 5290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7951895604822934e-05,
+      "loss": 0.4041,
+      "step": 5291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.794780967696808e-05,
+      "loss": 0.299,
+      "step": 5292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7943723521589037e-05,
+      "loss": 0.3749,
+      "step": 5293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7939637139001184e-05,
+      "loss": 0.3892,
+      "step": 5294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7935550529519935e-05,
+      "loss": 0.4027,
+      "step": 5295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7931463693460702e-05,
+      "loss": 0.3449,
+      "step": 5296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7927376631138925e-05,
+      "loss": 0.3731,
+      "step": 5297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7923289342870062e-05,
+      "loss": 0.3161,
+      "step": 5298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7919201828969586e-05,
+      "loss": 0.3271,
+      "step": 5299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7915114089752996e-05,
+      "loss": 0.3397,
+      "step": 5300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7911026125535785e-05,
+      "loss": 0.3658,
+      "step": 5301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7906937936633493e-05,
+      "loss": 0.3245,
+      "step": 5302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7902849523361653e-05,
+      "loss": 0.346,
+      "step": 5303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.789876088603584e-05,
+      "loss": 0.347,
+      "step": 5304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7894672024971616e-05,
+      "loss": 0.3751,
+      "step": 5305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7890582940484587e-05,
+      "loss": 0.3731,
+      "step": 5306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.788649363289036e-05,
+      "loss": 0.3375,
+      "step": 5307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7882404102504567e-05,
+      "loss": 0.3521,
+      "step": 5308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7878314349642857e-05,
+      "loss": 0.3511,
+      "step": 5309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.787422437462089e-05,
+      "loss": 0.3449,
+      "step": 5310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7870134177754346e-05,
+      "loss": 0.4297,
+      "step": 5311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.786604375935893e-05,
+      "loss": 0.3199,
+      "step": 5312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.786195311975035e-05,
+      "loss": 0.352,
+      "step": 5313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7857862259244347e-05,
+      "loss": 0.3734,
+      "step": 5314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7853771178156662e-05,
+      "loss": 0.4018,
+      "step": 5315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.784967987680307e-05,
+      "loss": 0.3535,
+      "step": 5316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.784558835549935e-05,
+      "loss": 0.3823,
+      "step": 5317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.78414966145613e-05,
+      "loss": 0.3593,
+      "step": 5318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.783740465430474e-05,
+      "loss": 0.3423,
+      "step": 5319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7833312475045506e-05,
+      "loss": 0.4035,
+      "step": 5320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7829220077099453e-05,
+      "loss": 0.3653,
+      "step": 5321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7825127460782437e-05,
+      "loss": 0.3922,
+      "step": 5322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7821034626410352e-05,
+      "loss": 0.3342,
+      "step": 5323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7816941574299103e-05,
+      "loss": 0.3289,
+      "step": 5324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.78128483047646e-05,
+      "loss": 0.3406,
+      "step": 5325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7808754818122783e-05,
+      "loss": 0.3732,
+      "step": 5326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7804661114689605e-05,
+      "loss": 0.3272,
+      "step": 5327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7800567194781037e-05,
+      "loss": 0.3695,
+      "step": 5328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7796473058713062e-05,
+      "loss": 0.4192,
+      "step": 5329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.779237870680168e-05,
+      "loss": 0.3965,
+      "step": 5330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.778828413936291e-05,
+      "loss": 0.3749,
+      "step": 5331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7784189356712796e-05,
+      "loss": 0.4063,
+      "step": 5332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7780094359167384e-05,
+      "loss": 0.3548,
+      "step": 5333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777599914704274e-05,
+      "loss": 0.3331,
+      "step": 5334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777190372065495e-05,
+      "loss": 0.3081,
+      "step": 5335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7767808080320128e-05,
+      "loss": 0.3734,
+      "step": 5336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.776371222635438e-05,
+      "loss": 0.3659,
+      "step": 5337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7759616159073844e-05,
+      "loss": 0.3425,
+      "step": 5338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.775551987879467e-05,
+      "loss": 0.3396,
+      "step": 5339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.775142338583303e-05,
+      "loss": 0.3888,
+      "step": 5340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7747326680505104e-05,
+      "loss": 0.3463,
+      "step": 5341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7743229763127096e-05,
+      "loss": 0.3669,
+      "step": 5342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7739132634015222e-05,
+      "loss": 0.4205,
+      "step": 5343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.773503529348572e-05,
+      "loss": 0.3108,
+      "step": 5344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.773093774185483e-05,
+      "loss": 0.347,
+      "step": 5345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7726839979438825e-05,
+      "loss": 0.3647,
+      "step": 5346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7722742006553986e-05,
+      "loss": 0.4138,
+      "step": 5347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771864382351661e-05,
+      "loss": 0.3057,
+      "step": 5348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7714545430643017e-05,
+      "loss": 0.3582,
+      "step": 5349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771044682824953e-05,
+      "loss": 0.3427,
+      "step": 5350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7706348016652503e-05,
+      "loss": 0.3561,
+      "step": 5351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770224899616829e-05,
+      "loss": 0.3164,
+      "step": 5352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7698149767113285e-05,
+      "loss": 0.4356,
+      "step": 5353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.769405032980387e-05,
+      "loss": 0.3714,
+      "step": 5354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7689950684556465e-05,
+      "loss": 0.3741,
+      "step": 5355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.768585083168749e-05,
+      "loss": 0.3611,
+      "step": 5356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.76817507715134e-05,
+      "loss": 0.2984,
+      "step": 5357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7677650504350646e-05,
+      "loss": 0.3243,
+      "step": 5358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7673550030515707e-05,
+      "loss": 0.3192,
+      "step": 5359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7669449350325066e-05,
+      "loss": 0.3529,
+      "step": 5360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7665348464095243e-05,
+      "loss": 0.3477,
+      "step": 5361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.766124737214276e-05,
+      "loss": 0.3976,
+      "step": 5362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7657146074784146e-05,
+      "loss": 0.3086,
+      "step": 5363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7653044572335963e-05,
+      "loss": 0.3858,
+      "step": 5364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.764894286511478e-05,
+      "loss": 0.3497,
+      "step": 5365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7644840953437196e-05,
+      "loss": 0.3682,
+      "step": 5366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7640738837619798e-05,
+      "loss": 0.3631,
+      "step": 5367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7636636517979206e-05,
+      "loss": 0.3541,
+      "step": 5368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7632533994832056e-05,
+      "loss": 0.3307,
+      "step": 5369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7628431268495008e-05,
+      "loss": 0.3771,
+      "step": 5370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7624328339284712e-05,
+      "loss": 0.3514,
+      "step": 5371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7620225207517854e-05,
+      "loss": 0.3534,
+      "step": 5372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7616121873511136e-05,
+      "loss": 0.3799,
+      "step": 5373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.761201833758127e-05,
+      "loss": 0.3687,
+      "step": 5374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7607914600044977e-05,
+      "loss": 0.3718,
+      "step": 5375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7603810661219e-05,
+      "loss": 0.3411,
+      "step": 5376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.759970652142011e-05,
+      "loss": 0.3688,
+      "step": 5377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.759560218096507e-05,
+      "loss": 0.3533,
+      "step": 5378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7591497640170677e-05,
+      "loss": 0.3566,
+      "step": 5379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7587392899353727e-05,
+      "loss": 0.4008,
+      "step": 5380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7583287958831057e-05,
+      "loss": 0.3536,
+      "step": 5381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7579182818919487e-05,
+      "loss": 0.3299,
+      "step": 5382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7575077479935876e-05,
+      "loss": 0.3916,
+      "step": 5383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.757097194219709e-05,
+      "loss": 0.3427,
+      "step": 5384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7566866206020015e-05,
+      "loss": 0.3299,
+      "step": 5385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7562760271721547e-05,
+      "loss": 0.3474,
+      "step": 5386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7558654139618598e-05,
+      "loss": 0.3535,
+      "step": 5387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.755454781002809e-05,
+      "loss": 0.3232,
+      "step": 5388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.755044128326698e-05,
+      "loss": 0.3307,
+      "step": 5389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7546334559652217e-05,
+      "loss": 0.4425,
+      "step": 5390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7542227639500775e-05,
+      "loss": 0.3266,
+      "step": 5391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7538120523129654e-05,
+      "loss": 0.3601,
+      "step": 5392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.753401321085584e-05,
+      "loss": 0.392,
+      "step": 5393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7529905702996367e-05,
+      "loss": 0.3503,
+      "step": 5394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7525797999868268e-05,
+      "loss": 0.3835,
+      "step": 5395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7521690101788588e-05,
+      "loss": 0.3677,
+      "step": 5396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7517582009074386e-05,
+      "loss": 0.3259,
+      "step": 5397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7513473722042757e-05,
+      "loss": 0.3568,
+      "step": 5398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7509365241010787e-05,
+      "loss": 0.3574,
+      "step": 5399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7505256566295582e-05,
+      "loss": 0.3494,
+      "step": 5400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7501147698214272e-05,
+      "loss": 0.3941,
+      "step": 5401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7497038637083996e-05,
+      "loss": 0.3975,
+      "step": 5402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.749292938322191e-05,
+      "loss": 0.3541,
+      "step": 5403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.748881993694518e-05,
+      "loss": 0.3383,
+      "step": 5404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.748471029857099e-05,
+      "loss": 0.2944,
+      "step": 5405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7480600468416538e-05,
+      "loss": 0.4125,
+      "step": 5406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.747649044679904e-05,
+      "loss": 0.3674,
+      "step": 5407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.747238023403573e-05,
+      "loss": 0.3253,
+      "step": 5408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.746826983044384e-05,
+      "loss": 0.3224,
+      "step": 5409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7464159236340636e-05,
+      "loss": 0.388,
+      "step": 5410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7460048452043388e-05,
+      "loss": 0.3681,
+      "step": 5411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7455937477869383e-05,
+      "loss": 0.3449,
+      "step": 5412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7451826314135927e-05,
+      "loss": 0.3713,
+      "step": 5413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7447714961160327e-05,
+      "loss": 0.3634,
+      "step": 5414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7443603419259928e-05,
+      "loss": 0.3378,
+      "step": 5415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7439491688752064e-05,
+      "loss": 0.3904,
+      "step": 5416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.74353797699541e-05,
+      "loss": 0.3573,
+      "step": 5417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.743126766318341e-05,
+      "loss": 0.3577,
+      "step": 5418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7427155368757388e-05,
+      "loss": 0.3174,
+      "step": 5419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7423042886993438e-05,
+      "loss": 0.3378,
+      "step": 5420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7418930218208968e-05,
+      "loss": 0.3587,
+      "step": 5421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7414817362721424e-05,
+      "loss": 0.3534,
+      "step": 5422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.741070432084824e-05,
+      "loss": 0.3246,
+      "step": 5423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.740659109290689e-05,
+      "loss": 0.372,
+      "step": 5424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.740247767921484e-05,
+      "loss": 0.3768,
+      "step": 5425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7398364080089592e-05,
+      "loss": 0.4044,
+      "step": 5426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7394250295848636e-05,
+      "loss": 0.3586,
+      "step": 5427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.739013632680951e-05,
+      "loss": 0.3396,
+      "step": 5428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7386022173289727e-05,
+      "loss": 0.5016,
+      "step": 5429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7381907835606845e-05,
+      "loss": 0.3537,
+      "step": 5430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.737779331407842e-05,
+      "loss": 0.3647,
+      "step": 5431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7373678609022038e-05,
+      "loss": 0.4416,
+      "step": 5432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7369563720755286e-05,
+      "loss": 0.3532,
+      "step": 5433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7365448649595757e-05,
+      "loss": 0.3312,
+      "step": 5434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7361333395861078e-05,
+      "loss": 0.3308,
+      "step": 5435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7357217959868883e-05,
+      "loss": 0.3735,
+      "step": 5436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.735310234193682e-05,
+      "loss": 0.4185,
+      "step": 5437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.734898654238254e-05,
+      "loss": 0.36,
+      "step": 5438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7344870561523723e-05,
+      "loss": 0.3266,
+      "step": 5439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7340754399678062e-05,
+      "loss": 0.3278,
+      "step": 5440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7336638057163255e-05,
+      "loss": 0.3431,
+      "step": 5441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7332521534297017e-05,
+      "loss": 0.3519,
+      "step": 5442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7328404831397074e-05,
+      "loss": 0.4109,
+      "step": 5443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7324287948781188e-05,
+      "loss": 0.3381,
+      "step": 5444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.73201708867671e-05,
+      "loss": 0.3685,
+      "step": 5445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7316053645672582e-05,
+      "loss": 0.3603,
+      "step": 5446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7311936225815427e-05,
+      "loss": 0.3559,
+      "step": 5447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.730781862751344e-05,
+      "loss": 0.3779,
+      "step": 5448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.730370085108442e-05,
+      "loss": 0.3266,
+      "step": 5449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7299582896846203e-05,
+      "loss": 0.3905,
+      "step": 5450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7295464765116627e-05,
+      "loss": 0.2954,
+      "step": 5451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.729134645621355e-05,
+      "loss": 0.3104,
+      "step": 5452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7287227970454836e-05,
+      "loss": 0.3739,
+      "step": 5453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7283109308158364e-05,
+      "loss": 0.3462,
+      "step": 5454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7278990469642037e-05,
+      "loss": 0.4022,
+      "step": 5455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.727487145522376e-05,
+      "loss": 0.4249,
+      "step": 5456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7270752265221458e-05,
+      "loss": 0.365,
+      "step": 5457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7266632899953065e-05,
+      "loss": 0.3254,
+      "step": 5458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7262513359736528e-05,
+      "loss": 0.3662,
+      "step": 5459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.725839364488982e-05,
+      "loss": 0.3806,
+      "step": 5460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7254273755730908e-05,
+      "loss": 0.3917,
+      "step": 5461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7250153692577785e-05,
+      "loss": 0.3537,
+      "step": 5462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7246033455748457e-05,
+      "loss": 0.3586,
+      "step": 5463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7241913045560934e-05,
+      "loss": 0.3891,
+      "step": 5464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7237792462333257e-05,
+      "loss": 0.3083,
+      "step": 5465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7233671706383463e-05,
+      "loss": 0.348,
+      "step": 5466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722955077802961e-05,
+      "loss": 0.3483,
+      "step": 5467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7225429677589765e-05,
+      "loss": 0.3299,
+      "step": 5468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722130840538202e-05,
+      "loss": 0.3962,
+      "step": 5469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.721718696172446e-05,
+      "loss": 0.3335,
+      "step": 5470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.721306534693521e-05,
+      "loss": 0.341,
+      "step": 5471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.720894356133238e-05,
+      "loss": 0.3779,
+      "step": 5472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7204821605234117e-05,
+      "loss": 0.3597,
+      "step": 5473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.720069947895856e-05,
+      "loss": 0.3573,
+      "step": 5474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7196577182823884e-05,
+      "loss": 0.3372,
+      "step": 5475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7192454717148254e-05,
+      "loss": 0.3588,
+      "step": 5476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7188332082249867e-05,
+      "loss": 0.3784,
+      "step": 5477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.718420927844692e-05,
+      "loss": 0.4025,
+      "step": 5478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.718008630605763e-05,
+      "loss": 0.3541,
+      "step": 5479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.717596316540022e-05,
+      "loss": 0.278,
+      "step": 5480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7171839856792942e-05,
+      "loss": 0.3736,
+      "step": 5481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7167716380554046e-05,
+      "loss": 0.3522,
+      "step": 5482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.716359273700179e-05,
+      "loss": 0.241,
+      "step": 5483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7159468926454466e-05,
+      "loss": 0.3778,
+      "step": 5484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.715534494923036e-05,
+      "loss": 0.3342,
+      "step": 5485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7151220805647782e-05,
+      "loss": 0.3929,
+      "step": 5486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7147096496025045e-05,
+      "loss": 0.3746,
+      "step": 5487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7142972020680484e-05,
+      "loss": 0.3902,
+      "step": 5488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7138847379932442e-05,
+      "loss": 0.3268,
+      "step": 5489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.713472257409928e-05,
+      "loss": 0.3771,
+      "step": 5490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.713059760349936e-05,
+      "loss": 0.3986,
+      "step": 5491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.712647246845107e-05,
+      "loss": 0.3799,
+      "step": 5492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.71223471692728e-05,
+      "loss": 0.3596,
+      "step": 5493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7118221706282972e-05,
+      "loss": 0.3645,
+      "step": 5494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7114096079799985e-05,
+      "loss": 0.3368,
+      "step": 5495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7109970290142288e-05,
+      "loss": 0.3769,
+      "step": 5496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710584433762832e-05,
+      "loss": 0.3428,
+      "step": 5497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.710171822257654e-05,
+      "loss": 0.3345,
+      "step": 5498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7097591945305418e-05,
+      "loss": 0.3469,
+      "step": 5499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7093465506133437e-05,
+      "loss": 0.3251,
+      "step": 5500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7089338905379097e-05,
+      "loss": 0.3434,
+      "step": 5501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7085212143360906e-05,
+      "loss": 0.3748,
+      "step": 5502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7081085220397374e-05,
+      "loss": 0.3456,
+      "step": 5503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7076958136807047e-05,
+      "loss": 0.3379,
+      "step": 5504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.707283089290846e-05,
+      "loss": 0.3218,
+      "step": 5505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7068703489020187e-05,
+      "loss": 0.366,
+      "step": 5506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.706457592546078e-05,
+      "loss": 0.4709,
+      "step": 5507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.706044820254883e-05,
+      "loss": 0.3629,
+      "step": 5508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7056320320602925e-05,
+      "loss": 0.4005,
+      "step": 5509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7052192279941683e-05,
+      "loss": 0.3302,
+      "step": 5510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7048064080883717e-05,
+      "loss": 0.3095,
+      "step": 5511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7043935723747665e-05,
+      "loss": 0.3416,
+      "step": 5512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7039807208852156e-05,
+      "loss": 0.4024,
+      "step": 5513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7035678536515856e-05,
+      "loss": 0.3972,
+      "step": 5514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.703154970705744e-05,
+      "loss": 0.3569,
+      "step": 5515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.702742072079558e-05,
+      "loss": 0.3568,
+      "step": 5516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7023291578048963e-05,
+      "loss": 0.3539,
+      "step": 5517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.70191622791363e-05,
+      "loss": 0.3304,
+      "step": 5518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.701503282437631e-05,
+      "loss": 0.3922,
+      "step": 5519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7010903214087715e-05,
+      "loss": 0.3393,
+      "step": 5520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.700677344858926e-05,
+      "loss": 0.3446,
+      "step": 5521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7002643528199693e-05,
+      "loss": 0.4018,
+      "step": 5522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6998513453237788e-05,
+      "loss": 0.3436,
+      "step": 5523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.699438322402231e-05,
+      "loss": 0.3754,
+      "step": 5524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.699025284087205e-05,
+      "loss": 0.3537,
+      "step": 5525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.698612230410581e-05,
+      "loss": 0.3202,
+      "step": 5526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.698199161404241e-05,
+      "loss": 0.3113,
+      "step": 5527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.697786077100066e-05,
+      "loss": 0.3785,
+      "step": 5528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.69737297752994e-05,
+      "loss": 0.3759,
+      "step": 5529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.696959862725748e-05,
+      "loss": 0.3761,
+      "step": 5530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.696546732719376e-05,
+      "loss": 0.3821,
+      "step": 5531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6961335875427113e-05,
+      "loss": 0.353,
+      "step": 5532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.695720427227641e-05,
+      "loss": 0.3099,
+      "step": 5533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6953072518060555e-05,
+      "loss": 0.3475,
+      "step": 5534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6948940613098452e-05,
+      "loss": 0.3624,
+      "step": 5535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6944808557709023e-05,
+      "loss": 0.3811,
+      "step": 5536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6940676352211187e-05,
+      "loss": 0.3337,
+      "step": 5537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6936543996923895e-05,
+      "loss": 0.388,
+      "step": 5538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.693241149216609e-05,
+      "loss": 0.3684,
+      "step": 5539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6928278838256742e-05,
+      "loss": 0.3481,
+      "step": 5540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6924146035514826e-05,
+      "loss": 0.3358,
+      "step": 5541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6920013084259327e-05,
+      "loss": 0.3379,
+      "step": 5542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6915879984809244e-05,
+      "loss": 0.3548,
+      "step": 5543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.691174673748359e-05,
+      "loss": 0.3294,
+      "step": 5544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6907613342601377e-05,
+      "loss": 0.3471,
+      "step": 5545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6903479800481654e-05,
+      "loss": 0.3234,
+      "step": 5546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6899346111443442e-05,
+      "loss": 0.3385,
+      "step": 5547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6895212275805817e-05,
+      "loss": 0.3945,
+      "step": 5548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6891078293887837e-05,
+      "loss": 0.3938,
+      "step": 5549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.688694416600858e-05,
+      "loss": 0.358,
+      "step": 5550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6882809892487136e-05,
+      "loss": 0.3623,
+      "step": 5551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6878675473642606e-05,
+      "loss": 0.3631,
+      "step": 5552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.68745409097941e-05,
+      "loss": 0.3863,
+      "step": 5553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6870406201260737e-05,
+      "loss": 0.3508,
+      "step": 5554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.686627134836166e-05,
+      "loss": 0.3771,
+      "step": 5555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.686213635141601e-05,
+      "loss": 0.347,
+      "step": 5556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6858001210742945e-05,
+      "loss": 0.4212,
+      "step": 5557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6853865926661626e-05,
+      "loss": 0.3285,
+      "step": 5558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6849730499491234e-05,
+      "loss": 0.338,
+      "step": 5559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6845594929550965e-05,
+      "loss": 0.3684,
+      "step": 5560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6841459217160012e-05,
+      "loss": 0.3831,
+      "step": 5561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.683732336263759e-05,
+      "loss": 0.2992,
+      "step": 5562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6833187366302916e-05,
+      "loss": 0.3181,
+      "step": 5563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6829051228475236e-05,
+      "loss": 0.3639,
+      "step": 5564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.682491494947378e-05,
+      "loss": 0.3644,
+      "step": 5565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.682077852961781e-05,
+      "loss": 0.3492,
+      "step": 5566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6816641969226587e-05,
+      "loss": 0.3494,
+      "step": 5567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6812505268619402e-05,
+      "loss": 0.3818,
+      "step": 5568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6808368428115526e-05,
+      "loss": 0.3781,
+      "step": 5569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6804231448034265e-05,
+      "loss": 0.3329,
+      "step": 5570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.680009432869493e-05,
+      "loss": 0.2949,
+      "step": 5571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6795957070416836e-05,
+      "loss": 0.3233,
+      "step": 5572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.679181967351932e-05,
+      "loss": 0.3451,
+      "step": 5573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.678768213832172e-05,
+      "loss": 0.3122,
+      "step": 5574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6783544465143386e-05,
+      "loss": 0.3809,
+      "step": 5575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6779406654303685e-05,
+      "loss": 0.3755,
+      "step": 5576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6775268706121987e-05,
+      "loss": 0.3113,
+      "step": 5577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6771130620917676e-05,
+      "loss": 0.389,
+      "step": 5578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6766992399010152e-05,
+      "loss": 0.3719,
+      "step": 5579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6762854040718814e-05,
+      "loss": 0.3178,
+      "step": 5580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6758715546363082e-05,
+      "loss": 0.3707,
+      "step": 5581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6754576916262378e-05,
+      "loss": 0.3512,
+      "step": 5582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.675043815073614e-05,
+      "loss": 0.3715,
+      "step": 5583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6746299250103815e-05,
+      "loss": 0.3234,
+      "step": 5584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.674216021468487e-05,
+      "loss": 0.4168,
+      "step": 5585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.673802104479875e-05,
+      "loss": 0.3401,
+      "step": 5586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6733881740764957e-05,
+      "loss": 0.3506,
+      "step": 5587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6729742302902968e-05,
+      "loss": 0.3617,
+      "step": 5588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6725602731532287e-05,
+      "loss": 0.3837,
+      "step": 5589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6721463026972417e-05,
+      "loss": 0.3223,
+      "step": 5590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6717323189542886e-05,
+      "loss": 0.3243,
+      "step": 5591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6713183219563213e-05,
+      "loss": 0.3557,
+      "step": 5592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.670904311735295e-05,
+      "loss": 0.3611,
+      "step": 5593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6704902883231638e-05,
+      "loss": 0.402,
+      "step": 5594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.670076251751884e-05,
+      "loss": 0.3448,
+      "step": 5595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6696622020534127e-05,
+      "loss": 0.3588,
+      "step": 5596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6692481392597085e-05,
+      "loss": 0.3668,
+      "step": 5597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.66883406340273e-05,
+      "loss": 0.3291,
+      "step": 5598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.668419974514437e-05,
+      "loss": 0.3636,
+      "step": 5599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6680058726267907e-05,
+      "loss": 0.3512,
+      "step": 5600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6675917577717538e-05,
+      "loss": 0.3508,
+      "step": 5601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.667177629981289e-05,
+      "loss": 0.3444,
+      "step": 5602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.66676348928736e-05,
+      "loss": 0.3537,
+      "step": 5603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.666349335721933e-05,
+      "loss": 0.3671,
+      "step": 5604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6659351693169732e-05,
+      "loss": 0.352,
+      "step": 5605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6655209901044478e-05,
+      "loss": 0.3857,
+      "step": 5606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6651067981163247e-05,
+      "loss": 0.3704,
+      "step": 5607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6646925933845733e-05,
+      "loss": 0.35,
+      "step": 5608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6642783759411637e-05,
+      "loss": 0.3168,
+      "step": 5609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6638641458180673e-05,
+      "loss": 0.2612,
+      "step": 5610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6634499030472544e-05,
+      "loss": 0.3701,
+      "step": 5611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6630356476607e-05,
+      "loss": 0.3527,
+      "step": 5612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.662621379690377e-05,
+      "loss": 0.3286,
+      "step": 5613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.662207099168261e-05,
+      "loss": 0.331,
+      "step": 5614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.661792806126327e-05,
+      "loss": 0.3493,
+      "step": 5615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.661378500596552e-05,
+      "loss": 0.3206,
+      "step": 5616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6609641826109146e-05,
+      "loss": 0.406,
+      "step": 5617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6605498522013934e-05,
+      "loss": 0.3822,
+      "step": 5618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.660135509399967e-05,
+      "loss": 0.3341,
+      "step": 5619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.659721154238617e-05,
+      "loss": 0.354,
+      "step": 5620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6593067867493257e-05,
+      "loss": 0.3036,
+      "step": 5621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6588924069640744e-05,
+      "loss": 0.3183,
+      "step": 5622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6584780149148475e-05,
+      "loss": 0.3678,
+      "step": 5623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6580636106336293e-05,
+      "loss": 0.3751,
+      "step": 5624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.657649194152405e-05,
+      "loss": 0.3992,
+      "step": 5625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6572347655031614e-05,
+      "loss": 0.3975,
+      "step": 5626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.656820324717885e-05,
+      "loss": 0.3535,
+      "step": 5627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6564058718285655e-05,
+      "loss": 0.3689,
+      "step": 5628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6559914068671904e-05,
+      "loss": 0.3493,
+      "step": 5629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6555769298657515e-05,
+      "loss": 0.3648,
+      "step": 5630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6551624408562386e-05,
+      "loss": 0.3671,
+      "step": 5631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.654747939870644e-05,
+      "loss": 0.3706,
+      "step": 5632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6543334269409608e-05,
+      "loss": 0.3529,
+      "step": 5633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.653918902099183e-05,
+      "loss": 0.3326,
+      "step": 5634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6535043653773047e-05,
+      "loss": 0.3325,
+      "step": 5635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6530898168073227e-05,
+      "loss": 0.3603,
+      "step": 5636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.652675256421232e-05,
+      "loss": 0.3627,
+      "step": 5637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6522606842510313e-05,
+      "loss": 0.3917,
+      "step": 5638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.651846100328719e-05,
+      "loss": 0.3814,
+      "step": 5639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6514315046862934e-05,
+      "loss": 0.3393,
+      "step": 5640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6510168973557558e-05,
+      "loss": 0.3633,
+      "step": 5641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6506022783691066e-05,
+      "loss": 0.3606,
+      "step": 5642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6501876477583488e-05,
+      "loss": 0.3251,
+      "step": 5643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.649773005555484e-05,
+      "loss": 0.3248,
+      "step": 5644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.649358351792516e-05,
+      "loss": 0.3925,
+      "step": 5645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6489436865014508e-05,
+      "loss": 0.3879,
+      "step": 5646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6485290097142938e-05,
+      "loss": 0.3554,
+      "step": 5647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6481143214630503e-05,
+      "loss": 0.335,
+      "step": 5648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6476996217797284e-05,
+      "loss": 0.4276,
+      "step": 5649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.647284910696336e-05,
+      "loss": 0.364,
+      "step": 5650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6468701882448838e-05,
+      "loss": 0.3599,
+      "step": 5651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6464554544573795e-05,
+      "loss": 0.3717,
+      "step": 5652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6460407093658346e-05,
+      "loss": 0.3672,
+      "step": 5653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6456259530022618e-05,
+      "loss": 0.3565,
+      "step": 5654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6452111853986733e-05,
+      "loss": 0.3918,
+      "step": 5655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6447964065870822e-05,
+      "loss": 0.3523,
+      "step": 5656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.644381616599503e-05,
+      "loss": 0.3292,
+      "step": 5657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6439668154679512e-05,
+      "loss": 0.284,
+      "step": 5658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6435520032244427e-05,
+      "loss": 0.3498,
+      "step": 5659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.643137179900994e-05,
+      "loss": 0.3862,
+      "step": 5660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.642722345529624e-05,
+      "loss": 0.3573,
+      "step": 5661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6423075001423504e-05,
+      "loss": 0.4041,
+      "step": 5662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6418926437711927e-05,
+      "loss": 0.3938,
+      "step": 5663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6414777764481717e-05,
+      "loss": 0.3832,
+      "step": 5664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.641062898205308e-05,
+      "loss": 0.337,
+      "step": 5665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6406480090746243e-05,
+      "loss": 0.3658,
+      "step": 5666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6402331090881426e-05,
+      "loss": 0.3831,
+      "step": 5667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.639818198277888e-05,
+      "loss": 0.3546,
+      "step": 5668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.639403276675884e-05,
+      "loss": 0.3331,
+      "step": 5669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6389883443141557e-05,
+      "loss": 0.3753,
+      "step": 5670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6385734012247298e-05,
+      "loss": 0.3544,
+      "step": 5671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6381584474396335e-05,
+      "loss": 0.3723,
+      "step": 5672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.637743482990895e-05,
+      "loss": 0.4042,
+      "step": 5673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6373285079105413e-05,
+      "loss": 0.3713,
+      "step": 5674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6369135222306034e-05,
+      "loss": 0.3361,
+      "step": 5675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6364985259831116e-05,
+      "loss": 0.3803,
+      "step": 5676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.636083519200097e-05,
+      "loss": 0.3226,
+      "step": 5677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6356685019135902e-05,
+      "loss": 0.3506,
+      "step": 5678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6352534741556254e-05,
+      "loss": 0.3882,
+      "step": 5679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.634838435958236e-05,
+      "loss": 0.4144,
+      "step": 5680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6344233873534564e-05,
+      "loss": 0.3349,
+      "step": 5681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6340083283733208e-05,
+      "loss": 0.3396,
+      "step": 5682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.633593259049866e-05,
+      "loss": 0.3521,
+      "step": 5683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.633178179415129e-05,
+      "loss": 0.3578,
+      "step": 5684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.632763089501147e-05,
+      "loss": 0.3738,
+      "step": 5685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.632347989339958e-05,
+      "loss": 0.3549,
+      "step": 5686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6319328789636014e-05,
+      "loss": 0.41,
+      "step": 5687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6315177584041177e-05,
+      "loss": 0.3202,
+      "step": 5688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6311026276935472e-05,
+      "loss": 0.3558,
+      "step": 5689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6306874868639315e-05,
+      "loss": 0.3682,
+      "step": 5690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6302723359473124e-05,
+      "loss": 0.3602,
+      "step": 5691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.629857174975734e-05,
+      "loss": 0.3877,
+      "step": 5692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6294420039812394e-05,
+      "loss": 0.3443,
+      "step": 5693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.629026822995873e-05,
+      "loss": 0.3902,
+      "step": 5694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6286116320516803e-05,
+      "loss": 0.3258,
+      "step": 5695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.628196431180708e-05,
+      "loss": 0.3113,
+      "step": 5696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6277812204150032e-05,
+      "loss": 0.3461,
+      "step": 5697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6273659997866124e-05,
+      "loss": 0.3165,
+      "step": 5698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6269507693275854e-05,
+      "loss": 0.3567,
+      "step": 5699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6265355290699704e-05,
+      "loss": 0.3465,
+      "step": 5700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6261202790458182e-05,
+      "loss": 0.3586,
+      "step": 5701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6257050192871784e-05,
+      "loss": 0.3132,
+      "step": 5702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6252897498261036e-05,
+      "loss": 0.3649,
+      "step": 5703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6248744706946454e-05,
+      "loss": 0.3284,
+      "step": 5704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.624459181924857e-05,
+      "loss": 0.3527,
+      "step": 5705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6240438835487917e-05,
+      "loss": 0.3133,
+      "step": 5706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6236285755985052e-05,
+      "loss": 0.3185,
+      "step": 5707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.623213258106051e-05,
+      "loss": 0.3321,
+      "step": 5708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6227979311034863e-05,
+      "loss": 0.3193,
+      "step": 5709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.622382594622867e-05,
+      "loss": 0.3335,
+      "step": 5710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6219672486962508e-05,
+      "loss": 0.3656,
+      "step": 5711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.621551893355696e-05,
+      "loss": 0.3789,
+      "step": 5712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6211365286332613e-05,
+      "loss": 0.3822,
+      "step": 5713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.620721154561006e-05,
+      "loss": 0.3297,
+      "step": 5714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6203057711709906e-05,
+      "loss": 0.3467,
+      "step": 5715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.619890378495276e-05,
+      "loss": 0.3758,
+      "step": 5716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6194749765659248e-05,
+      "loss": 0.3408,
+      "step": 5717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6190595654149983e-05,
+      "loss": 0.3871,
+      "step": 5718
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6186441450745598e-05,
+      "loss": 0.3747,
+      "step": 5719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6182287155766734e-05,
+      "loss": 0.3623,
+      "step": 5720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6178132769534045e-05,
+      "loss": 0.3626,
+      "step": 5721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.617397829236817e-05,
+      "loss": 0.3533,
+      "step": 5722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6169823724589776e-05,
+      "loss": 0.364,
+      "step": 5723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6165669066519525e-05,
+      "loss": 0.3627,
+      "step": 5724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6161514318478102e-05,
+      "loss": 0.371,
+      "step": 5725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6157359480786177e-05,
+      "loss": 0.3806,
+      "step": 5726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6153204553764436e-05,
+      "loss": 0.4207,
+      "step": 5727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6149049537733583e-05,
+      "loss": 0.3564,
+      "step": 5728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6144894433014314e-05,
+      "loss": 0.3371,
+      "step": 5729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.614073923992734e-05,
+      "loss": 0.3779,
+      "step": 5730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613658395879337e-05,
+      "loss": 0.3114,
+      "step": 5731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613242858993313e-05,
+      "loss": 0.385,
+      "step": 5732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612827313366735e-05,
+      "loss": 0.365,
+      "step": 5733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6124117590316763e-05,
+      "loss": 0.3392,
+      "step": 5734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.611996196020211e-05,
+      "loss": 0.3241,
+      "step": 5735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6115806243644144e-05,
+      "loss": 0.3708,
+      "step": 5736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6111650440963618e-05,
+      "loss": 0.3802,
+      "step": 5737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6107494552481297e-05,
+      "loss": 0.3148,
+      "step": 5738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6103338578517942e-05,
+      "loss": 0.3936,
+      "step": 5739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6099182519394336e-05,
+      "loss": 0.3838,
+      "step": 5740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6095026375431256e-05,
+      "loss": 0.3682,
+      "step": 5741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6090870146949493e-05,
+      "loss": 0.3423,
+      "step": 5742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6086713834269847e-05,
+      "loss": 0.364,
+      "step": 5743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.608255743771311e-05,
+      "loss": 0.3578,
+      "step": 5744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.607840095760009e-05,
+      "loss": 0.3244,
+      "step": 5745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.607424439425161e-05,
+      "loss": 0.3361,
+      "step": 5746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6070087747988485e-05,
+      "loss": 0.35,
+      "step": 5747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6065931019131547e-05,
+      "loss": 0.368,
+      "step": 5748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.606177420800162e-05,
+      "loss": 0.3636,
+      "step": 5749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6057617314919555e-05,
+      "loss": 0.3711,
+      "step": 5750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.605346034020619e-05,
+      "loss": 0.3619,
+      "step": 5751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6049303284182387e-05,
+      "loss": 0.3631,
+      "step": 5752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604514614716899e-05,
+      "loss": 0.3537,
+      "step": 5753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604098892948688e-05,
+      "loss": 0.3388,
+      "step": 5754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603683163145692e-05,
+      "loss": 0.385,
+      "step": 5755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603267425339998e-05,
+      "loss": 0.3413,
+      "step": 5756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602851679563696e-05,
+      "loss": 0.3699,
+      "step": 5757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6024359258488743e-05,
+      "loss": 0.3441,
+      "step": 5758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6020201642276223e-05,
+      "loss": 0.3668,
+      "step": 5759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.60160439473203e-05,
+      "loss": 0.3793,
+      "step": 5760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.601188617394189e-05,
+      "loss": 0.3719,
+      "step": 5761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.60077283224619e-05,
+      "loss": 0.3633,
+      "step": 5762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6003570393201258e-05,
+      "loss": 0.3942,
+      "step": 5763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.599941238648088e-05,
+      "loss": 0.3337,
+      "step": 5764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.59952543026217e-05,
+      "loss": 0.3401,
+      "step": 5765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5991096141944666e-05,
+      "loss": 0.4229,
+      "step": 5766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5986937904770717e-05,
+      "loss": 0.3554,
+      "step": 5767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5982779591420797e-05,
+      "loss": 0.3235,
+      "step": 5768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5978621202215866e-05,
+      "loss": 0.4046,
+      "step": 5769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.597446273747689e-05,
+      "loss": 0.4321,
+      "step": 5770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5970304197524834e-05,
+      "loss": 0.3525,
+      "step": 5771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5966145582680665e-05,
+      "loss": 0.3741,
+      "step": 5772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5961986893265367e-05,
+      "loss": 0.3316,
+      "step": 5773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5957828129599932e-05,
+      "loss": 0.3887,
+      "step": 5774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5953669292005343e-05,
+      "loss": 0.4007,
+      "step": 5775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5949510380802592e-05,
+      "loss": 0.313,
+      "step": 5776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5945351396312687e-05,
+      "loss": 0.3352,
+      "step": 5777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.594119233885664e-05,
+      "loss": 0.4013,
+      "step": 5778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5937033208755458e-05,
+      "loss": 0.3573,
+      "step": 5779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5932874006330162e-05,
+      "loss": 0.3551,
+      "step": 5780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.592871473190177e-05,
+      "loss": 0.3562,
+      "step": 5781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5924555385791324e-05,
+      "loss": 0.3955,
+      "step": 5782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5920395968319853e-05,
+      "loss": 0.3798,
+      "step": 5783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.59162364798084e-05,
+      "loss": 0.3396,
+      "step": 5784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5912076920578007e-05,
+      "loss": 0.3355,
+      "step": 5785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.590791729094973e-05,
+      "loss": 0.3792,
+      "step": 5786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5903757591244624e-05,
+      "loss": 0.3282,
+      "step": 5787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5899597821783753e-05,
+      "loss": 0.3843,
+      "step": 5788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5895437982888194e-05,
+      "loss": 0.3175,
+      "step": 5789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5891278074879005e-05,
+      "loss": 0.3623,
+      "step": 5790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5887118098077272e-05,
+      "loss": 0.345,
+      "step": 5791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5882958052804087e-05,
+      "loss": 0.3073,
+      "step": 5792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.587879793938053e-05,
+      "loss": 0.3686,
+      "step": 5793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5874637758127695e-05,
+      "loss": 0.3827,
+      "step": 5794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5870477509366685e-05,
+      "loss": 0.3924,
+      "step": 5795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5866317193418612e-05,
+      "loss": 0.3943,
+      "step": 5796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5862156810604576e-05,
+      "loss": 0.3461,
+      "step": 5797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5857996361245697e-05,
+      "loss": 0.342,
+      "step": 5798
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.585383584566309e-05,
+      "loss": 0.3649,
+      "step": 5799
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.58496752641779e-05,
+      "loss": 0.3208,
+      "step": 5800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5845514617111235e-05,
+      "loss": 0.3304,
+      "step": 5801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.584135390478424e-05,
+      "loss": 0.3231,
+      "step": 5802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5837193127518058e-05,
+      "loss": 0.367,
+      "step": 5803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5833032285633838e-05,
+      "loss": 0.3397,
+      "step": 5804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5828871379452723e-05,
+      "loss": 0.3049,
+      "step": 5805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.582471040929587e-05,
+      "loss": 0.3392,
+      "step": 5806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.582054937548444e-05,
+      "loss": 0.3341,
+      "step": 5807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5816388278339614e-05,
+      "loss": 0.3955,
+      "step": 5808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5812227118182538e-05,
+      "loss": 0.3348,
+      "step": 5809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5808065895334404e-05,
+      "loss": 0.3558,
+      "step": 5810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5803904610116386e-05,
+      "loss": 0.3225,
+      "step": 5811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5799743262849677e-05,
+      "loss": 0.3603,
+      "step": 5812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.579558185385546e-05,
+      "loss": 0.3501,
+      "step": 5813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5791420383454926e-05,
+      "loss": 0.406,
+      "step": 5814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5787258851969285e-05,
+      "loss": 0.3199,
+      "step": 5815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5783097259719736e-05,
+      "loss": 0.392,
+      "step": 5816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5778935607027488e-05,
+      "loss": 0.3392,
+      "step": 5817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.577477389421375e-05,
+      "loss": 0.366,
+      "step": 5818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5770612121599752e-05,
+      "loss": 0.4207,
+      "step": 5819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.576645028950671e-05,
+      "loss": 0.3692,
+      "step": 5820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5762288398255847e-05,
+      "loss": 0.323,
+      "step": 5821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57581264481684e-05,
+      "loss": 0.3231,
+      "step": 5822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5753964439565607e-05,
+      "loss": 0.3863,
+      "step": 5823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.574980237276871e-05,
+      "loss": 0.3531,
+      "step": 5824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5745640248098945e-05,
+      "loss": 0.3689,
+      "step": 5825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5741478065877572e-05,
+      "loss": 0.3529,
+      "step": 5826
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.573731582642585e-05,
+      "loss": 0.357,
+      "step": 5827
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5733153530065018e-05,
+      "loss": 0.3429,
+      "step": 5828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.572899117711636e-05,
+      "loss": 0.3324,
+      "step": 5829
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5724828767901133e-05,
+      "loss": 0.3429,
+      "step": 5830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.572066630274061e-05,
+      "loss": 0.2742,
+      "step": 5831
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.571650378195607e-05,
+      "loss": 0.3052,
+      "step": 5832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5712341205868794e-05,
+      "loss": 0.2695,
+      "step": 5833
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5708178574800068e-05,
+      "loss": 0.2507,
+      "step": 5834
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.570401588907117e-05,
+      "loss": 0.27,
+      "step": 5835
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5699853149003402e-05,
+      "loss": 0.2549,
+      "step": 5836
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.569569035491807e-05,
+      "loss": 0.2489,
+      "step": 5837
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5691527507136465e-05,
+      "loss": 0.2364,
+      "step": 5838
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5687364605979893e-05,
+      "loss": 0.2563,
+      "step": 5839
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.568320165176966e-05,
+      "loss": 0.249,
+      "step": 5840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5679038644827098e-05,
+      "loss": 0.2717,
+      "step": 5841
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.567487558547351e-05,
+      "loss": 0.252,
+      "step": 5842
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5670712474030214e-05,
+      "loss": 0.2646,
+      "step": 5843
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5666549310818547e-05,
+      "loss": 0.2414,
+      "step": 5844
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.566238609615984e-05,
+      "loss": 0.243,
+      "step": 5845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5658222830375423e-05,
+      "loss": 0.2114,
+      "step": 5846
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.565405951378663e-05,
+      "loss": 0.2573,
+      "step": 5847
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5649896146714807e-05,
+      "loss": 0.2661,
+      "step": 5848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5645732729481307e-05,
+      "loss": 0.2849,
+      "step": 5849
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5641569262407468e-05,
+      "loss": 0.2314,
+      "step": 5850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.563740574581465e-05,
+      "loss": 0.2779,
+      "step": 5851
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5633242180024208e-05,
+      "loss": 0.2292,
+      "step": 5852
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5629078565357506e-05,
+      "loss": 0.2542,
+      "step": 5853
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5624914902135908e-05,
+      "loss": 0.2725,
+      "step": 5854
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5620751190680785e-05,
+      "loss": 0.2607,
+      "step": 5855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.56165874313135e-05,
+      "loss": 0.2655,
+      "step": 5856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5612423624355445e-05,
+      "loss": 0.2766,
+      "step": 5857
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.560825977012799e-05,
+      "loss": 0.2745,
+      "step": 5858
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5604095868952515e-05,
+      "loss": 0.297,
+      "step": 5859
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5599931921150415e-05,
+      "loss": 0.2665,
+      "step": 5860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5595767927043073e-05,
+      "loss": 0.2878,
+      "step": 5861
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5591603886951892e-05,
+      "loss": 0.2788,
+      "step": 5862
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5587439801198266e-05,
+      "loss": 0.2665,
+      "step": 5863
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5583275670103596e-05,
+      "loss": 0.3084,
+      "step": 5864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5579111493989288e-05,
+      "loss": 0.2147,
+      "step": 5865
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.557494727317675e-05,
+      "loss": 0.2727,
+      "step": 5866
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5570783007987392e-05,
+      "loss": 0.2336,
+      "step": 5867
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5566618698742632e-05,
+      "loss": 0.1994,
+      "step": 5868
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5562454345763887e-05,
+      "loss": 0.24,
+      "step": 5869
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.555828994937258e-05,
+      "loss": 0.2516,
+      "step": 5870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5554125509890137e-05,
+      "loss": 0.2396,
+      "step": 5871
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5549961027637986e-05,
+      "loss": 0.254,
+      "step": 5872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5545796502937558e-05,
+      "loss": 0.2139,
+      "step": 5873
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.554163193611029e-05,
+      "loss": 0.2667,
+      "step": 5874
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.553746732747762e-05,
+      "loss": 0.2777,
+      "step": 5875
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.553330267736099e-05,
+      "loss": 0.2418,
+      "step": 5876
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.552913798608184e-05,
+      "loss": 0.2402,
+      "step": 5877
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5524973253961627e-05,
+      "loss": 0.2409,
+      "step": 5878
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.55208084813218e-05,
+      "loss": 0.2382,
+      "step": 5879
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5516643668483807e-05,
+      "loss": 0.2633,
+      "step": 5880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5512478815769108e-05,
+      "loss": 0.2436,
+      "step": 5881
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550831392349917e-05,
+      "loss": 0.2468,
+      "step": 5882
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5504148991995454e-05,
+      "loss": 0.2772,
+      "step": 5883
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549998402157942e-05,
+      "loss": 0.2673,
+      "step": 5884
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495819012572544e-05,
+      "loss": 0.2556,
+      "step": 5885
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54916539652963e-05,
+      "loss": 0.2994,
+      "step": 5886
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5487488880072163e-05,
+      "loss": 0.2443,
+      "step": 5887
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54833237572216e-05,
+      "loss": 0.2478,
+      "step": 5888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479158597066112e-05,
+      "loss": 0.2445,
+      "step": 5889
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547499339992717e-05,
+      "loss": 0.2392,
+      "step": 5890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5470828166126266e-05,
+      "loss": 0.2414,
+      "step": 5891
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5466662895984887e-05,
+      "loss": 0.2626,
+      "step": 5892
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5462497589824527e-05,
+      "loss": 0.292,
+      "step": 5893
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545833224796668e-05,
+      "loss": 0.2731,
+      "step": 5894
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454166870732857e-05,
+      "loss": 0.2337,
+      "step": 5895
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545000145844454e-05,
+      "loss": 0.2215,
+      "step": 5896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544583601142324e-05,
+      "loss": 0.2614,
+      "step": 5897
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544167052999047e-05,
+      "loss": 0.2381,
+      "step": 5898
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437505014467732e-05,
+      "loss": 0.2731,
+      "step": 5899
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5433339465176542e-05,
+      "loss": 0.2424,
+      "step": 5900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5429173882438407e-05,
+      "loss": 0.3107,
+      "step": 5901
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542500826657486e-05,
+      "loss": 0.2662,
+      "step": 5902
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5420842617907402e-05,
+      "loss": 0.2221,
+      "step": 5903
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5416676936757564e-05,
+      "loss": 0.2614,
+      "step": 5904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541251122344687e-05,
+      "loss": 0.236,
+      "step": 5905
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5408345478296845e-05,
+      "loss": 0.2826,
+      "step": 5906
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5404179701629022e-05,
+      "loss": 0.3073,
+      "step": 5907
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5400013893764928e-05,
+      "loss": 0.2468,
+      "step": 5908
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5395848055026107e-05,
+      "loss": 0.2457,
+      "step": 5909
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5391682185734082e-05,
+      "loss": 0.2303,
+      "step": 5910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5387516286210405e-05,
+      "loss": 0.2467,
+      "step": 5911
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5383350356776605e-05,
+      "loss": 0.2493,
+      "step": 5912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5379184397754237e-05,
+      "loss": 0.2387,
+      "step": 5913
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.537501840946484e-05,
+      "loss": 0.2587,
+      "step": 5914
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.537085239222997e-05,
+      "loss": 0.2775,
+      "step": 5915
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5366686346371167e-05,
+      "loss": 0.2445,
+      "step": 5916
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5362520272209994e-05,
+      "loss": 0.2416,
+      "step": 5917
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358354170067993e-05,
+      "loss": 0.2796,
+      "step": 5918
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5354188040266733e-05,
+      "loss": 0.2435,
+      "step": 5919
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5350021883127774e-05,
+      "loss": 0.2743,
+      "step": 5920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5345855698972664e-05,
+      "loss": 0.2423,
+      "step": 5921
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534168948812298e-05,
+      "loss": 0.2451,
+      "step": 5922
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533752325090028e-05,
+      "loss": 0.2386,
+      "step": 5923
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333356987626136e-05,
+      "loss": 0.2796,
+      "step": 5924
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329190698622114e-05,
+      "loss": 0.2313,
+      "step": 5925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5325024384209787e-05,
+      "loss": 0.2566,
+      "step": 5926
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5320858044710725e-05,
+      "loss": 0.3128,
+      "step": 5927
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5316691680446514e-05,
+      "loss": 0.3122,
+      "step": 5928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5312525291738718e-05,
+      "loss": 0.2309,
+      "step": 5929
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.530835887890892e-05,
+      "loss": 0.2272,
+      "step": 5930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.530419244227871e-05,
+      "loss": 0.2871,
+      "step": 5931
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.530002598216966e-05,
+      "loss": 0.2498,
+      "step": 5932
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5295859498903356e-05,
+      "loss": 0.3078,
+      "step": 5933
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.529169299280139e-05,
+      "loss": 0.2887,
+      "step": 5934
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5287526464185348e-05,
+      "loss": 0.264,
+      "step": 5935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.528335991337682e-05,
+      "loss": 0.2683,
+      "step": 5936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5279193340697395e-05,
+      "loss": 0.2339,
+      "step": 5937
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.527502674646867e-05,
+      "loss": 0.2451,
+      "step": 5938
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.527086013101224e-05,
+      "loss": 0.2861,
+      "step": 5939
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5266693494649692e-05,
+      "loss": 0.2738,
+      "step": 5940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.526252683770264e-05,
+      "loss": 0.2839,
+      "step": 5941
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5258360160492677e-05,
+      "loss": 0.2219,
+      "step": 5942
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5254193463341403e-05,
+      "loss": 0.2835,
+      "step": 5943
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5250026746570417e-05,
+      "loss": 0.2281,
+      "step": 5944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5245860010501336e-05,
+      "loss": 0.2757,
+      "step": 5945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5241693255455756e-05,
+      "loss": 0.2627,
+      "step": 5946
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5237526481755283e-05,
+      "loss": 0.2388,
+      "step": 5947
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.523335968972153e-05,
+      "loss": 0.2383,
+      "step": 5948
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.522919287967611e-05,
+      "loss": 0.2613,
+      "step": 5949
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5225026051940633e-05,
+      "loss": 0.2462,
+      "step": 5950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5220859206836704e-05,
+      "loss": 0.2422,
+      "step": 5951
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5216692344685949e-05,
+      "loss": 0.2781,
+      "step": 5952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5212525465809981e-05,
+      "loss": 0.2363,
+      "step": 5953
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5208358570530412e-05,
+      "loss": 0.242,
+      "step": 5954
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5204191659168864e-05,
+      "loss": 0.2498,
+      "step": 5955
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5200024732046952e-05,
+      "loss": 0.2529,
+      "step": 5956
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5195857789486303e-05,
+      "loss": 0.2674,
+      "step": 5957
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5191690831808541e-05,
+      "loss": 0.2683,
+      "step": 5958
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5187523859335277e-05,
+      "loss": 0.2632,
+      "step": 5959
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5183356872388147e-05,
+      "loss": 0.2494,
+      "step": 5960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5179189871288773e-05,
+      "loss": 0.2661,
+      "step": 5961
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.517502285635878e-05,
+      "loss": 0.2734,
+      "step": 5962
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5170855827919794e-05,
+      "loss": 0.265,
+      "step": 5963
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5166688786293443e-05,
+      "loss": 0.2406,
+      "step": 5964
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5162521731801364e-05,
+      "loss": 0.2521,
+      "step": 5965
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5158354664765185e-05,
+      "loss": 0.2713,
+      "step": 5966
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5154187585506531e-05,
+      "loss": 0.2515,
+      "step": 5967
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5150020494347036e-05,
+      "loss": 0.287,
+      "step": 5968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5145853391608344e-05,
+      "loss": 0.2803,
+      "step": 5969
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5141686277612076e-05,
+      "loss": 0.2716,
+      "step": 5970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5137519152679875e-05,
+      "loss": 0.3084,
+      "step": 5971
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5133352017133371e-05,
+      "loss": 0.2135,
+      "step": 5972
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5129184871294209e-05,
+      "loss": 0.2327,
+      "step": 5973
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5125017715484023e-05,
+      "loss": 0.2478,
+      "step": 5974
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5120850550024445e-05,
+      "loss": 0.2298,
+      "step": 5975
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5116683375237126e-05,
+      "loss": 0.2594,
+      "step": 5976
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5112516191443693e-05,
+      "loss": 0.2587,
+      "step": 5977
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5108348998965798e-05,
+      "loss": 0.2268,
+      "step": 5978
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5104181798125073e-05,
+      "loss": 0.2228,
+      "step": 5979
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.510001458924317e-05,
+      "loss": 0.239,
+      "step": 5980
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5095847372641717e-05,
+      "loss": 0.2447,
+      "step": 5981
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5091680148642373e-05,
+      "loss": 0.2116,
+      "step": 5982
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5087512917566774e-05,
+      "loss": 0.2562,
+      "step": 5983
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5083345679736564e-05,
+      "loss": 0.2323,
+      "step": 5984
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5079178435473385e-05,
+      "loss": 0.2468,
+      "step": 5985
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5075011185098888e-05,
+      "loss": 0.2751,
+      "step": 5986
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5070843928934715e-05,
+      "loss": 0.275,
+      "step": 5987
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5066676667302517e-05,
+      "loss": 0.2063,
+      "step": 5988
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.506250940052393e-05,
+      "loss": 0.2475,
+      "step": 5989
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5058342128920615e-05,
+      "loss": 0.2744,
+      "step": 5990
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5054174852814213e-05,
+      "loss": 0.2494,
+      "step": 5991
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5050007572526365e-05,
+      "loss": 0.2456,
+      "step": 5992
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5045840288378723e-05,
+      "loss": 0.254,
+      "step": 5993
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5041673000692944e-05,
+      "loss": 0.2742,
+      "step": 5994
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5037505709790667e-05,
+      "loss": 0.2543,
+      "step": 5995
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5033338415993542e-05,
+      "loss": 0.2659,
+      "step": 5996
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5029171119623218e-05,
+      "loss": 0.2461,
+      "step": 5997
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5025003821001352e-05,
+      "loss": 0.2645,
+      "step": 5998
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5020836520449585e-05,
+      "loss": 0.2678,
+      "step": 5999
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5016669218289566e-05,
+      "loss": 0.2279,
+      "step": 6000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.501250191484295e-05,
+      "loss": 0.2962,
+      "step": 6001
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5008334610431383e-05,
+      "loss": 0.25,
+      "step": 6002
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5004167305376519e-05,
+      "loss": 0.2611,
+      "step": 6003
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5e-05,
+      "loss": 0.2722,
+      "step": 6004
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4995832694623486e-05,
+      "loss": 0.267,
+      "step": 6005
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4991665389568618e-05,
+      "loss": 0.3053,
+      "step": 6006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4987498085157055e-05,
+      "loss": 0.2548,
+      "step": 6007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.498333078171044e-05,
+      "loss": 0.2436,
+      "step": 6008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4979163479550418e-05,
+      "loss": 0.24,
+      "step": 6009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4974996178998652e-05,
+      "loss": 0.2272,
+      "step": 6010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4970828880376781e-05,
+      "loss": 0.2937,
+      "step": 6011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4966661584006461e-05,
+      "loss": 0.2525,
+      "step": 6012
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4962494290209339e-05,
+      "loss": 0.2497,
+      "step": 6013
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.495832699930706e-05,
+      "loss": 0.2468,
+      "step": 6014
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4954159711621281e-05,
+      "loss": 0.2901,
+      "step": 6015
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4949992427473643e-05,
+      "loss": 0.2295,
+      "step": 6016
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4945825147185792e-05,
+      "loss": 0.2903,
+      "step": 6017
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4941657871079388e-05,
+      "loss": 0.2355,
+      "step": 6018
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.493749059947607e-05,
+      "loss": 0.2645,
+      "step": 6019
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4933323332697487e-05,
+      "loss": 0.2644,
+      "step": 6020
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4929156071065287e-05,
+      "loss": 0.2965,
+      "step": 6021
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4924988814901116e-05,
+      "loss": 0.262,
+      "step": 6022
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.492082156452662e-05,
+      "loss": 0.2111,
+      "step": 6023
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4916654320263442e-05,
+      "loss": 0.2565,
+      "step": 6024
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4912487082433227e-05,
+      "loss": 0.2491,
+      "step": 6025
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.490831985135763e-05,
+      "loss": 0.267,
+      "step": 6026
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4904152627358284e-05,
+      "loss": 0.2354,
+      "step": 6027
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4899985410756834e-05,
+      "loss": 0.225,
+      "step": 6028
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.489581820187493e-05,
+      "loss": 0.2838,
+      "step": 6029
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4891651001034203e-05,
+      "loss": 0.2454,
+      "step": 6030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.488748380855631e-05,
+      "loss": 0.2439,
+      "step": 6031
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4883316624762882e-05,
+      "loss": 0.2763,
+      "step": 6032
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4879149449975554e-05,
+      "loss": 0.2713,
+      "step": 6033
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4874982284515982e-05,
+      "loss": 0.2184,
+      "step": 6034
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4870815128705793e-05,
+      "loss": 0.2543,
+      "step": 6035
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4866647982866628e-05,
+      "loss": 0.2552,
+      "step": 6036
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4862480847320128e-05,
+      "loss": 0.224,
+      "step": 6037
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.485831372238793e-05,
+      "loss": 0.267,
+      "step": 6038
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.485414660839166e-05,
+      "loss": 0.2699,
+      "step": 6039
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4849979505652966e-05,
+      "loss": 0.2293,
+      "step": 6040
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.484581241449347e-05,
+      "loss": 0.2223,
+      "step": 6041
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4841645335234819e-05,
+      "loss": 0.2681,
+      "step": 6042
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4837478268198637e-05,
+      "loss": 0.2612,
+      "step": 6043
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4833311213706556e-05,
+      "loss": 0.2818,
+      "step": 6044
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.482914417208021e-05,
+      "loss": 0.2502,
+      "step": 6045
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4824977143641226e-05,
+      "loss": 0.293,
+      "step": 6046
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.482081012871123e-05,
+      "loss": 0.255,
+      "step": 6047
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4816643127611857e-05,
+      "loss": 0.2712,
+      "step": 6048
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4812476140664727e-05,
+      "loss": 0.2645,
+      "step": 6049
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4808309168191461e-05,
+      "loss": 0.2268,
+      "step": 6050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4804142210513697e-05,
+      "loss": 0.2695,
+      "step": 6051
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4799975267953048e-05,
+      "loss": 0.2461,
+      "step": 6052
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.479580834083114e-05,
+      "loss": 0.2349,
+      "step": 6053
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4791641429469594e-05,
+      "loss": 0.2591,
+      "step": 6054
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.478747453419002e-05,
+      "loss": 0.2472,
+      "step": 6055
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4783307655314053e-05,
+      "loss": 0.2421,
+      "step": 6056
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.47791407931633e-05,
+      "loss": 0.258,
+      "step": 6057
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4774973948059371e-05,
+      "loss": 0.2564,
+      "step": 6058
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4770807120323892e-05,
+      "loss": 0.2784,
+      "step": 6059
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4766640310278468e-05,
+      "loss": 0.2696,
+      "step": 6060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.476247351824472e-05,
+      "loss": 0.2903,
+      "step": 6061
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.475830674454425e-05,
+      "loss": 0.2572,
+      "step": 6062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4754139989498664e-05,
+      "loss": 0.2596,
+      "step": 6063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4749973253429583e-05,
+      "loss": 0.2603,
+      "step": 6064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4745806536658603e-05,
+      "loss": 0.2562,
+      "step": 6065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4741639839507323e-05,
+      "loss": 0.2347,
+      "step": 6066
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.473747316229736e-05,
+      "loss": 0.246,
+      "step": 6067
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4733306505350307e-05,
+      "loss": 0.253,
+      "step": 6068
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4729139868987764e-05,
+      "loss": 0.2777,
+      "step": 6069
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4724973253531334e-05,
+      "loss": 0.2604,
+      "step": 6070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4720806659302602e-05,
+      "loss": 0.3111,
+      "step": 6071
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4716640086623181e-05,
+      "loss": 0.216,
+      "step": 6072
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4712473535814658e-05,
+      "loss": 0.2327,
+      "step": 6073
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.470830700719861e-05,
+      "loss": 0.2485,
+      "step": 6074
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4704140501096645e-05,
+      "loss": 0.2639,
+      "step": 6075
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4699974017830346e-05,
+      "loss": 0.2546,
+      "step": 6076
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4695807557721295e-05,
+      "loss": 0.2546,
+      "step": 6077
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4691641121091082e-05,
+      "loss": 0.2585,
+      "step": 6078
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.468747470826129e-05,
+      "loss": 0.2575,
+      "step": 6079
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4683308319553488e-05,
+      "loss": 0.2454,
+      "step": 6080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4679141955289279e-05,
+      "loss": 0.2407,
+      "step": 6081
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4674975615790214e-05,
+      "loss": 0.2666,
+      "step": 6082
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4670809301377888e-05,
+      "loss": 0.2611,
+      "step": 6083
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4666643012373868e-05,
+      "loss": 0.2678,
+      "step": 6084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4662476749099721e-05,
+      "loss": 0.2819,
+      "step": 6085
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4658310511877023e-05,
+      "loss": 0.252,
+      "step": 6086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4654144301027339e-05,
+      "loss": 0.2254,
+      "step": 6087
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4649978116872227e-05,
+      "loss": 0.2354,
+      "step": 6088
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.464581195973327e-05,
+      "loss": 0.2143,
+      "step": 6089
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4641645829932005e-05,
+      "loss": 0.258,
+      "step": 6090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.463747972779001e-05,
+      "loss": 0.272,
+      "step": 6091
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4633313653628835e-05,
+      "loss": 0.2535,
+      "step": 6092
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4629147607770032e-05,
+      "loss": 0.2461,
+      "step": 6093
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4624981590535161e-05,
+      "loss": 0.2958,
+      "step": 6094
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4620815602245767e-05,
+      "loss": 0.2349,
+      "step": 6095
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4616649643223392e-05,
+      "loss": 0.246,
+      "step": 6096
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4612483713789597e-05,
+      "loss": 0.2645,
+      "step": 6097
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4608317814265923e-05,
+      "loss": 0.2556,
+      "step": 6098
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4604151944973897e-05,
+      "loss": 0.2797,
+      "step": 6099
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4599986106235073e-05,
+      "loss": 0.2265,
+      "step": 6100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.459582029837098e-05,
+      "loss": 0.2256,
+      "step": 6101
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4591654521703158e-05,
+      "loss": 0.2517,
+      "step": 6102
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4587488776553136e-05,
+      "loss": 0.2573,
+      "step": 6103
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4583323063242437e-05,
+      "loss": 0.2604,
+      "step": 6104
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4579157382092602e-05,
+      "loss": 0.274,
+      "step": 6105
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4574991733425148e-05,
+      "loss": 0.2311,
+      "step": 6106
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.457082611756159e-05,
+      "loss": 0.2494,
+      "step": 6107
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4566660534823462e-05,
+      "loss": 0.2791,
+      "step": 6108
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.456249498553227e-05,
+      "loss": 0.2519,
+      "step": 6109
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4558329470009533e-05,
+      "loss": 0.2705,
+      "step": 6110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4554163988576763e-05,
+      "loss": 0.2529,
+      "step": 6111
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4549998541555462e-05,
+      "loss": 0.2514,
+      "step": 6112
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4545833129267147e-05,
+      "loss": 0.2608,
+      "step": 6113
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4541667752033318e-05,
+      "loss": 0.2555,
+      "step": 6114
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4537502410175474e-05,
+      "loss": 0.2597,
+      "step": 6115
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4533337104015115e-05,
+      "loss": 0.2306,
+      "step": 6116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4529171833873739e-05,
+      "loss": 0.2928,
+      "step": 6117
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4525006600072833e-05,
+      "loss": 0.2298,
+      "step": 6118
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4520841402933892e-05,
+      "loss": 0.263,
+      "step": 6119
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4516676242778395e-05,
+      "loss": 0.2775,
+      "step": 6120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4512511119927841e-05,
+      "loss": 0.2412,
+      "step": 6121
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4508346034703703e-05,
+      "loss": 0.2669,
+      "step": 6122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4504180987427455e-05,
+      "loss": 0.2476,
+      "step": 6123
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4500015978420583e-05,
+      "loss": 0.2828,
+      "step": 6124
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4495851008004552e-05,
+      "loss": 0.2524,
+      "step": 6125
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4491686076500833e-05,
+      "loss": 0.2655,
+      "step": 6126
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4487521184230895e-05,
+      "loss": 0.2313,
+      "step": 6127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.44833563315162e-05,
+      "loss": 0.2743,
+      "step": 6128
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4479191518678205e-05,
+      "loss": 0.2424,
+      "step": 6129
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4475026746038375e-05,
+      "loss": 0.2706,
+      "step": 6130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4470862013918159e-05,
+      "loss": 0.2446,
+      "step": 6131
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4466697322639014e-05,
+      "loss": 0.2524,
+      "step": 6132
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4462532672522385e-05,
+      "loss": 0.2326,
+      "step": 6133
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4458368063889711e-05,
+      "loss": 0.2597,
+      "step": 6134
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4454203497062445e-05,
+      "loss": 0.2525,
+      "step": 6135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4450038972362019e-05,
+      "loss": 0.2516,
+      "step": 6136
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4445874490109862e-05,
+      "loss": 0.2658,
+      "step": 6137
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.444171005062742e-05,
+      "loss": 0.2327,
+      "step": 6138
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4437545654236116e-05,
+      "loss": 0.2782,
+      "step": 6139
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4433381301257369e-05,
+      "loss": 0.2459,
+      "step": 6140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4429216992012612e-05,
+      "loss": 0.2359,
+      "step": 6141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4425052726823253e-05,
+      "loss": 0.294,
+      "step": 6142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4420888506010717e-05,
+      "loss": 0.2478,
+      "step": 6143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4416724329896408e-05,
+      "loss": 0.2578,
+      "step": 6144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4412560198801733e-05,
+      "loss": 0.2538,
+      "step": 6145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.440839611304811e-05,
+      "loss": 0.2712,
+      "step": 6146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4404232072956929e-05,
+      "loss": 0.252,
+      "step": 6147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.440006807884959e-05,
+      "loss": 0.2916,
+      "step": 6148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4395904131047491e-05,
+      "loss": 0.2498,
+      "step": 6149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4391740229872015e-05,
+      "loss": 0.2581,
+      "step": 6150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4387576375644559e-05,
+      "loss": 0.2896,
+      "step": 6151
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.43834125686865e-05,
+      "loss": 0.2597,
+      "step": 6152
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4379248809319216e-05,
+      "loss": 0.2547,
+      "step": 6153
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.437508509786409e-05,
+      "loss": 0.2524,
+      "step": 6154
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4370921434642494e-05,
+      "loss": 0.2403,
+      "step": 6155
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4366757819975793e-05,
+      "loss": 0.2721,
+      "step": 6156
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4362594254185352e-05,
+      "loss": 0.2578,
+      "step": 6157
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4358430737592536e-05,
+      "loss": 0.2171,
+      "step": 6158
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4354267270518697e-05,
+      "loss": 0.255,
+      "step": 6159
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4350103853285194e-05,
+      "loss": 0.2167,
+      "step": 6160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4345940486213368e-05,
+      "loss": 0.2165,
+      "step": 6161
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.434177716962458e-05,
+      "loss": 0.263,
+      "step": 6162
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4337613903840161e-05,
+      "loss": 0.2391,
+      "step": 6163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4333450689181452e-05,
+      "loss": 0.282,
+      "step": 6164
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4329287525969787e-05,
+      "loss": 0.2265,
+      "step": 6165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.43251244145265e-05,
+      "loss": 0.2999,
+      "step": 6166
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4320961355172908e-05,
+      "loss": 0.261,
+      "step": 6167
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.431679834823034e-05,
+      "loss": 0.2587,
+      "step": 6168
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4312635394020115e-05,
+      "loss": 0.259,
+      "step": 6169
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4308472492863537e-05,
+      "loss": 0.2553,
+      "step": 6170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4304309645081932e-05,
+      "loss": 0.278,
+      "step": 6171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4300146850996594e-05,
+      "loss": 0.2846,
+      "step": 6172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4295984110928832e-05,
+      "loss": 0.2337,
+      "step": 6173
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4291821425199942e-05,
+      "loss": 0.2654,
+      "step": 6174
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4287658794131209e-05,
+      "loss": 0.2406,
+      "step": 6175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4283496218043933e-05,
+      "loss": 0.2328,
+      "step": 6176
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4279333697259395e-05,
+      "loss": 0.2465,
+      "step": 6177
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4275171232098868e-05,
+      "loss": 0.252,
+      "step": 6178
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4271008822883643e-05,
+      "loss": 0.2526,
+      "step": 6179
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4266846469934983e-05,
+      "loss": 0.2407,
+      "step": 6180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4262684173574157e-05,
+      "loss": 0.2907,
+      "step": 6181
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4258521934122429e-05,
+      "loss": 0.241,
+      "step": 6182
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4254359751901052e-05,
+      "loss": 0.2647,
+      "step": 6183
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4250197627231295e-05,
+      "loss": 0.2185,
+      "step": 6184
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4246035560434397e-05,
+      "loss": 0.2404,
+      "step": 6185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.42418735518316e-05,
+      "loss": 0.2435,
+      "step": 6186
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4237711601744157e-05,
+      "loss": 0.2591,
+      "step": 6187
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4233549710493297e-05,
+      "loss": 0.2412,
+      "step": 6188
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.422938787840025e-05,
+      "loss": 0.2661,
+      "step": 6189
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4225226105786251e-05,
+      "loss": 0.242,
+      "step": 6190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4221064392972511e-05,
+      "loss": 0.2595,
+      "step": 6191
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4216902740280268e-05,
+      "loss": 0.2435,
+      "step": 6192
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.421274114803072e-05,
+      "loss": 0.2813,
+      "step": 6193
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4208579616545073e-05,
+      "loss": 0.2498,
+      "step": 6194
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4204418146144544e-05,
+      "loss": 0.2154,
+      "step": 6195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4200256737150327e-05,
+      "loss": 0.2654,
+      "step": 6196
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4196095389883613e-05,
+      "loss": 0.2354,
+      "step": 6197
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4191934104665598e-05,
+      "loss": 0.2656,
+      "step": 6198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4187772881817467e-05,
+      "loss": 0.2444,
+      "step": 6199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.418361172166039e-05,
+      "loss": 0.2504,
+      "step": 6200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4179450624515561e-05,
+      "loss": 0.2381,
+      "step": 6201
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.417528959070413e-05,
+      "loss": 0.2174,
+      "step": 6202
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4171128620547281e-05,
+      "loss": 0.2574,
+      "step": 6203
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4166967714366168e-05,
+      "loss": 0.2609,
+      "step": 6204
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4162806872481941e-05,
+      "loss": 0.2145,
+      "step": 6205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.415864609521576e-05,
+      "loss": 0.2223,
+      "step": 6206
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4154485382888769e-05,
+      "loss": 0.2523,
+      "step": 6207
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4150324735822102e-05,
+      "loss": 0.2442,
+      "step": 6208
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4146164154336908e-05,
+      "loss": 0.2807,
+      "step": 6209
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.414200363875431e-05,
+      "loss": 0.278,
+      "step": 6210
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4137843189395425e-05,
+      "loss": 0.2462,
+      "step": 6211
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4133682806581392e-05,
+      "loss": 0.24,
+      "step": 6212
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4129522490633312e-05,
+      "loss": 0.2379,
+      "step": 6213
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4125362241872306e-05,
+      "loss": 0.2717,
+      "step": 6214
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4121202060619475e-05,
+      "loss": 0.259,
+      "step": 6215
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4117041947195912e-05,
+      "loss": 0.2529,
+      "step": 6216
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4112881901922728e-05,
+      "loss": 0.2493,
+      "step": 6217
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4108721925121001e-05,
+      "loss": 0.2523,
+      "step": 6218
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4104562017111809e-05,
+      "loss": 0.2242,
+      "step": 6219
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4100402178216246e-05,
+      "loss": 0.2839,
+      "step": 6220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4096242408755377e-05,
+      "loss": 0.2659,
+      "step": 6221
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4092082709050275e-05,
+      "loss": 0.2649,
+      "step": 6222
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4087923079422001e-05,
+      "loss": 0.2766,
+      "step": 6223
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4083763520191603e-05,
+      "loss": 0.2621,
+      "step": 6224
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.407960403168015e-05,
+      "loss": 0.2472,
+      "step": 6225
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4075444614208682e-05,
+      "loss": 0.2308,
+      "step": 6226
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.407128526809823e-05,
+      "loss": 0.2791,
+      "step": 6227
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4067125993669844e-05,
+      "loss": 0.2602,
+      "step": 6228
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4062966791244548e-05,
+      "loss": 0.2888,
+      "step": 6229
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4058807661143362e-05,
+      "loss": 0.2743,
+      "step": 6230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4054648603687316e-05,
+      "loss": 0.2998,
+      "step": 6231
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4050489619197407e-05,
+      "loss": 0.2407,
+      "step": 6232
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.404633070799466e-05,
+      "loss": 0.237,
+      "step": 6233
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4042171870400074e-05,
+      "loss": 0.2547,
+      "step": 6234
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.403801310673463e-05,
+      "loss": 0.2562,
+      "step": 6235
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4033854417319337e-05,
+      "loss": 0.2519,
+      "step": 6236
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4029695802475172e-05,
+      "loss": 0.2454,
+      "step": 6237
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4025537262523113e-05,
+      "loss": 0.213,
+      "step": 6238
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4021378797784137e-05,
+      "loss": 0.2824,
+      "step": 6239
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.401722040857921e-05,
+      "loss": 0.2507,
+      "step": 6240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4013062095229286e-05,
+      "loss": 0.2226,
+      "step": 6241
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4008903858055335e-05,
+      "loss": 0.2302,
+      "step": 6242
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4004745697378298e-05,
+      "loss": 0.2771,
+      "step": 6243
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4000587613519124e-05,
+      "loss": 0.2329,
+      "step": 6244
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3996429606798748e-05,
+      "loss": 0.2589,
+      "step": 6245
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3992271677538101e-05,
+      "loss": 0.3003,
+      "step": 6246
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3988113826058112e-05,
+      "loss": 0.2701,
+      "step": 6247
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3983956052679703e-05,
+      "loss": 0.2746,
+      "step": 6248
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3979798357723778e-05,
+      "loss": 0.2829,
+      "step": 6249
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3975640741511258e-05,
+      "loss": 0.2514,
+      "step": 6250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.397148320436304e-05,
+      "loss": 0.2513,
+      "step": 6251
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.396732574660002e-05,
+      "loss": 0.308,
+      "step": 6252
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3963168368543088e-05,
+      "loss": 0.2148,
+      "step": 6253
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3959011070513124e-05,
+      "loss": 0.2636,
+      "step": 6254
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3954853852831012e-05,
+      "loss": 0.2377,
+      "step": 6255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.395069671581762e-05,
+      "loss": 0.2794,
+      "step": 6256
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3946539659793808e-05,
+      "loss": 0.2516,
+      "step": 6257
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3942382685080447e-05,
+      "loss": 0.322,
+      "step": 6258
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3938225791998386e-05,
+      "loss": 0.2439,
+      "step": 6259
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3934068980868458e-05,
+      "loss": 0.2377,
+      "step": 6260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3929912252011516e-05,
+      "loss": 0.2672,
+      "step": 6261
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.392575560574839e-05,
+      "loss": 0.254,
+      "step": 6262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3921599042399913e-05,
+      "loss": 0.2705,
+      "step": 6263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3917442562286898e-05,
+      "loss": 0.26,
+      "step": 6264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3913286165730156e-05,
+      "loss": 0.2654,
+      "step": 6265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3909129853050506e-05,
+      "loss": 0.2726,
+      "step": 6266
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3904973624568745e-05,
+      "loss": 0.2625,
+      "step": 6267
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3900817480605667e-05,
+      "loss": 0.2456,
+      "step": 6268
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3896661421482062e-05,
+      "loss": 0.2751,
+      "step": 6269
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.389250544751871e-05,
+      "loss": 0.2451,
+      "step": 6270
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3888349559036385e-05,
+      "loss": 0.3134,
+      "step": 6271
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.388419375635586e-05,
+      "loss": 0.26,
+      "step": 6272
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3880038039797889e-05,
+      "loss": 0.2366,
+      "step": 6273
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.387588240968324e-05,
+      "loss": 0.2161,
+      "step": 6274
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3871726866332655e-05,
+      "loss": 0.2349,
+      "step": 6275
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3867571410066871e-05,
+      "loss": 0.3004,
+      "step": 6276
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3863416041206634e-05,
+      "loss": 0.2722,
+      "step": 6277
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3859260760072667e-05,
+      "loss": 0.2912,
+      "step": 6278
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3855105566985689e-05,
+      "loss": 0.246,
+      "step": 6279
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.385095046226642e-05,
+      "loss": 0.2513,
+      "step": 6280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3846795446235561e-05,
+      "loss": 0.1915,
+      "step": 6281
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3842640519213825e-05,
+      "loss": 0.2341,
+      "step": 6282
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3838485681521902e-05,
+      "loss": 0.2369,
+      "step": 6283
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3834330933480474e-05,
+      "loss": 0.2639,
+      "step": 6284
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3830176275410226e-05,
+      "loss": 0.2386,
+      "step": 6285
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3826021707631834e-05,
+      "loss": 0.2131,
+      "step": 6286
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3821867230465958e-05,
+      "loss": 0.2828,
+      "step": 6287
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3817712844233266e-05,
+      "loss": 0.2521,
+      "step": 6288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3813558549254408e-05,
+      "loss": 0.2408,
+      "step": 6289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.380940434585002e-05,
+      "loss": 0.2543,
+      "step": 6290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3805250234340755e-05,
+      "loss": 0.2411,
+      "step": 6291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3801096215047237e-05,
+      "loss": 0.2753,
+      "step": 6292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3796942288290094e-05,
+      "loss": 0.2249,
+      "step": 6293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3792788454389944e-05,
+      "loss": 0.2409,
+      "step": 6294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.378863471366739e-05,
+      "loss": 0.2453,
+      "step": 6295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3784481066443044e-05,
+      "loss": 0.2567,
+      "step": 6296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3780327513037497e-05,
+      "loss": 0.2529,
+      "step": 6297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.377617405377133e-05,
+      "loss": 0.2588,
+      "step": 6298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.377202068896514e-05,
+      "loss": 0.2961,
+      "step": 6299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3767867418939493e-05,
+      "loss": 0.2568,
+      "step": 6300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3763714244014952e-05,
+      "loss": 0.2468,
+      "step": 6301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3759561164512082e-05,
+      "loss": 0.2207,
+      "step": 6302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3755408180751431e-05,
+      "loss": 0.2344,
+      "step": 6303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.375125529305355e-05,
+      "loss": 0.2883,
+      "step": 6304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.374710250173897e-05,
+      "loss": 0.275,
+      "step": 6305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3742949807128215e-05,
+      "loss": 0.2645,
+      "step": 6306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3738797209541824e-05,
+      "loss": 0.2765,
+      "step": 6307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3734644709300299e-05,
+      "loss": 0.2345,
+      "step": 6308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3730492306724149e-05,
+      "loss": 0.236,
+      "step": 6309
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3726340002133878e-05,
+      "loss": 0.2383,
+      "step": 6310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3722187795849968e-05,
+      "loss": 0.3086,
+      "step": 6311
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3718035688192922e-05,
+      "loss": 0.2529,
+      "step": 6312
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.37138836794832e-05,
+      "loss": 0.2341,
+      "step": 6313
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3709731770041271e-05,
+      "loss": 0.2861,
+      "step": 6314
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.370557996018761e-05,
+      "loss": 0.2233,
+      "step": 6315
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3701428250242664e-05,
+      "loss": 0.2548,
+      "step": 6316
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3697276640526875e-05,
+      "loss": 0.2647,
+      "step": 6317
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3693125131360688e-05,
+      "loss": 0.249,
+      "step": 6318
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.368897372306453e-05,
+      "loss": 0.239,
+      "step": 6319
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3684822415958824e-05,
+      "loss": 0.2533,
+      "step": 6320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3680671210363987e-05,
+      "loss": 0.2424,
+      "step": 6321
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.367652010660042e-05,
+      "loss": 0.2457,
+      "step": 6322
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3672369104988534e-05,
+      "loss": 0.3167,
+      "step": 6323
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3668218205848713e-05,
+      "loss": 0.2437,
+      "step": 6324
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.366406740950134e-05,
+      "loss": 0.2778,
+      "step": 6325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3659916716266795e-05,
+      "loss": 0.2328,
+      "step": 6326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3655766126465444e-05,
+      "loss": 0.3054,
+      "step": 6327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3651615640417641e-05,
+      "loss": 0.2762,
+      "step": 6328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3647465258443747e-05,
+      "loss": 0.2381,
+      "step": 6329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3643314980864102e-05,
+      "loss": 0.3006,
+      "step": 6330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3639164807999034e-05,
+      "loss": 0.2532,
+      "step": 6331
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3635014740168886e-05,
+      "loss": 0.269,
+      "step": 6332
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3630864777693965e-05,
+      "loss": 0.2428,
+      "step": 6333
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3626714920894587e-05,
+      "loss": 0.2577,
+      "step": 6334
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.362256517009106e-05,
+      "loss": 0.242,
+      "step": 6335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3618415525603664e-05,
+      "loss": 0.2482,
+      "step": 6336
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3614265987752705e-05,
+      "loss": 0.2994,
+      "step": 6337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3610116556858449e-05,
+      "loss": 0.245,
+      "step": 6338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3605967233241164e-05,
+      "loss": 0.2618,
+      "step": 6339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3601818017221123e-05,
+      "loss": 0.238,
+      "step": 6340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3597668909118573e-05,
+      "loss": 0.2376,
+      "step": 6341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.359351990925376e-05,
+      "loss": 0.2712,
+      "step": 6342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3589371017946924e-05,
+      "loss": 0.2731,
+      "step": 6343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3585222235518284e-05,
+      "loss": 0.2764,
+      "step": 6344
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3581073562288077e-05,
+      "loss": 0.25,
+      "step": 6345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3576924998576504e-05,
+      "loss": 0.2502,
+      "step": 6346
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3572776544703762e-05,
+      "loss": 0.2624,
+      "step": 6347
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3568628200990058e-05,
+      "loss": 0.2861,
+      "step": 6348
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3564479967755577e-05,
+      "loss": 0.2374,
+      "step": 6349
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.356033184532049e-05,
+      "loss": 0.2603,
+      "step": 6350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3556183834004973e-05,
+      "loss": 0.2694,
+      "step": 6351
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3552035934129177e-05,
+      "loss": 0.2696,
+      "step": 6352
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3547888146013267e-05,
+      "loss": 0.249,
+      "step": 6353
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3543740469977386e-05,
+      "loss": 0.2163,
+      "step": 6354
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3539592906341654e-05,
+      "loss": 0.2477,
+      "step": 6355
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3535445455426211e-05,
+      "loss": 0.2764,
+      "step": 6356
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3531298117551171e-05,
+      "loss": 0.3318,
+      "step": 6357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3527150893036638e-05,
+      "loss": 0.2634,
+      "step": 6358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.352300378220272e-05,
+      "loss": 0.2836,
+      "step": 6359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3518856785369504e-05,
+      "loss": 0.2359,
+      "step": 6360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3514709902857064e-05,
+      "loss": 0.2143,
+      "step": 6361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3510563134985494e-05,
+      "loss": 0.26,
+      "step": 6362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3506416482074838e-05,
+      "loss": 0.2357,
+      "step": 6363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3502269944445166e-05,
+      "loss": 0.247,
+      "step": 6364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.349812352241652e-05,
+      "loss": 0.2661,
+      "step": 6365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3493977216308933e-05,
+      "loss": 0.2355,
+      "step": 6366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3489831026442446e-05,
+      "loss": 0.2753,
+      "step": 6367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.348568495313707e-05,
+      "loss": 0.2853,
+      "step": 6368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3481538996712812e-05,
+      "loss": 0.2261,
+      "step": 6369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3477393157489691e-05,
+      "loss": 0.2732,
+      "step": 6370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3473247435787684e-05,
+      "loss": 0.2514,
+      "step": 6371
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3469101831926776e-05,
+      "loss": 0.3235,
+      "step": 6372
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3464956346226954e-05,
+      "loss": 0.2423,
+      "step": 6373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.346081097900817e-05,
+      "loss": 0.2688,
+      "step": 6374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3456665730590393e-05,
+      "loss": 0.2783,
+      "step": 6375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3452520601293563e-05,
+      "loss": 0.2531,
+      "step": 6376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3448375591437614e-05,
+      "loss": 0.2471,
+      "step": 6377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3444230701342489e-05,
+      "loss": 0.2458,
+      "step": 6378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3440085931328098e-05,
+      "loss": 0.2503,
+      "step": 6379
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3435941281714347e-05,
+      "loss": 0.2361,
+      "step": 6380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3431796752821152e-05,
+      "loss": 0.2465,
+      "step": 6381
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3427652344968388e-05,
+      "loss": 0.2358,
+      "step": 6382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3423508058475953e-05,
+      "loss": 0.2503,
+      "step": 6383
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3419363893663711e-05,
+      "loss": 0.2441,
+      "step": 6384
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3415219850851524e-05,
+      "loss": 0.2736,
+      "step": 6385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3411075930359255e-05,
+      "loss": 0.2581,
+      "step": 6386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3406932132506749e-05,
+      "loss": 0.2364,
+      "step": 6387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3402788457613828e-05,
+      "loss": 0.2548,
+      "step": 6388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3398644906000333e-05,
+      "loss": 0.2319,
+      "step": 6389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3394501477986074e-05,
+      "loss": 0.2423,
+      "step": 6390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3390358173890856e-05,
+      "loss": 0.2456,
+      "step": 6391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3386214994034483e-05,
+      "loss": 0.2468,
+      "step": 6392
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.338207193873673e-05,
+      "loss": 0.2745,
+      "step": 6393
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3377929008317393e-05,
+      "loss": 0.2811,
+      "step": 6394
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3373786203096231e-05,
+      "loss": 0.2744,
+      "step": 6395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3369643523392999e-05,
+      "loss": 0.2465,
+      "step": 6396
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3365500969527455e-05,
+      "loss": 0.2723,
+      "step": 6397
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3361358541819335e-05,
+      "loss": 0.281,
+      "step": 6398
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3357216240588365e-05,
+      "loss": 0.2501,
+      "step": 6399
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.335307406615427e-05,
+      "loss": 0.2329,
+      "step": 6400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.334893201883676e-05,
+      "loss": 0.2732,
+      "step": 6401
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3344790098955526e-05,
+      "loss": 0.2343,
+      "step": 6402
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3340648306830274e-05,
+      "loss": 0.2801,
+      "step": 6403
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3336506642780672e-05,
+      "loss": 0.3189,
+      "step": 6404
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.33323651071264e-05,
+      "loss": 0.2449,
+      "step": 6405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3328223700187116e-05,
+      "loss": 0.2323,
+      "step": 6406
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3324082422282464e-05,
+      "loss": 0.2369,
+      "step": 6407
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3319941273732095e-05,
+      "loss": 0.2758,
+      "step": 6408
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3315800254855638e-05,
+      "loss": 0.2446,
+      "step": 6409
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3311659365972704e-05,
+      "loss": 0.2502,
+      "step": 6410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3307518607402917e-05,
+      "loss": 0.2336,
+      "step": 6411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.330337797946587e-05,
+      "loss": 0.2953,
+      "step": 6412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3299237482481162e-05,
+      "loss": 0.2676,
+      "step": 6413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3295097116768367e-05,
+      "loss": 0.249,
+      "step": 6414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3290956882647054e-05,
+      "loss": 0.2163,
+      "step": 6415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.328681678043679e-05,
+      "loss": 0.2796,
+      "step": 6416
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3282676810457121e-05,
+      "loss": 0.2373,
+      "step": 6417
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3278536973027582e-05,
+      "loss": 0.2421,
+      "step": 6418
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3274397268467716e-05,
+      "loss": 0.2409,
+      "step": 6419
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3270257697097034e-05,
+      "loss": 0.2591,
+      "step": 6420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3266118259235044e-05,
+      "loss": 0.2507,
+      "step": 6421
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3261978955201251e-05,
+      "loss": 0.2563,
+      "step": 6422
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3257839785315137e-05,
+      "loss": 0.2224,
+      "step": 6423
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3253700749896186e-05,
+      "loss": 0.2716,
+      "step": 6424
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3249561849263865e-05,
+      "loss": 0.2493,
+      "step": 6425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3245423083737623e-05,
+      "loss": 0.1937,
+      "step": 6426
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.324128445363692e-05,
+      "loss": 0.2313,
+      "step": 6427
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.323714595928119e-05,
+      "loss": 0.2513,
+      "step": 6428
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.323300760098985e-05,
+      "loss": 0.2344,
+      "step": 6429
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3228869379082324e-05,
+      "loss": 0.2406,
+      "step": 6430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3224731293878017e-05,
+      "loss": 0.2224,
+      "step": 6431
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3220593345696317e-05,
+      "loss": 0.2638,
+      "step": 6432
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3216455534856618e-05,
+      "loss": 0.2523,
+      "step": 6433
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.321231786167828e-05,
+      "loss": 0.2468,
+      "step": 6434
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3208180326480681e-05,
+      "loss": 0.3003,
+      "step": 6435
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3204042929583163e-05,
+      "loss": 0.2745,
+      "step": 6436
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.319990567130507e-05,
+      "loss": 0.2487,
+      "step": 6437
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3195768551965736e-05,
+      "loss": 0.2624,
+      "step": 6438
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3191631571884478e-05,
+      "loss": 0.2466,
+      "step": 6439
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3187494731380602e-05,
+      "loss": 0.2256,
+      "step": 6440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3183358030773413e-05,
+      "loss": 0.2776,
+      "step": 6441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3179221470382192e-05,
+      "loss": 0.2563,
+      "step": 6442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3175085050526223e-05,
+      "loss": 0.2585,
+      "step": 6443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.317094877152477e-05,
+      "loss": 0.2651,
+      "step": 6444
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3166812633697083e-05,
+      "loss": 0.255,
+      "step": 6445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3162676637362414e-05,
+      "loss": 0.2318,
+      "step": 6446
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3158540782839993e-05,
+      "loss": 0.2368,
+      "step": 6447
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3154405070449039e-05,
+      "loss": 0.2379,
+      "step": 6448
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3150269500508769e-05,
+      "loss": 0.2797,
+      "step": 6449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3146134073338382e-05,
+      "loss": 0.2447,
+      "step": 6450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3141998789257057e-05,
+      "loss": 0.2357,
+      "step": 6451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.313786364858399e-05,
+      "loss": 0.231,
+      "step": 6452
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.313372865163834e-05,
+      "loss": 0.2666,
+      "step": 6453
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3129593798739262e-05,
+      "loss": 0.2543,
+      "step": 6454
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3125459090205905e-05,
+      "loss": 0.2714,
+      "step": 6455
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3121324526357397e-05,
+      "loss": 0.2373,
+      "step": 6456
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3117190107512867e-05,
+      "loss": 0.2358,
+      "step": 6457
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3113055833991423e-05,
+      "loss": 0.2358,
+      "step": 6458
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3108921706112162e-05,
+      "loss": 0.2838,
+      "step": 6459
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3104787724194183e-05,
+      "loss": 0.2338,
+      "step": 6460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3100653888556557e-05,
+      "loss": 0.252,
+      "step": 6461
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3096520199518352e-05,
+      "loss": 0.2434,
+      "step": 6462
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3092386657398623e-05,
+      "loss": 0.2573,
+      "step": 6463
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.308825326251641e-05,
+      "loss": 0.2585,
+      "step": 6464
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3084120015190759e-05,
+      "loss": 0.2261,
+      "step": 6465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3079986915740677e-05,
+      "loss": 0.246,
+      "step": 6466
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3075853964485174e-05,
+      "loss": 0.2402,
+      "step": 6467
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.307172116174326e-05,
+      "loss": 0.2404,
+      "step": 6468
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3067588507833913e-05,
+      "loss": 0.2337,
+      "step": 6469
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.306345600307611e-05,
+      "loss": 0.2267,
+      "step": 6470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3059323647788816e-05,
+      "loss": 0.2325,
+      "step": 6471
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3055191442290978e-05,
+      "loss": 0.2296,
+      "step": 6472
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3051059386901549e-05,
+      "loss": 0.2558,
+      "step": 6473
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3046927481939449e-05,
+      "loss": 0.2261,
+      "step": 6474
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3042795727723589e-05,
+      "loss": 0.2364,
+      "step": 6475
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3038664124572893e-05,
+      "loss": 0.2587,
+      "step": 6476
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.303453267280624e-05,
+      "loss": 0.2662,
+      "step": 6477
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.303040137274252e-05,
+      "loss": 0.2958,
+      "step": 6478
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3026270224700603e-05,
+      "loss": 0.2471,
+      "step": 6479
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3022139228999347e-05,
+      "loss": 0.2413,
+      "step": 6480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3018008385957593e-05,
+      "loss": 0.293,
+      "step": 6481
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3013877695894192e-05,
+      "loss": 0.2814,
+      "step": 6482
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.300974715912795e-05,
+      "loss": 0.2579,
+      "step": 6483
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3005616775977695e-05,
+      "loss": 0.2753,
+      "step": 6484
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3001486546762218e-05,
+      "loss": 0.2608,
+      "step": 6485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2997356471800308e-05,
+      "loss": 0.2348,
+      "step": 6486
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2993226551410745e-05,
+      "loss": 0.2508,
+      "step": 6487
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.298909678591229e-05,
+      "loss": 0.282,
+      "step": 6488
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2984967175623693e-05,
+      "loss": 0.2247,
+      "step": 6489
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2980837720863704e-05,
+      "loss": 0.255,
+      "step": 6490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2976708421951044e-05,
+      "loss": 0.2634,
+      "step": 6491
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2972579279204425e-05,
+      "loss": 0.2579,
+      "step": 6492
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2968450292942563e-05,
+      "loss": 0.2628,
+      "step": 6493
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.296432146348414e-05,
+      "loss": 0.2405,
+      "step": 6494
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2960192791147846e-05,
+      "loss": 0.2769,
+      "step": 6495
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2956064276252343e-05,
+      "loss": 0.2514,
+      "step": 6496
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.295193591911628e-05,
+      "loss": 0.2523,
+      "step": 6497
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2947807720058322e-05,
+      "loss": 0.2613,
+      "step": 6498
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2943679679397079e-05,
+      "loss": 0.2451,
+      "step": 6499
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2939551797451174e-05,
+      "loss": 0.2942,
+      "step": 6500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2935424074539227e-05,
+      "loss": 0.2328,
+      "step": 6501
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2931296510979817e-05,
+      "loss": 0.2983,
+      "step": 6502
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2927169107091538e-05,
+      "loss": 0.2545,
+      "step": 6503
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2923041863192959e-05,
+      "loss": 0.2762,
+      "step": 6504
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2918914779602624e-05,
+      "loss": 0.2437,
+      "step": 6505
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2914787856639099e-05,
+      "loss": 0.2511,
+      "step": 6506
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2910661094620909e-05,
+      "loss": 0.2461,
+      "step": 6507
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2906534493866562e-05,
+      "loss": 0.2403,
+      "step": 6508
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2902408054694585e-05,
+      "loss": 0.2183,
+      "step": 6509
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2898281777423466e-05,
+      "loss": 0.2323,
+      "step": 6510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2894155662371684e-05,
+      "loss": 0.2463,
+      "step": 6511
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2890029709857716e-05,
+      "loss": 0.2446,
+      "step": 6512
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2885903920200014e-05,
+      "loss": 0.2325,
+      "step": 6513
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2881778293717032e-05,
+      "loss": 0.2164,
+      "step": 6514
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2877652830727201e-05,
+      "loss": 0.2433,
+      "step": 6515
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.287352753154893e-05,
+      "loss": 0.2348,
+      "step": 6516
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2869402396500642e-05,
+      "loss": 0.2349,
+      "step": 6517
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2865277425900725e-05,
+      "loss": 0.2695,
+      "step": 6518
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.286115262006756e-05,
+      "loss": 0.2525,
+      "step": 6519
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2857027979319522e-05,
+      "loss": 0.2542,
+      "step": 6520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2852903503974961e-05,
+      "loss": 0.2419,
+      "step": 6521
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2848779194352222e-05,
+      "loss": 0.2327,
+      "step": 6522
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2844655050769641e-05,
+      "loss": 0.2555,
+      "step": 6523
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2840531073545535e-05,
+      "loss": 0.2417,
+      "step": 6524
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2836407262998211e-05,
+      "loss": 0.2863,
+      "step": 6525
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.283228361944596e-05,
+      "loss": 0.244,
+      "step": 6526
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2828160143207058e-05,
+      "loss": 0.2517,
+      "step": 6527
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.282403683459978e-05,
+      "loss": 0.2819,
+      "step": 6528
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2819913693942377e-05,
+      "loss": 0.2892,
+      "step": 6529
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2815790721553083e-05,
+      "loss": 0.2249,
+      "step": 6530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2811667917750137e-05,
+      "loss": 0.2201,
+      "step": 6531
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2807545282851752e-05,
+      "loss": 0.2658,
+      "step": 6532
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2803422817176119e-05,
+      "loss": 0.2374,
+      "step": 6533
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2799300521041441e-05,
+      "loss": 0.2896,
+      "step": 6534
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2795178394765886e-05,
+      "loss": 0.2346,
+      "step": 6535
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2791056438667621e-05,
+      "loss": 0.2725,
+      "step": 6536
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2786934653064797e-05,
+      "loss": 0.2497,
+      "step": 6537
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2782813038275538e-05,
+      "loss": 0.2699,
+      "step": 6538
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2778691594617985e-05,
+      "loss": 0.2428,
+      "step": 6539
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2774570322410243e-05,
+      "loss": 0.2229,
+      "step": 6540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2770449221970394e-05,
+      "loss": 0.2436,
+      "step": 6541
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2766328293616541e-05,
+      "loss": 0.2369,
+      "step": 6542
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2762207537666745e-05,
+      "loss": 0.2107,
+      "step": 6543
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2758086954439067e-05,
+      "loss": 0.246,
+      "step": 6544
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.275396654425155e-05,
+      "loss": 0.2776,
+      "step": 6545
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2749846307422214e-05,
+      "loss": 0.2325,
+      "step": 6546
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2745726244269094e-05,
+      "loss": 0.2596,
+      "step": 6547
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2741606355110182e-05,
+      "loss": 0.2372,
+      "step": 6548
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.273748664026347e-05,
+      "loss": 0.2073,
+      "step": 6549
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2733367100046938e-05,
+      "loss": 0.2334,
+      "step": 6550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2729247734778548e-05,
+      "loss": 0.2472,
+      "step": 6551
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2725128544776241e-05,
+      "loss": 0.2379,
+      "step": 6552
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2721009530357967e-05,
+      "loss": 0.2348,
+      "step": 6553
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2716890691841635e-05,
+      "loss": 0.246,
+      "step": 6554
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.271277202954517e-05,
+      "loss": 0.2581,
+      "step": 6555
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2708653543786455e-05,
+      "loss": 0.2447,
+      "step": 6556
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2704535234883374e-05,
+      "loss": 0.2802,
+      "step": 6557
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.27004171031538e-05,
+      "loss": 0.2564,
+      "step": 6558
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2696299148915583e-05,
+      "loss": 0.2121,
+      "step": 6559
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2692181372486564e-05,
+      "loss": 0.2901,
+      "step": 6560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2688063774184572e-05,
+      "loss": 0.24,
+      "step": 6561
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.268394635432742e-05,
+      "loss": 0.2724,
+      "step": 6562
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2679829113232902e-05,
+      "loss": 0.261,
+      "step": 6563
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2675712051218816e-05,
+      "loss": 0.2419,
+      "step": 6564
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2671595168602922e-05,
+      "loss": 0.2429,
+      "step": 6565
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2667478465702987e-05,
+      "loss": 0.2419,
+      "step": 6566
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2663361942836749e-05,
+      "loss": 0.2444,
+      "step": 6567
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.265924560032194e-05,
+      "loss": 0.2408,
+      "step": 6568
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2655129438476278e-05,
+      "loss": 0.247,
+      "step": 6569
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2651013457617465e-05,
+      "loss": 0.2658,
+      "step": 6570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2646897658063182e-05,
+      "loss": 0.255,
+      "step": 6571
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2642782040131116e-05,
+      "loss": 0.2527,
+      "step": 6572
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2638666604138921e-05,
+      "loss": 0.2335,
+      "step": 6573
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2634551350404246e-05,
+      "loss": 0.2349,
+      "step": 6574
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2630436279244723e-05,
+      "loss": 0.2477,
+      "step": 6575
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2626321390977963e-05,
+      "loss": 0.2285,
+      "step": 6576
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2622206685921582e-05,
+      "loss": 0.2798,
+      "step": 6577
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2618092164393163e-05,
+      "loss": 0.2541,
+      "step": 6578
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2613977826710276e-05,
+      "loss": 0.2496,
+      "step": 6579
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2609863673190495e-05,
+      "loss": 0.2745,
+      "step": 6580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2605749704151363e-05,
+      "loss": 0.2408,
+      "step": 6581
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2601635919910409e-05,
+      "loss": 0.2455,
+      "step": 6582
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.259752232078516e-05,
+      "loss": 0.2631,
+      "step": 6583
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2593408907093112e-05,
+      "loss": 0.2493,
+      "step": 6584
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2589295679151762e-05,
+      "loss": 0.2346,
+      "step": 6585
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2585182637278582e-05,
+      "loss": 0.2608,
+      "step": 6586
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2581069781791031e-05,
+      "loss": 0.2464,
+      "step": 6587
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2576957113006564e-05,
+      "loss": 0.2579,
+      "step": 6588
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2572844631242611e-05,
+      "loss": 0.2752,
+      "step": 6589
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2568732336816588e-05,
+      "loss": 0.2446,
+      "step": 6590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2564620230045901e-05,
+      "loss": 0.2824,
+      "step": 6591
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2560508311247941e-05,
+      "loss": 0.2409,
+      "step": 6592
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2556396580740078e-05,
+      "loss": 0.2553,
+      "step": 6593
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2552285038839677e-05,
+      "loss": 0.2516,
+      "step": 6594
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2548173685864076e-05,
+      "loss": 0.2461,
+      "step": 6595
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.254406252213062e-05,
+      "loss": 0.2664,
+      "step": 6596
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2539951547956616e-05,
+      "loss": 0.25,
+      "step": 6597
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2535840763659366e-05,
+      "loss": 0.2466,
+      "step": 6598
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2531730169556161e-05,
+      "loss": 0.2754,
+      "step": 6599
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2527619765964275e-05,
+      "loss": 0.2748,
+      "step": 6600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2523509553200961e-05,
+      "loss": 0.2627,
+      "step": 6601
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2519399531583466e-05,
+      "loss": 0.2332,
+      "step": 6602
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.251528970142901e-05,
+      "loss": 0.2387,
+      "step": 6603
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2511180063054823e-05,
+      "loss": 0.2237,
+      "step": 6604
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2507070616778093e-05,
+      "loss": 0.2164,
+      "step": 6605
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2502961362916003e-05,
+      "loss": 0.2614,
+      "step": 6606
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.249885230178573e-05,
+      "loss": 0.2645,
+      "step": 6607
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2494743433704422e-05,
+      "loss": 0.2774,
+      "step": 6608
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2490634758989217e-05,
+      "loss": 0.2413,
+      "step": 6609
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2486526277957247e-05,
+      "loss": 0.25,
+      "step": 6610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2482417990925618e-05,
+      "loss": 0.2427,
+      "step": 6611
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2478309898211417e-05,
+      "loss": 0.2232,
+      "step": 6612
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2474202000131738e-05,
+      "loss": 0.2558,
+      "step": 6613
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2470094297003632e-05,
+      "loss": 0.2535,
+      "step": 6614
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2465986789144162e-05,
+      "loss": 0.2687,
+      "step": 6615
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2461879476870353e-05,
+      "loss": 0.2569,
+      "step": 6616
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2457772360499221e-05,
+      "loss": 0.2678,
+      "step": 6617
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2453665440347787e-05,
+      "loss": 0.2769,
+      "step": 6618
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2449558716733026e-05,
+      "loss": 0.2629,
+      "step": 6619
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2445452189971909e-05,
+      "loss": 0.2621,
+      "step": 6620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2441345860381406e-05,
+      "loss": 0.263,
+      "step": 6621
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2437239728278456e-05,
+      "loss": 0.2239,
+      "step": 6622
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2433133793979985e-05,
+      "loss": 0.2606,
+      "step": 6623
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2429028057802911e-05,
+      "loss": 0.2621,
+      "step": 6624
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2424922520064122e-05,
+      "loss": 0.2482,
+      "step": 6625
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2420817181080519e-05,
+      "loss": 0.2347,
+      "step": 6626
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2416712041168951e-05,
+      "loss": 0.246,
+      "step": 6627
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.241260710064627e-05,
+      "loss": 0.2268,
+      "step": 6628
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2408502359829329e-05,
+      "loss": 0.2705,
+      "step": 6629
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2404397819034933e-05,
+      "loss": 0.2279,
+      "step": 6630
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2400293478579893e-05,
+      "loss": 0.2617,
+      "step": 6631
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2396189338781002e-05,
+      "loss": 0.3196,
+      "step": 6632
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2392085399955026e-05,
+      "loss": 0.2396,
+      "step": 6633
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2387981662418734e-05,
+      "loss": 0.2442,
+      "step": 6634
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2383878126488868e-05,
+      "loss": 0.2127,
+      "step": 6635
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2379774792482145e-05,
+      "loss": 0.2348,
+      "step": 6636
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2375671660715292e-05,
+      "loss": 0.2608,
+      "step": 6637
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2371568731505e-05,
+      "loss": 0.2587,
+      "step": 6638
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2367466005167945e-05,
+      "loss": 0.2334,
+      "step": 6639
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2363363482020798e-05,
+      "loss": 0.2651,
+      "step": 6640
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.235926116238021e-05,
+      "loss": 0.2618,
+      "step": 6641
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2355159046562805e-05,
+      "loss": 0.2201,
+      "step": 6642
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2351057134885218e-05,
+      "loss": 0.2738,
+      "step": 6643
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2346955427664036e-05,
+      "loss": 0.2685,
+      "step": 6644
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2342853925215857e-05,
+      "loss": 0.2537,
+      "step": 6645
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2338752627857248e-05,
+      "loss": 0.2458,
+      "step": 6646
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2334651535904759e-05,
+      "loss": 0.2809,
+      "step": 6647
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2330550649674935e-05,
+      "loss": 0.2432,
+      "step": 6648
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2326449969484301e-05,
+      "loss": 0.2131,
+      "step": 6649
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2322349495649355e-05,
+      "loss": 0.3041,
+      "step": 6650
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2318249228486601e-05,
+      "loss": 0.2791,
+      "step": 6651
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2314149168312513e-05,
+      "loss": 0.2332,
+      "step": 6652
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2310049315443538e-05,
+      "loss": 0.2781,
+      "step": 6653
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2305949670196133e-05,
+      "loss": 0.2542,
+      "step": 6654
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2301850232886718e-05,
+      "loss": 0.2679,
+      "step": 6655
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2297751003831711e-05,
+      "loss": 0.248,
+      "step": 6656
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2293651983347506e-05,
+      "loss": 0.2614,
+      "step": 6657
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2289553171750472e-05,
+      "loss": 0.2431,
+      "step": 6658
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2285454569356987e-05,
+      "loss": 0.2682,
+      "step": 6659
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2281356176483396e-05,
+      "loss": 0.2352,
+      "step": 6660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2277257993446015e-05,
+      "loss": 0.2601,
+      "step": 6661
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2273160020561178e-05,
+      "loss": 0.2371,
+      "step": 6662
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2269062258145171e-05,
+      "loss": 0.273,
+      "step": 6663
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2264964706514284e-05,
+      "loss": 0.2147,
+      "step": 6664
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.226086736598478e-05,
+      "loss": 0.2752,
+      "step": 6665
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2256770236872901e-05,
+      "loss": 0.237,
+      "step": 6666
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2252673319494898e-05,
+      "loss": 0.2347,
+      "step": 6667
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2248576614166977e-05,
+      "loss": 0.257,
+      "step": 6668
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2244480121205332e-05,
+      "loss": 0.2735,
+      "step": 6669
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.224038384092616e-05,
+      "loss": 0.2712,
+      "step": 6670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2236287773645626e-05,
+      "loss": 0.2436,
+      "step": 6671
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2232191919679875e-05,
+      "loss": 0.2624,
+      "step": 6672
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.222809627934505e-05,
+      "loss": 0.2465,
+      "step": 6673
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.222400085295726e-05,
+      "loss": 0.2531,
+      "step": 6674
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2219905640832619e-05,
+      "loss": 0.2035,
+      "step": 6675
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2215810643287207e-05,
+      "loss": 0.2345,
+      "step": 6676
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2211715860637087e-05,
+      "loss": 0.2401,
+      "step": 6677
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2207621293198324e-05,
+      "loss": 0.2475,
+      "step": 6678
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2203526941286944e-05,
+      "loss": 0.2673,
+      "step": 6679
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2199432805218964e-05,
+      "loss": 0.2683,
+      "step": 6680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2195338885310397e-05,
+      "loss": 0.2333,
+      "step": 6681
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2191245181877221e-05,
+      "loss": 0.2865,
+      "step": 6682
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.21871516952354e-05,
+      "loss": 0.2513,
+      "step": 6683
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2183058425700901e-05,
+      "loss": 0.2454,
+      "step": 6684
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2178965373589647e-05,
+      "loss": 0.2344,
+      "step": 6685
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2174872539217567e-05,
+      "loss": 0.2123,
+      "step": 6686
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2170779922900555e-05,
+      "loss": 0.2386,
+      "step": 6687
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2166687524954496e-05,
+      "loss": 0.2242,
+      "step": 6688
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2162595345695263e-05,
+      "loss": 0.2091,
+      "step": 6689
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2158503385438706e-05,
+      "loss": 0.2375,
+      "step": 6690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2154411644500653e-05,
+      "loss": 0.2925,
+      "step": 6691
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2150320123196932e-05,
+      "loss": 0.2316,
+      "step": 6692
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2146228821843338e-05,
+      "loss": 0.2284,
+      "step": 6693
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2142137740755655e-05,
+      "loss": 0.2661,
+      "step": 6694
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2138046880249651e-05,
+      "loss": 0.215,
+      "step": 6695
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2133956240641073e-05,
+      "loss": 0.2701,
+      "step": 6696
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2129865822245656e-05,
+      "loss": 0.2245,
+      "step": 6697
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2125775625379118e-05,
+      "loss": 0.2595,
+      "step": 6698
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2121685650357145e-05,
+      "loss": 0.2503,
+      "step": 6699
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2117595897495433e-05,
+      "loss": 0.2615,
+      "step": 6700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2113506367109643e-05,
+      "loss": 0.2849,
+      "step": 6701
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2109417059515416e-05,
+      "loss": 0.2758,
+      "step": 6702
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2105327975028387e-05,
+      "loss": 0.2663,
+      "step": 6703
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2101239113964163e-05,
+      "loss": 0.2929,
+      "step": 6704
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2097150476638346e-05,
+      "loss": 0.2373,
+      "step": 6705
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2093062063366513e-05,
+      "loss": 0.2798,
+      "step": 6706
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2088973874464214e-05,
+      "loss": 0.2345,
+      "step": 6707
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.208488591024701e-05,
+      "loss": 0.2148,
+      "step": 6708
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2080798171030415e-05,
+      "loss": 0.2455,
+      "step": 6709
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2076710657129939e-05,
+      "loss": 0.2579,
+      "step": 6710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2072623368861078e-05,
+      "loss": 0.237,
+      "step": 6711
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2068536306539304e-05,
+      "loss": 0.2357,
+      "step": 6712
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2064449470480068e-05,
+      "loss": 0.2319,
+      "step": 6713
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2060362860998816e-05,
+      "loss": 0.2938,
+      "step": 6714
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2056276478410962e-05,
+      "loss": 0.2407,
+      "step": 6715
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2052190323031919e-05,
+      "loss": 0.2656,
+      "step": 6716
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2048104395177072e-05,
+      "loss": 0.2696,
+      "step": 6717
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2044018695161782e-05,
+      "loss": 0.2436,
+      "step": 6718
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.203993322330141e-05,
+      "loss": 0.2621,
+      "step": 6719
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2035847979911284e-05,
+      "loss": 0.2358,
+      "step": 6720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2031762965306719e-05,
+      "loss": 0.2178,
+      "step": 6721
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2027678179803018e-05,
+      "loss": 0.2559,
+      "step": 6722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.202359362371546e-05,
+      "loss": 0.2618,
+      "step": 6723
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2019509297359302e-05,
+      "loss": 0.2685,
+      "step": 6724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.20154252010498e-05,
+      "loss": 0.2626,
+      "step": 6725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.201134133510217e-05,
+      "loss": 0.2943,
+      "step": 6726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2007257699831635e-05,
+      "loss": 0.2238,
+      "step": 6727
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2003174295553378e-05,
+      "loss": 0.2583,
+      "step": 6728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1999091122582571e-05,
+      "loss": 0.2547,
+      "step": 6729
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1995008181234376e-05,
+      "loss": 0.2306,
+      "step": 6730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1990925471823932e-05,
+      "loss": 0.2253,
+      "step": 6731
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.198684299466635e-05,
+      "loss": 0.2252,
+      "step": 6732
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.1982760750076747e-05,
+      "loss": 0.2164,
+      "step": 6733
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1978678738370195e-05,
+      "loss": 0.2582,
+      "step": 6734
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.197459695986177e-05,
+      "loss": 0.2293,
+      "step": 6735
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1970515414866515e-05,
+      "loss": 0.2397,
+      "step": 6736
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1966434103699458e-05,
+      "loss": 0.2555,
+      "step": 6737
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1962353026675621e-05,
+      "loss": 0.2457,
+      "step": 6738
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1958272184109993e-05,
+      "loss": 0.2717,
+      "step": 6739
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1954191576317543e-05,
+      "loss": 0.2496,
+      "step": 6740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1950111203613242e-05,
+      "loss": 0.2419,
+      "step": 6741
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1946031066312026e-05,
+      "loss": 0.2258,
+      "step": 6742
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.194195116472881e-05,
+      "loss": 0.2417,
+      "step": 6743
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1937871499178509e-05,
+      "loss": 0.2371,
+      "step": 6744
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1933792069975996e-05,
+      "loss": 0.2143,
+      "step": 6745
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1929712877436155e-05,
+      "loss": 0.1985,
+      "step": 6746
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.192563392187382e-05,
+      "loss": 0.301,
+      "step": 6747
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1921555203603825e-05,
+      "loss": 0.2919,
+      "step": 6748
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1917476722940986e-05,
+      "loss": 0.2438,
+      "step": 6749
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1913398480200098e-05,
+      "loss": 0.2383,
+      "step": 6750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.190932047569593e-05,
+      "loss": 0.2515,
+      "step": 6751
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1905242709743246e-05,
+      "loss": 0.2448,
+      "step": 6752
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1901165182656785e-05,
+      "loss": 0.239,
+      "step": 6753
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1897087894751258e-05,
+      "loss": 0.2448,
+      "step": 6754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1893010846341379e-05,
+      "loss": 0.2257,
+      "step": 6755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.188893403774182e-05,
+      "loss": 0.2932,
+      "step": 6756
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1884857469267257e-05,
+      "loss": 0.2894,
+      "step": 6757
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1880781141232335e-05,
+      "loss": 0.3125,
+      "step": 6758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1876705053951673e-05,
+      "loss": 0.2594,
+      "step": 6759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.187262920773989e-05,
+      "loss": 0.263,
+      "step": 6760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1868553602911574e-05,
+      "loss": 0.2229,
+      "step": 6761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.186447823978129e-05,
+      "loss": 0.2702,
+      "step": 6762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1860403118663604e-05,
+      "loss": 0.2991,
+      "step": 6763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1856328239873036e-05,
+      "loss": 0.2737,
+      "step": 6764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1852253603724119e-05,
+      "loss": 0.2564,
+      "step": 6765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1848179210531339e-05,
+      "loss": 0.288,
+      "step": 6766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1844105060609174e-05,
+      "loss": 0.2282,
+      "step": 6767
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1840031154272091e-05,
+      "loss": 0.25,
+      "step": 6768
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1835957491834526e-05,
+      "loss": 0.2805,
+      "step": 6769
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1831884073610897e-05,
+      "loss": 0.3134,
+      "step": 6770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1827810899915623e-05,
+      "loss": 0.2565,
+      "step": 6771
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1823737971063072e-05,
+      "loss": 0.2855,
+      "step": 6772
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.181966528736761e-05,
+      "loss": 0.2333,
+      "step": 6773
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1815592849143598e-05,
+      "loss": 0.2302,
+      "step": 6774
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.181152065670535e-05,
+      "loss": 0.2684,
+      "step": 6775
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1807448710367183e-05,
+      "loss": 0.2402,
+      "step": 6776
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1803377010443384e-05,
+      "loss": 0.2692,
+      "step": 6777
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1799305557248215e-05,
+      "loss": 0.2573,
+      "step": 6778
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1795234351095946e-05,
+      "loss": 0.2384,
+      "step": 6779
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1791163392300796e-05,
+      "loss": 0.2605,
+      "step": 6780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1787092681176977e-05,
+      "loss": 0.2717,
+      "step": 6781
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1783022218038694e-05,
+      "loss": 0.2516,
+      "step": 6782
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1778952003200115e-05,
+      "loss": 0.2531,
+      "step": 6783
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1774882036975396e-05,
+      "loss": 0.2103,
+      "step": 6784
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1770812319678678e-05,
+      "loss": 0.2262,
+      "step": 6785
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.176674285162407e-05,
+      "loss": 0.2336,
+      "step": 6786
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.176267363312568e-05,
+      "loss": 0.2941,
+      "step": 6787
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1758604664497592e-05,
+      "loss": 0.2337,
+      "step": 6788
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1754535946053845e-05,
+      "loss": 0.2533,
+      "step": 6789
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.1750467478108501e-05,
+      "loss": 0.2546,
+      "step": 6790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.174639926097557e-05,
+      "loss": 0.2649,
+      "step": 6791
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1742331294969054e-05,
+      "loss": 0.2332,
+      "step": 6792
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.173826358040294e-05,
+      "loss": 0.2755,
+      "step": 6793
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1734196117591186e-05,
+      "loss": 0.221,
+      "step": 6794
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1730128906847742e-05,
+      "loss": 0.2387,
+      "step": 6795
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1726061948486533e-05,
+      "loss": 0.2805,
+      "step": 6796
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1721995242821451e-05,
+      "loss": 0.2287,
+      "step": 6797
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1717928790166394e-05,
+      "loss": 0.2267,
+      "step": 6798
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1713862590835226e-05,
+      "loss": 0.2381,
+      "step": 6799
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1709796645141787e-05,
+      "loss": 0.2491,
+      "step": 6800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.170573095339991e-05,
+      "loss": 0.2357,
+      "step": 6801
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1701665515923397e-05,
+      "loss": 0.2596,
+      "step": 6802
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1697600333026034e-05,
+      "loss": 0.3026,
+      "step": 6803
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1693535405021599e-05,
+      "loss": 0.2721,
+      "step": 6804
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.168947073222383e-05,
+      "loss": 0.2262,
+      "step": 6805
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.168540631494646e-05,
+      "loss": 0.2959,
+      "step": 6806
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.16813421535032e-05,
+      "loss": 0.2383,
+      "step": 6807
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1677278248207727e-05,
+      "loss": 0.2731,
+      "step": 6808
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1673214599373724e-05,
+      "loss": 0.2894,
+      "step": 6809
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1669151207314837e-05,
+      "loss": 0.2631,
+      "step": 6810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1665088072344683e-05,
+      "loss": 0.2825,
+      "step": 6811
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1661025194776889e-05,
+      "loss": 0.1967,
+      "step": 6812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1656962574925039e-05,
+      "loss": 0.2826,
+      "step": 6813
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1652900213102695e-05,
+      "loss": 0.2244,
+      "step": 6814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1648838109623419e-05,
+      "loss": 0.2258,
+      "step": 6815
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.164477626480073e-05,
+      "loss": 0.2714,
+      "step": 6816
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1640714678948148e-05,
+      "loss": 0.2411,
+      "step": 6817
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1636653352379158e-05,
+      "loss": 0.2643,
+      "step": 6818
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1632592285407224e-05,
+      "loss": 0.2224,
+      "step": 6819
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1628531478345808e-05,
+      "loss": 0.2697,
+      "step": 6820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1624470931508339e-05,
+      "loss": 0.2511,
+      "step": 6821
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.162041064520821e-05,
+      "loss": 0.2328,
+      "step": 6822
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1616350619758833e-05,
+      "loss": 0.2498,
+      "step": 6823
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1612290855473563e-05,
+      "loss": 0.2702,
+      "step": 6824
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1608231352665756e-05,
+      "loss": 0.2228,
+      "step": 6825
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.160417211164874e-05,
+      "loss": 0.28,
+      "step": 6826
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1600113132735817e-05,
+      "loss": 0.2568,
+      "step": 6827
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1596054416240291e-05,
+      "loss": 0.2727,
+      "step": 6828
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.159199596247542e-05,
+      "loss": 0.2651,
+      "step": 6829
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1587937771754452e-05,
+      "loss": 0.2391,
+      "step": 6830
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1583879844390618e-05,
+      "loss": 0.2589,
+      "step": 6831
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1579822180697128e-05,
+      "loss": 0.2158,
+      "step": 6832
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1575764780987161e-05,
+      "loss": 0.2776,
+      "step": 6833
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1571707645573893e-05,
+      "loss": 0.2628,
+      "step": 6834
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.156765077477046e-05,
+      "loss": 0.2711,
+      "step": 6835
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1563594168890002e-05,
+      "loss": 0.2395,
+      "step": 6836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1559537828245619e-05,
+      "loss": 0.2759,
+      "step": 6837
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1555481753150388e-05,
+      "loss": 0.2361,
+      "step": 6838
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1551425943917388e-05,
+      "loss": 0.2342,
+      "step": 6839
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1547370400859652e-05,
+      "loss": 0.2543,
+      "step": 6840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1543315124290207e-05,
+      "loss": 0.2477,
+      "step": 6841
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1539260114522057e-05,
+      "loss": 0.2783,
+      "step": 6842
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1535205371868185e-05,
+      "loss": 0.2677,
+      "step": 6843
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1531150896641547e-05,
+      "loss": 0.2413,
+      "step": 6844
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1527096689155094e-05,
+      "loss": 0.2434,
+      "step": 6845
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.152304274972174e-05,
+      "loss": 0.2415,
+      "step": 6846
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1518989078654388e-05,
+      "loss": 0.2553,
+      "step": 6847
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1514935676265916e-05,
+      "loss": 0.2322,
+      "step": 6848
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.1510882542869178e-05,
+      "loss": 0.2143,
+      "step": 6849
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1506829678777022e-05,
+      "loss": 0.2641,
+      "step": 6850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1502777084302257e-05,
+      "loss": 0.2424,
+      "step": 6851
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1498724759757674e-05,
+      "loss": 0.2337,
+      "step": 6852
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1494672705456061e-05,
+      "loss": 0.273,
+      "step": 6853
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1490620921710167e-05,
+      "loss": 0.2656,
+      "step": 6854
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1486569408832722e-05,
+      "loss": 0.2376,
+      "step": 6855
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1482518167136444e-05,
+      "loss": 0.2323,
+      "step": 6856
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.147846719693402e-05,
+      "loss": 0.2423,
+      "step": 6857
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1474416498538126e-05,
+      "loss": 0.2477,
+      "step": 6858
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1470366072261409e-05,
+      "loss": 0.2381,
+      "step": 6859
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1466315918416488e-05,
+      "loss": 0.2294,
+      "step": 6860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1462266037315988e-05,
+      "loss": 0.2564,
+      "step": 6861
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.145821642927249e-05,
+      "loss": 0.2292,
+      "step": 6862
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.145416709459855e-05,
+      "loss": 0.2357,
+      "step": 6863
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1450118033606727e-05,
+      "loss": 0.2382,
+      "step": 6864
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.144606924660953e-05,
+      "loss": 0.2342,
+      "step": 6865
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1442020733919474e-05,
+      "loss": 0.2447,
+      "step": 6866
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1437972495849033e-05,
+      "loss": 0.2584,
+      "step": 6867
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1433924532710662e-05,
+      "loss": 0.2647,
+      "step": 6868
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1429876844816812e-05,
+      "loss": 0.2416,
+      "step": 6869
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1425829432479892e-05,
+      "loss": 0.2558,
+      "step": 6870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1421782296012301e-05,
+      "loss": 0.2546,
+      "step": 6871
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1417735435726413e-05,
+      "loss": 0.2413,
+      "step": 6872
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1413688851934582e-05,
+      "loss": 0.2309,
+      "step": 6873
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1409642544949138e-05,
+      "loss": 0.2731,
+      "step": 6874
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1405596515082394e-05,
+      "loss": 0.224,
+      "step": 6875
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1401550762646635e-05,
+      "loss": 0.2245,
+      "step": 6876
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1397505287954137e-05,
+      "loss": 0.2418,
+      "step": 6877
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1393460091317142e-05,
+      "loss": 0.2173,
+      "step": 6878
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1389415173047875e-05,
+      "loss": 0.2258,
+      "step": 6879
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1385370533458539e-05,
+      "loss": 0.2526,
+      "step": 6880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.138132617286132e-05,
+      "loss": 0.2383,
+      "step": 6881
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.137728209156837e-05,
+      "loss": 0.2588,
+      "step": 6882
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.137323828989184e-05,
+      "loss": 0.2643,
+      "step": 6883
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1369194768143839e-05,
+      "loss": 0.2864,
+      "step": 6884
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1365151526636457e-05,
+      "loss": 0.2683,
+      "step": 6885
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1361108565681783e-05,
+      "loss": 0.231,
+      "step": 6886
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.135706588559186e-05,
+      "loss": 0.2191,
+      "step": 6887
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1353023486678724e-05,
+      "loss": 0.2891,
+      "step": 6888
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1348981369254382e-05,
+      "loss": 0.2548,
+      "step": 6889
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1344939533630817e-05,
+      "loss": 0.2534,
+      "step": 6890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1340897980120001e-05,
+      "loss": 0.2379,
+      "step": 6891
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1336856709033876e-05,
+      "loss": 0.2177,
+      "step": 6892
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1332815720684358e-05,
+      "loss": 0.2698,
+      "step": 6893
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1328775015383357e-05,
+      "loss": 0.2447,
+      "step": 6894
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1324734593442745e-05,
+      "loss": 0.247,
+      "step": 6895
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1320694455174384e-05,
+      "loss": 0.2553,
+      "step": 6896
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1316654600890106e-05,
+      "loss": 0.2513,
+      "step": 6897
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1312615030901715e-05,
+      "loss": 0.2703,
+      "step": 6898
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1308575745521021e-05,
+      "loss": 0.2547,
+      "step": 6899
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1304536745059777e-05,
+      "loss": 0.2601,
+      "step": 6900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.130049802982973e-05,
+      "loss": 0.2137,
+      "step": 6901
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1296459600142613e-05,
+      "loss": 0.2551,
+      "step": 6902
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1292421456310126e-05,
+      "loss": 0.2543,
+      "step": 6903
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1288383598643948e-05,
+      "loss": 0.2695,
+      "step": 6904
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1284346027455738e-05,
+      "loss": 0.2394,
+      "step": 6905
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1280308743057128e-05,
+      "loss": 0.2469,
+      "step": 6906
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.1276271745759748e-05,
+      "loss": 0.2578,
+      "step": 6907
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1272235035875175e-05,
+      "loss": 0.2269,
+      "step": 6908
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1268198613714975e-05,
+      "loss": 0.2109,
+      "step": 6909
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1264162479590716e-05,
+      "loss": 0.25,
+      "step": 6910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1260126633813908e-05,
+      "loss": 0.2183,
+      "step": 6911
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1256091076696056e-05,
+      "loss": 0.2379,
+      "step": 6912
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1252055808548648e-05,
+      "loss": 0.2425,
+      "step": 6913
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1248020829683135e-05,
+      "loss": 0.2623,
+      "step": 6914
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1243986140410954e-05,
+      "loss": 0.227,
+      "step": 6915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1239951741043527e-05,
+      "loss": 0.238,
+      "step": 6916
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1235917631892234e-05,
+      "loss": 0.25,
+      "step": 6917
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1231883813268456e-05,
+      "loss": 0.258,
+      "step": 6918
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1227850285483533e-05,
+      "loss": 0.2159,
+      "step": 6919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1223817048848789e-05,
+      "loss": 0.2245,
+      "step": 6920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1219784103675532e-05,
+      "loss": 0.258,
+      "step": 6921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1215751450275035e-05,
+      "loss": 0.2308,
+      "step": 6922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1211719088958553e-05,
+      "loss": 0.272,
+      "step": 6923
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1207687020037334e-05,
+      "loss": 0.2693,
+      "step": 6924
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1203655243822568e-05,
+      "loss": 0.2348,
+      "step": 6925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1199623760625466e-05,
+      "loss": 0.2754,
+      "step": 6926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1195592570757186e-05,
+      "loss": 0.2635,
+      "step": 6927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1191561674528868e-05,
+      "loss": 0.2789,
+      "step": 6928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1187531072251635e-05,
+      "loss": 0.2373,
+      "step": 6929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1183500764236592e-05,
+      "loss": 0.2416,
+      "step": 6930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1179470750794803e-05,
+      "loss": 0.2356,
+      "step": 6931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1175441032237334e-05,
+      "loss": 0.2732,
+      "step": 6932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1171411608875212e-05,
+      "loss": 0.2368,
+      "step": 6933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1167382481019436e-05,
+      "loss": 0.2103,
+      "step": 6934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1163353648981e-05,
+      "loss": 0.264,
+      "step": 6935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1159325113070862e-05,
+      "loss": 0.2584,
+      "step": 6936
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1155296873599963e-05,
+      "loss": 0.2318,
+      "step": 6937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1151268930879221e-05,
+      "loss": 0.2603,
+      "step": 6938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.114724128521952e-05,
+      "loss": 0.2404,
+      "step": 6939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1143213936931743e-05,
+      "loss": 0.2567,
+      "step": 6940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1139186886326736e-05,
+      "loss": 0.2525,
+      "step": 6941
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1135160133715312e-05,
+      "loss": 0.2262,
+      "step": 6942
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.113113367940828e-05,
+      "loss": 0.2346,
+      "step": 6943
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1127107523716422e-05,
+      "loss": 0.2463,
+      "step": 6944
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1123081666950487e-05,
+      "loss": 0.2562,
+      "step": 6945
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.111905610942121e-05,
+      "loss": 0.2643,
+      "step": 6946
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1115030851439299e-05,
+      "loss": 0.2432,
+      "step": 6947
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1111005893315445e-05,
+      "loss": 0.2297,
+      "step": 6948
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1106981235360309e-05,
+      "loss": 0.2357,
+      "step": 6949
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1102956877884521e-05,
+      "loss": 0.2394,
+      "step": 6950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1098932821198712e-05,
+      "loss": 0.2575,
+      "step": 6951
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.109490906561347e-05,
+      "loss": 0.28,
+      "step": 6952
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1090885611439359e-05,
+      "loss": 0.2712,
+      "step": 6953
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1086862458986935e-05,
+      "loss": 0.2355,
+      "step": 6954
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1082839608566713e-05,
+      "loss": 0.2444,
+      "step": 6955
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1078817060489204e-05,
+      "loss": 0.2127,
+      "step": 6956
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.107479481506488e-05,
+      "loss": 0.2264,
+      "step": 6957
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1070772872604188e-05,
+      "loss": 0.2275,
+      "step": 6958
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.106675123341757e-05,
+      "loss": 0.2171,
+      "step": 6959
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1062729897815426e-05,
+      "loss": 0.2304,
+      "step": 6960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1058708866108136e-05,
+      "loss": 0.2303,
+      "step": 6961
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1054688138606066e-05,
+      "loss": 0.2351,
+      "step": 6962
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1050667715619555e-05,
+      "loss": 0.2581,
+      "step": 6963
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1046647597458901e-05,
+      "loss": 0.2271,
+      "step": 6964
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1042627784434413e-05,
+      "loss": 0.2943,
+      "step": 6965
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.103860827685634e-05,
+      "loss": 0.2236,
+      "step": 6966
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.103458907503494e-05,
+      "loss": 0.2302,
+      "step": 6967
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.103057017928042e-05,
+      "loss": 0.2333,
+      "step": 6968
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.102655158990298e-05,
+      "loss": 0.2096,
+      "step": 6969
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1022533307212792e-05,
+      "loss": 0.246,
+      "step": 6970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.101851533152e-05,
+      "loss": 0.2531,
+      "step": 6971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1014497663134725e-05,
+      "loss": 0.246,
+      "step": 6972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1010480302367078e-05,
+      "loss": 0.2697,
+      "step": 6973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1006463249527132e-05,
+      "loss": 0.2461,
+      "step": 6974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1002446504924933e-05,
+      "loss": 0.2164,
+      "step": 6975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0998430068870518e-05,
+      "loss": 0.2335,
+      "step": 6976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0994413941673886e-05,
+      "loss": 0.1963,
+      "step": 6977
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0990398123645023e-05,
+      "loss": 0.2513,
+      "step": 6978
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0986382615093887e-05,
+      "loss": 0.2487,
+      "step": 6979
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0982367416330403e-05,
+      "loss": 0.2392,
+      "step": 6980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0978352527664494e-05,
+      "loss": 0.2766,
+      "step": 6981
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0974337949406038e-05,
+      "loss": 0.2445,
+      "step": 6982
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0970323681864896e-05,
+      "loss": 0.269,
+      "step": 6983
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0966309725350908e-05,
+      "loss": 0.2816,
+      "step": 6984
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0962296080173887e-05,
+      "loss": 0.2485,
+      "step": 6985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0958282746643624e-05,
+      "loss": 0.226,
+      "step": 6986
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0954269725069886e-05,
+      "loss": 0.2667,
+      "step": 6987
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0950257015762405e-05,
+      "loss": 0.2252,
+      "step": 6988
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0946244619030915e-05,
+      "loss": 0.234,
+      "step": 6989
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0942232535185099e-05,
+      "loss": 0.2408,
+      "step": 6990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0938220764534624e-05,
+      "loss": 0.2834,
+      "step": 6991
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0934209307389145e-05,
+      "loss": 0.24,
+      "step": 6992
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0930198164058275e-05,
+      "loss": 0.2698,
+      "step": 6993
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.092618733485161e-05,
+      "loss": 0.2378,
+      "step": 6994
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0922176820078729e-05,
+      "loss": 0.266,
+      "step": 6995
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.091816662004917e-05,
+      "loss": 0.2112,
+      "step": 6996
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.091415673507247e-05,
+      "loss": 0.2301,
+      "step": 6997
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0910147165458121e-05,
+      "loss": 0.2637,
+      "step": 6998
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0906137911515595e-05,
+      "loss": 0.2701,
+      "step": 6999
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0902128973554352e-05,
+      "loss": 0.2137,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0898120351883815e-05,
+      "loss": 0.2407,
+      "step": 7001
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0894112046813377e-05,
+      "loss": 0.2429,
+      "step": 7002
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0890104058652429e-05,
+      "loss": 0.2522,
+      "step": 7003
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0886096387710319e-05,
+      "loss": 0.2742,
+      "step": 7004
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0882089034296366e-05,
+      "loss": 0.2223,
+      "step": 7005
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0878081998719892e-05,
+      "loss": 0.2626,
+      "step": 7006
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0874075281290163e-05,
+      "loss": 0.2689,
+      "step": 7007
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0870068882316444e-05,
+      "loss": 0.2905,
+      "step": 7008
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.086606280210796e-05,
+      "loss": 0.2946,
+      "step": 7009
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0862057040973914e-05,
+      "loss": 0.2588,
+      "step": 7010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0858051599223494e-05,
+      "loss": 0.2766,
+      "step": 7011
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0854046477165854e-05,
+      "loss": 0.2738,
+      "step": 7012
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0850041675110118e-05,
+      "loss": 0.2099,
+      "step": 7013
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0846037193365406e-05,
+      "loss": 0.2581,
+      "step": 7014
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0842033032240794e-05,
+      "loss": 0.2348,
+      "step": 7015
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.083802919204534e-05,
+      "loss": 0.2544,
+      "step": 7016
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0834025673088082e-05,
+      "loss": 0.2647,
+      "step": 7017
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.083002247567802e-05,
+      "loss": 0.2839,
+      "step": 7018
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.082601960012414e-05,
+      "loss": 0.2431,
+      "step": 7019
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0822017046735408e-05,
+      "loss": 0.2173,
+      "step": 7020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0818014815820742e-05,
+      "loss": 0.2542,
+      "step": 7021
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.081401290768907e-05,
+      "loss": 0.2397,
+      "step": 7022
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0810011322649264e-05,
+      "loss": 0.2366,
+      "step": 7023
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.0806010061010183e-05,
+      "loss": 0.2918,
+      "step": 7024
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0802009123080666e-05,
+      "loss": 0.2176,
+      "step": 7025
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0798008509169512e-05,
+      "loss": 0.247,
+      "step": 7026
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0794008219585523e-05,
+      "loss": 0.2412,
+      "step": 7027
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0790008254637445e-05,
+      "loss": 0.234,
+      "step": 7028
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0786008614634008e-05,
+      "loss": 0.249,
+      "step": 7029
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0782009299883932e-05,
+      "loss": 0.2691,
+      "step": 7030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0778010310695896e-05,
+      "loss": 0.2449,
+      "step": 7031
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0774011647378554e-05,
+      "loss": 0.2455,
+      "step": 7032
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.077001331024055e-05,
+      "loss": 0.2638,
+      "step": 7033
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0766015299590484e-05,
+      "loss": 0.2556,
+      "step": 7034
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0762017615736938e-05,
+      "loss": 0.2265,
+      "step": 7035
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0758020258988476e-05,
+      "loss": 0.272,
+      "step": 7036
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0754023229653622e-05,
+      "loss": 0.2681,
+      "step": 7037
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0750026528040897e-05,
+      "loss": 0.2157,
+      "step": 7038
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0746030154458771e-05,
+      "loss": 0.2299,
+      "step": 7039
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0742034109215703e-05,
+      "loss": 0.2809,
+      "step": 7040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.073803839262013e-05,
+      "loss": 0.2628,
+      "step": 7041
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0734043004980454e-05,
+      "loss": 0.2279,
+      "step": 7042
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0730047946605047e-05,
+      "loss": 0.2229,
+      "step": 7043
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0726053217802286e-05,
+      "loss": 0.2477,
+      "step": 7044
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0722058818880481e-05,
+      "loss": 0.2637,
+      "step": 7045
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.071806475014794e-05,
+      "loss": 0.2198,
+      "step": 7046
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0714071011912948e-05,
+      "loss": 0.2381,
+      "step": 7047
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0710077604483753e-05,
+      "loss": 0.2368,
+      "step": 7048
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0706084528168589e-05,
+      "loss": 0.2636,
+      "step": 7049
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.070209178327565e-05,
+      "loss": 0.2463,
+      "step": 7050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0698099370113114e-05,
+      "loss": 0.2623,
+      "step": 7051
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0694107288989143e-05,
+      "loss": 0.26,
+      "step": 7052
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.069011554021185e-05,
+      "loss": 0.2588,
+      "step": 7053
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0686124124089334e-05,
+      "loss": 0.2222,
+      "step": 7054
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.068213304092968e-05,
+      "loss": 0.2754,
+      "step": 7055
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0678142291040926e-05,
+      "loss": 0.2602,
+      "step": 7056
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0674151874731104e-05,
+      "loss": 0.2786,
+      "step": 7057
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0670161792308203e-05,
+      "loss": 0.2318,
+      "step": 7058
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0666172044080194e-05,
+      "loss": 0.2563,
+      "step": 7059
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0662182630355031e-05,
+      "loss": 0.2949,
+      "step": 7060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0658193551440627e-05,
+      "loss": 0.2495,
+      "step": 7061
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.065420480764487e-05,
+      "loss": 0.2336,
+      "step": 7062
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0650216399275638e-05,
+      "loss": 0.2615,
+      "step": 7063
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0646228326640771e-05,
+      "loss": 0.2327,
+      "step": 7064
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0642240590048081e-05,
+      "loss": 0.2337,
+      "step": 7065
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0638253189805363e-05,
+      "loss": 0.2383,
+      "step": 7066
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0634266126220372e-05,
+      "loss": 0.2313,
+      "step": 7067
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0630279399600857e-05,
+      "loss": 0.261,
+      "step": 7068
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0626293010254533e-05,
+      "loss": 0.2675,
+      "step": 7069
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0622306958489064e-05,
+      "loss": 0.2236,
+      "step": 7070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0618321244612135e-05,
+      "loss": 0.2399,
+      "step": 7071
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0614335868931368e-05,
+      "loss": 0.2317,
+      "step": 7072
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0610350831754368e-05,
+      "loss": 0.2545,
+      "step": 7073
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0606366133388727e-05,
+      "loss": 0.2246,
+      "step": 7074
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0602381774141995e-05,
+      "loss": 0.272,
+      "step": 7075
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0598397754321691e-05,
+      "loss": 0.2533,
+      "step": 7076
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0594414074235343e-05,
+      "loss": 0.2642,
+      "step": 7077
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.05904307341904e-05,
+      "loss": 0.2671,
+      "step": 7078
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0586447734494333e-05,
+      "loss": 0.3008,
+      "step": 7079
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0582465075454562e-05,
+      "loss": 0.2331,
+      "step": 7080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.0578482757378477e-05,
+      "loss": 0.2642,
+      "step": 7081
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.057450078057346e-05,
+      "loss": 0.2731,
+      "step": 7082
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.057051914534685e-05,
+      "loss": 0.2379,
+      "step": 7083
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0566537852005965e-05,
+      "loss": 0.2255,
+      "step": 7084
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0562556900858106e-05,
+      "loss": 0.2598,
+      "step": 7085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0558576292210533e-05,
+      "loss": 0.2186,
+      "step": 7086
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0554596026370485e-05,
+      "loss": 0.2613,
+      "step": 7087
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0550616103645182e-05,
+      "loss": 0.262,
+      "step": 7088
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0546636524341802e-05,
+      "loss": 0.2352,
+      "step": 7089
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0542657288767515e-05,
+      "loss": 0.2635,
+      "step": 7090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.053867839722945e-05,
+      "loss": 0.2473,
+      "step": 7091
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0534699850034707e-05,
+      "loss": 0.2259,
+      "step": 7092
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0530721647490379e-05,
+      "loss": 0.2578,
+      "step": 7093
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.052674378990352e-05,
+      "loss": 0.2239,
+      "step": 7094
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0522766277581144e-05,
+      "loss": 0.2371,
+      "step": 7095
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0518789110830266e-05,
+      "loss": 0.2465,
+      "step": 7096
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0514812289957852e-05,
+      "loss": 0.244,
+      "step": 7097
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0510835815270857e-05,
+      "loss": 0.2365,
+      "step": 7098
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0506859687076197e-05,
+      "loss": 0.2394,
+      "step": 7099
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.050288390568076e-05,
+      "loss": 0.2678,
+      "step": 7100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0498908471391427e-05,
+      "loss": 0.2716,
+      "step": 7101
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0494933384515033e-05,
+      "loss": 0.2583,
+      "step": 7102
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0490958645358382e-05,
+      "loss": 0.2386,
+      "step": 7103
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0486984254228275e-05,
+      "loss": 0.2247,
+      "step": 7104
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0483010211431467e-05,
+      "loss": 0.2479,
+      "step": 7105
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0479036517274685e-05,
+      "loss": 0.3037,
+      "step": 7106
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0475063172064644e-05,
+      "loss": 0.2537,
+      "step": 7107
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0471090176108016e-05,
+      "loss": 0.2897,
+      "step": 7108
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0467117529711459e-05,
+      "loss": 0.2594,
+      "step": 7109
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0463145233181602e-05,
+      "loss": 0.2765,
+      "step": 7110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.045917328682503e-05,
+      "loss": 0.2374,
+      "step": 7111
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0455201690948325e-05,
+      "loss": 0.2311,
+      "step": 7112
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0451230445858031e-05,
+      "loss": 0.2361,
+      "step": 7113
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0447259551860656e-05,
+      "loss": 0.2059,
+      "step": 7114
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0443289009262702e-05,
+      "loss": 0.237,
+      "step": 7115
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.043931881837062e-05,
+      "loss": 0.2677,
+      "step": 7116
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0435348979490858e-05,
+      "loss": 0.2373,
+      "step": 7117
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0431379492929821e-05,
+      "loss": 0.2237,
+      "step": 7118
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0427410358993883e-05,
+      "loss": 0.246,
+      "step": 7119
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.042344157798941e-05,
+      "loss": 0.2836,
+      "step": 7120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0419473150222716e-05,
+      "loss": 0.238,
+      "step": 7121
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.041550507600011e-05,
+      "loss": 0.2268,
+      "step": 7122
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0411537355627863e-05,
+      "loss": 0.2378,
+      "step": 7123
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.040756998941222e-05,
+      "loss": 0.2305,
+      "step": 7124
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0403602977659392e-05,
+      "loss": 0.2638,
+      "step": 7125
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.039963632067558e-05,
+      "loss": 0.235,
+      "step": 7126
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0395670018766939e-05,
+      "loss": 0.265,
+      "step": 7127
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0391704072239612e-05,
+      "loss": 0.2321,
+      "step": 7128
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0387738481399703e-05,
+      "loss": 0.2328,
+      "step": 7129
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.038377324655329e-05,
+      "loss": 0.2895,
+      "step": 7130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0379808368006432e-05,
+      "loss": 0.2648,
+      "step": 7131
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0375843846065154e-05,
+      "loss": 0.2316,
+      "step": 7132
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0371879681035444e-05,
+      "loss": 0.2221,
+      "step": 7133
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0367915873223286e-05,
+      "loss": 0.2899,
+      "step": 7134
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0363952422934621e-05,
+      "loss": 0.2568,
+      "step": 7135
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0359989330475358e-05,
+      "loss": 0.242,
+      "step": 7136
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.035602659615139e-05,
+      "loss": 0.2522,
+      "step": 7137
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0352064220268575e-05,
+      "loss": 0.2454,
+      "step": 7138
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0348102203132748e-05,
+      "loss": 0.2788,
+      "step": 7139
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0344140545049714e-05,
+      "loss": 0.2592,
+      "step": 7140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.0340179246325242e-05,
+      "loss": 0.235,
+      "step": 7141
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0336218307265093e-05,
+      "loss": 0.2545,
+      "step": 7142
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0332257728174987e-05,
+      "loss": 0.244,
+      "step": 7143
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0328297509360609e-05,
+      "loss": 0.2468,
+      "step": 7144
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0324337651127635e-05,
+      "loss": 0.2395,
+      "step": 7145
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0320378153781698e-05,
+      "loss": 0.263,
+      "step": 7146
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0316419017628411e-05,
+      "loss": 0.2268,
+      "step": 7147
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0312460242973355e-05,
+      "loss": 0.2121,
+      "step": 7148
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.030850183012208e-05,
+      "loss": 0.2331,
+      "step": 7149
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0304543779380123e-05,
+      "loss": 0.2292,
+      "step": 7150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0300586091052978e-05,
+      "loss": 0.2307,
+      "step": 7151
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0296628765446111e-05,
+      "loss": 0.2508,
+      "step": 7152
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0292671802864975e-05,
+      "loss": 0.2416,
+      "step": 7153
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0288715203614979e-05,
+      "loss": 0.3073,
+      "step": 7154
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0284758968001502e-05,
+      "loss": 0.2586,
+      "step": 7155
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0280803096329918e-05,
+      "loss": 0.2397,
+      "step": 7156
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0276847588905543e-05,
+      "loss": 0.2447,
+      "step": 7157
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0272892446033694e-05,
+      "loss": 0.2716,
+      "step": 7158
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0268937668019636e-05,
+      "loss": 0.2346,
+      "step": 7159
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0264983255168614e-05,
+      "loss": 0.2754,
+      "step": 7160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0261029207785852e-05,
+      "loss": 0.2393,
+      "step": 7161
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0257075526176539e-05,
+      "loss": 0.2767,
+      "step": 7162
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.025312221064583e-05,
+      "loss": 0.2385,
+      "step": 7163
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0249169261498868e-05,
+      "loss": 0.2238,
+      "step": 7164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0245216679040749e-05,
+      "loss": 0.2139,
+      "step": 7165
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0241264463576549e-05,
+      "loss": 0.2925,
+      "step": 7166
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0237312615411328e-05,
+      "loss": 0.2468,
+      "step": 7167
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0233361134850096e-05,
+      "loss": 0.2513,
+      "step": 7168
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.022941002219785e-05,
+      "loss": 0.2475,
+      "step": 7169
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.022545927775955e-05,
+      "loss": 0.2623,
+      "step": 7170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.022150890184013e-05,
+      "loss": 0.2599,
+      "step": 7171
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0217558894744503e-05,
+      "loss": 0.2669,
+      "step": 7172
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.021360925677754e-05,
+      "loss": 0.2511,
+      "step": 7173
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0209659988244086e-05,
+      "loss": 0.2399,
+      "step": 7174
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0205711089448976e-05,
+      "loss": 0.2452,
+      "step": 7175
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0201762560696995e-05,
+      "loss": 0.2897,
+      "step": 7176
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0197814402292904e-05,
+      "loss": 0.2699,
+      "step": 7177
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0193866614541444e-05,
+      "loss": 0.2081,
+      "step": 7178
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0189919197747311e-05,
+      "loss": 0.2556,
+      "step": 7179
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0185972152215202e-05,
+      "loss": 0.231,
+      "step": 7180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.018202547824975e-05,
+      "loss": 0.2287,
+      "step": 7181
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0178079176155576e-05,
+      "loss": 0.2662,
+      "step": 7182
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0174133246237286e-05,
+      "loss": 0.2449,
+      "step": 7183
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0170187688799431e-05,
+      "loss": 0.276,
+      "step": 7184
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0166242504146547e-05,
+      "loss": 0.279,
+      "step": 7185
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0162297692583143e-05,
+      "loss": 0.2172,
+      "step": 7186
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0158353254413688e-05,
+      "loss": 0.2196,
+      "step": 7187
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0154409189942648e-05,
+      "loss": 0.2152,
+      "step": 7188
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0150465499474426e-05,
+      "loss": 0.2559,
+      "step": 7189
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.014652218331341e-05,
+      "loss": 0.2871,
+      "step": 7190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0142579241763977e-05,
+      "loss": 0.251,
+      "step": 7191
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0138636675130448e-05,
+      "loss": 0.238,
+      "step": 7192
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.013469448371713e-05,
+      "loss": 0.2305,
+      "step": 7193
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0130752667828297e-05,
+      "loss": 0.2568,
+      "step": 7194
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0126811227768197e-05,
+      "loss": 0.2339,
+      "step": 7195
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.012287016384104e-05,
+      "loss": 0.2548,
+      "step": 7196
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.011892947635103e-05,
+      "loss": 0.2282,
+      "step": 7197
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.01149891656023e-05,
+      "loss": 0.2573,
+      "step": 7198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.0111049231899004e-05,
+      "loss": 0.2509,
+      "step": 7199
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0107109675545232e-05,
+      "loss": 0.2583,
+      "step": 7200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0103170496845052e-05,
+      "loss": 0.2358,
+      "step": 7201
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0099231696102511e-05,
+      "loss": 0.2613,
+      "step": 7202
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0095293273621622e-05,
+      "loss": 0.2549,
+      "step": 7203
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0091355229706362e-05,
+      "loss": 0.2611,
+      "step": 7204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.00874175646607e-05,
+      "loss": 0.2362,
+      "step": 7205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0083480278788551e-05,
+      "loss": 0.2705,
+      "step": 7206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0079543372393805e-05,
+      "loss": 0.2394,
+      "step": 7207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0075606845780343e-05,
+      "loss": 0.316,
+      "step": 7208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0071670699251993e-05,
+      "loss": 0.2378,
+      "step": 7209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0067734933112567e-05,
+      "loss": 0.2396,
+      "step": 7210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0063799547665846e-05,
+      "loss": 0.2612,
+      "step": 7211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.005986454321557e-05,
+      "loss": 0.2438,
+      "step": 7212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.005592992006547e-05,
+      "loss": 0.2588,
+      "step": 7213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0051995678519233e-05,
+      "loss": 0.2399,
+      "step": 7214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.004806181888051e-05,
+      "loss": 0.2378,
+      "step": 7215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0044128341452951e-05,
+      "loss": 0.2516,
+      "step": 7216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0040195246540143e-05,
+      "loss": 0.2408,
+      "step": 7217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0036262534445667e-05,
+      "loss": 0.2468,
+      "step": 7218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0032330205473065e-05,
+      "loss": 0.2445,
+      "step": 7219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0028398259925843e-05,
+      "loss": 0.247,
+      "step": 7220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0024466698107495e-05,
+      "loss": 0.2456,
+      "step": 7221
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0020535520321473e-05,
+      "loss": 0.2542,
+      "step": 7222
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.001660472687119e-05,
+      "loss": 0.231,
+      "step": 7223
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.001267431806006e-05,
+      "loss": 0.2504,
+      "step": 7224
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0008744294191436e-05,
+      "loss": 0.2537,
+      "step": 7225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0004814655568655e-05,
+      "loss": 0.2201,
+      "step": 7226
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.0000885402495025e-05,
+      "loss": 0.2655,
+      "step": 7227
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996956535273818e-06,
+      "loss": 0.2487,
+      "step": 7228
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.993028054208288e-06,
+      "loss": 0.2637,
+      "step": 7229
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.989099959601652e-06,
+      "loss": 0.2208,
+      "step": 7230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.98517225175708e-06,
+      "loss": 0.2529,
+      "step": 7231
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.981244930977747e-06,
+      "loss": 0.2398,
+      "step": 7232
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.977317997566774e-06,
+      "loss": 0.2436,
+      "step": 7233
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.973391451827254e-06,
+      "loss": 0.2435,
+      "step": 7234
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.969465294062259e-06,
+      "loss": 0.2529,
+      "step": 7235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.965539524574826e-06,
+      "loss": 0.2518,
+      "step": 7236
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.961614143667955e-06,
+      "loss": 0.2408,
+      "step": 7237
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.957689151644635e-06,
+      "loss": 0.285,
+      "step": 7238
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.953764548807803e-06,
+      "loss": 0.2406,
+      "step": 7239
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.949840335460386e-06,
+      "loss": 0.2553,
+      "step": 7240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.945916511905262e-06,
+      "loss": 0.2797,
+      "step": 7241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.941993078445292e-06,
+      "loss": 0.2716,
+      "step": 7242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.938070035383304e-06,
+      "loss": 0.2658,
+      "step": 7243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.934147383022094e-06,
+      "loss": 0.2533,
+      "step": 7244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.93022512166442e-06,
+      "loss": 0.2668,
+      "step": 7245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.926303251613034e-06,
+      "loss": 0.2328,
+      "step": 7246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.92238177317063e-06,
+      "loss": 0.2426,
+      "step": 7247
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.918460686639892e-06,
+      "loss": 0.2667,
+      "step": 7248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.91453999232346e-06,
+      "loss": 0.2451,
+      "step": 7249
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.91061969052395e-06,
+      "loss": 0.2513,
+      "step": 7250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.906699781543953e-06,
+      "loss": 0.2403,
+      "step": 7251
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.902780265686017e-06,
+      "loss": 0.2387,
+      "step": 7252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.898861143252666e-06,
+      "loss": 0.2375,
+      "step": 7253
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.894942414546402e-06,
+      "loss": 0.2392,
+      "step": 7254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.891024079869686e-06,
+      "loss": 0.2645,
+      "step": 7255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.887106139524943e-06,
+      "loss": 0.257,
+      "step": 7256
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.883188593814587e-06,
+      "loss": 0.2445,
+      "step": 7257
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.879271443040982e-06,
+      "loss": 0.2565,
+      "step": 7258
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.875354687506477e-06,
+      "loss": 0.2266,
+      "step": 7259
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.871438327513379e-06,
+      "loss": 0.2413,
+      "step": 7260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.867522363363965e-06,
+      "loss": 0.2591,
+      "step": 7261
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.863606795360498e-06,
+      "loss": 0.2556,
+      "step": 7262
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.859691623805187e-06,
+      "loss": 0.2483,
+      "step": 7263
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.855776849000224e-06,
+      "loss": 0.243,
+      "step": 7264
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.851862471247772e-06,
+      "loss": 0.2879,
+      "step": 7265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.84794849084995e-06,
+      "loss": 0.2568,
+      "step": 7266
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.844034908108862e-06,
+      "loss": 0.2302,
+      "step": 7267
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.840121723326574e-06,
+      "loss": 0.2589,
+      "step": 7268
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.836208936805113e-06,
+      "loss": 0.2577,
+      "step": 7269
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.8322965488465e-06,
+      "loss": 0.2261,
+      "step": 7270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.828384559752702e-06,
+      "loss": 0.2386,
+      "step": 7271
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.824472969825657e-06,
+      "loss": 0.2326,
+      "step": 7272
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.820561779367287e-06,
+      "loss": 0.2053,
+      "step": 7273
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.816650988679467e-06,
+      "loss": 0.2503,
+      "step": 7274
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.812740598064048e-06,
+      "loss": 0.2633,
+      "step": 7275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.808830607822856e-06,
+      "loss": 0.2681,
+      "step": 7276
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.80492101825767e-06,
+      "loss": 0.2696,
+      "step": 7277
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.801011829670263e-06,
+      "loss": 0.2379,
+      "step": 7278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.797103042362353e-06,
+      "loss": 0.3109,
+      "step": 7279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.793194656635636e-06,
+      "loss": 0.2353,
+      "step": 7280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.789286672791783e-06,
+      "loss": 0.2137,
+      "step": 7281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.785379091132424e-06,
+      "loss": 0.2531,
+      "step": 7282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.781471911959161e-06,
+      "loss": 0.273,
+      "step": 7283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.777565135573571e-06,
+      "loss": 0.2281,
+      "step": 7284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.773658762277193e-06,
+      "loss": 0.2447,
+      "step": 7285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.76975279237153e-06,
+      "loss": 0.2293,
+      "step": 7286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.765847226158077e-06,
+      "loss": 0.3155,
+      "step": 7287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.761942063938265e-06,
+      "loss": 0.2246,
+      "step": 7288
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.758037306013527e-06,
+      "loss": 0.2419,
+      "step": 7289
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.754132952685237e-06,
+      "loss": 0.2395,
+      "step": 7290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750229004254748e-06,
+      "loss": 0.2401,
+      "step": 7291
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.746325461023391e-06,
+      "loss": 0.2388,
+      "step": 7292
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.742422323292454e-06,
+      "loss": 0.2194,
+      "step": 7293
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.738519591363191e-06,
+      "loss": 0.2426,
+      "step": 7294
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.734617265536843e-06,
+      "loss": 0.2432,
+      "step": 7295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.730715346114602e-06,
+      "loss": 0.2423,
+      "step": 7296
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.726813833397631e-06,
+      "loss": 0.2686,
+      "step": 7297
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.722912727687071e-06,
+      "loss": 0.2211,
+      "step": 7298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.719012029284022e-06,
+      "loss": 0.2471,
+      "step": 7299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.715111738489557e-06,
+      "loss": 0.2355,
+      "step": 7300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.711211855604715e-06,
+      "loss": 0.2566,
+      "step": 7301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.707312380930504e-06,
+      "loss": 0.2302,
+      "step": 7302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.703413314767909e-06,
+      "loss": 0.2171,
+      "step": 7303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.69951465741787e-06,
+      "loss": 0.2483,
+      "step": 7304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.695616409181303e-06,
+      "loss": 0.2688,
+      "step": 7305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.691718570359093e-06,
+      "loss": 0.2225,
+      "step": 7306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.687821141252081e-06,
+      "loss": 0.2382,
+      "step": 7307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.683924122161105e-06,
+      "loss": 0.2732,
+      "step": 7308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.680027513386941e-06,
+      "loss": 0.2461,
+      "step": 7309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.676131315230342e-06,
+      "loss": 0.2308,
+      "step": 7310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.672235527992044e-06,
+      "loss": 0.2321,
+      "step": 7311
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.668340151972733e-06,
+      "loss": 0.2502,
+      "step": 7312
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.66444518747307e-06,
+      "loss": 0.259,
+      "step": 7313
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.66055063479369e-06,
+      "loss": 0.2702,
+      "step": 7314
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.656656494235186e-06,
+      "loss": 0.2323,
+      "step": 7315
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.652762766098124e-06,
+      "loss": 0.2437,
+      "step": 7316
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.648869450683038e-06,
+      "loss": 0.2449,
+      "step": 7317
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.644976548290428e-06,
+      "loss": 0.2317,
+      "step": 7318
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.641084059220772e-06,
+      "loss": 0.2158,
+      "step": 7319
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.637191983774506e-06,
+      "loss": 0.2536,
+      "step": 7320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.633300322252033e-06,
+      "loss": 0.2437,
+      "step": 7321
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.629409074953728e-06,
+      "loss": 0.2414,
+      "step": 7322
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.625518242179937e-06,
+      "loss": 0.2359,
+      "step": 7323
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.621627824230963e-06,
+      "loss": 0.249,
+      "step": 7324
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.6177378214071e-06,
+      "loss": 0.2136,
+      "step": 7325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.61384823400858e-06,
+      "loss": 0.1995,
+      "step": 7326
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.609959062335617e-06,
+      "loss": 0.277,
+      "step": 7327
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.606070306688405e-06,
+      "loss": 0.224,
+      "step": 7328
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.602181967367083e-06,
+      "loss": 0.2408,
+      "step": 7329
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.59829404467178e-06,
+      "loss": 0.2667,
+      "step": 7330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.594406538902575e-06,
+      "loss": 0.2469,
+      "step": 7331
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.590519450359516e-06,
+      "loss": 0.2436,
+      "step": 7332
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.586632779342643e-06,
+      "loss": 0.2328,
+      "step": 7333
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.58274652615193e-06,
+      "loss": 0.2367,
+      "step": 7334
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.578860691087332e-06,
+      "loss": 0.2598,
+      "step": 7335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.574975274448786e-06,
+      "loss": 0.2281,
+      "step": 7336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.571090276536179e-06,
+      "loss": 0.2484,
+      "step": 7337
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.567205697649365e-06,
+      "loss": 0.2383,
+      "step": 7338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.563321538088184e-06,
+      "loss": 0.2548,
+      "step": 7339
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.55943779815242e-06,
+      "loss": 0.275,
+      "step": 7340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.555554478141852e-06,
+      "loss": 0.2677,
+      "step": 7341
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.551671578356193e-06,
+      "loss": 0.2309,
+      "step": 7342
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.547789099095145e-06,
+      "loss": 0.2548,
+      "step": 7343
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.543907040658384e-06,
+      "loss": 0.2334,
+      "step": 7344
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.540025403345538e-06,
+      "loss": 0.2579,
+      "step": 7345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.536144187456203e-06,
+      "loss": 0.2266,
+      "step": 7346
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.532263393289953e-06,
+      "loss": 0.2177,
+      "step": 7347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.528383021146317e-06,
+      "loss": 0.2409,
+      "step": 7348
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.524503071324809e-06,
+      "loss": 0.2318,
+      "step": 7349
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.520623544124898e-06,
+      "loss": 0.2379,
+      "step": 7350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.516744439846009e-06,
+      "loss": 0.257,
+      "step": 7351
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.51286575878756e-06,
+      "loss": 0.264,
+      "step": 7352
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.508987501248922e-06,
+      "loss": 0.2512,
+      "step": 7353
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.50510966752943e-06,
+      "loss": 0.274,
+      "step": 7354
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.501232257928397e-06,
+      "loss": 0.2758,
+      "step": 7355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.497355272745097e-06,
+      "loss": 0.2398,
+      "step": 7356
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.493478712278762e-06,
+      "loss": 0.2206,
+      "step": 7357
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.489602576828617e-06,
+      "loss": 0.2309,
+      "step": 7358
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.485726866693823e-06,
+      "loss": 0.1938,
+      "step": 7359
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.481851582173533e-06,
+      "loss": 0.2682,
+      "step": 7360
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.477976723566857e-06,
+      "loss": 0.2273,
+      "step": 7361
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.47410229117287e-06,
+      "loss": 0.2569,
+      "step": 7362
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.470228285290618e-06,
+      "loss": 0.244,
+      "step": 7363
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.466354706219114e-06,
+      "loss": 0.2335,
+      "step": 7364
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.462481554257328e-06,
+      "loss": 0.224,
+      "step": 7365
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.45860882970422e-06,
+      "loss": 0.2565,
+      "step": 7366
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.454736532858698e-06,
+      "loss": 0.221,
+      "step": 7367
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.450864664019634e-06,
+      "loss": 0.2673,
+      "step": 7368
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.446993223485883e-06,
+      "loss": 0.2435,
+      "step": 7369
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.443122211556256e-06,
+      "loss": 0.2194,
+      "step": 7370
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.439251628529537e-06,
+      "loss": 0.1934,
+      "step": 7371
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.435381474704471e-06,
+      "loss": 0.2282,
+      "step": 7372
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.43151175037977e-06,
+      "loss": 0.2951,
+      "step": 7373
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.42764245585412e-06,
+      "loss": 0.2389,
+      "step": 7374
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.423773591426175e-06,
+      "loss": 0.2453,
+      "step": 7375
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.41990515739453e-06,
+      "loss": 0.218,
+      "step": 7376
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.416037154057787e-06,
+      "loss": 0.2442,
+      "step": 7377
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.412169581714484e-06,
+      "loss": 0.2445,
+      "step": 7378
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.408302440663143e-06,
+      "loss": 0.2277,
+      "step": 7379
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.40443573120224e-06,
+      "loss": 0.2468,
+      "step": 7380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.40056945363022e-06,
+      "loss": 0.2546,
+      "step": 7381
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.396703608245512e-06,
+      "loss": 0.2339,
+      "step": 7382
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.392838195346489e-06,
+      "loss": 0.2527,
+      "step": 7383
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.3889732152315e-06,
+      "loss": 0.2784,
+      "step": 7384
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.385108668198862e-06,
+      "loss": 0.2317,
+      "step": 7385
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.381244554546856e-06,
+      "loss": 0.2372,
+      "step": 7386
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.377380874573729e-06,
+      "loss": 0.2413,
+      "step": 7387
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.373517628577698e-06,
+      "loss": 0.2241,
+      "step": 7388
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.369654816856938e-06,
+      "loss": 0.2349,
+      "step": 7389
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.36579243970961e-06,
+      "loss": 0.2605,
+      "step": 7390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.36193049743382e-06,
+      "loss": 0.2549,
+      "step": 7391
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.358068990327644e-06,
+      "loss": 0.2409,
+      "step": 7392
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.35420791868914e-06,
+      "loss": 0.2792,
+      "step": 7393
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.350347282816315e-06,
+      "loss": 0.2661,
+      "step": 7394
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.346487083007147e-06,
+      "loss": 0.2608,
+      "step": 7395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.34262731955959e-06,
+      "loss": 0.2186,
+      "step": 7396
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.33876799277155e-06,
+      "loss": 0.244,
+      "step": 7397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.334909102940902e-06,
+      "loss": 0.2431,
+      "step": 7398
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.331050650365503e-06,
+      "loss": 0.2261,
+      "step": 7399
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.327192635343155e-06,
+      "loss": 0.244,
+      "step": 7400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.323335058171643e-06,
+      "loss": 0.2549,
+      "step": 7401
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.319477919148705e-06,
+      "loss": 0.2439,
+      "step": 7402
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.31562121857205e-06,
+      "loss": 0.3073,
+      "step": 7403
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.311764956739361e-06,
+      "loss": 0.2754,
+      "step": 7404
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.307909133948275e-06,
+      "loss": 0.2249,
+      "step": 7405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.304053750496397e-06,
+      "loss": 0.2232,
+      "step": 7406
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.300198806681311e-06,
+      "loss": 0.2461,
+      "step": 7407
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.29634430280055e-06,
+      "loss": 0.2486,
+      "step": 7408
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.292490239151627e-06,
+      "loss": 0.2177,
+      "step": 7409
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.288636616032009e-06,
+      "loss": 0.2269,
+      "step": 7410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.284783433739134e-06,
+      "loss": 0.2714,
+      "step": 7411
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.280930692570414e-06,
+      "loss": 0.2223,
+      "step": 7412
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.277078392823213e-06,
+      "loss": 0.2527,
+      "step": 7413
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.273226534794864e-06,
+      "loss": 0.2349,
+      "step": 7414
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.269375118782679e-06,
+      "loss": 0.2613,
+      "step": 7415
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.265524145083925e-06,
+      "loss": 0.2454,
+      "step": 7416
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.261673613995826e-06,
+      "loss": 0.2254,
+      "step": 7417
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.257823525815595e-06,
+      "loss": 0.2719,
+      "step": 7418
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.253973880840388e-06,
+      "loss": 0.2341,
+      "step": 7419
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.250124679367345e-06,
+      "loss": 0.2284,
+      "step": 7420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.246275921693557e-06,
+      "loss": 0.2332,
+      "step": 7421
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.242427608116082e-06,
+      "loss": 0.2466,
+      "step": 7422
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.238579738931962e-06,
+      "loss": 0.27,
+      "step": 7423
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.234732314438187e-06,
+      "loss": 0.2089,
+      "step": 7424
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.230885334931712e-06,
+      "loss": 0.2734,
+      "step": 7425
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.227038800709468e-06,
+      "loss": 0.2515,
+      "step": 7426
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.223192712068347e-06,
+      "loss": 0.281,
+      "step": 7427
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.2193470693052e-06,
+      "loss": 0.2126,
+      "step": 7428
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.215501872716854e-06,
+      "loss": 0.2448,
+      "step": 7429
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.211657122600092e-06,
+      "loss": 0.2684,
+      "step": 7430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.20781281925168e-06,
+      "loss": 0.2652,
+      "step": 7431
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.20396896296833e-06,
+      "loss": 0.24,
+      "step": 7432
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.200125554046723e-06,
+      "loss": 0.2321,
+      "step": 7433
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.196282592783514e-06,
+      "loss": 0.204,
+      "step": 7434
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.19244007947532e-06,
+      "loss": 0.2687,
+      "step": 7435
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.188598014418715e-06,
+      "loss": 0.2208,
+      "step": 7436
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.184756397910254e-06,
+      "loss": 0.2207,
+      "step": 7437
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.180915230246442e-06,
+      "loss": 0.2308,
+      "step": 7438
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.177074511723762e-06,
+      "loss": 0.2295,
+      "step": 7439
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.173234242638654e-06,
+      "loss": 0.2779,
+      "step": 7440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.169394423287523e-06,
+      "loss": 0.2369,
+      "step": 7441
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.165555053966752e-06,
+      "loss": 0.2092,
+      "step": 7442
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.16171613497267e-06,
+      "loss": 0.272,
+      "step": 7443
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.15787766660158e-06,
+      "loss": 0.2167,
+      "step": 7444
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.154039649149758e-06,
+      "loss": 0.2354,
+      "step": 7445
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.150202082913433e-06,
+      "loss": 0.25,
+      "step": 7446
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.146364968188799e-06,
+      "loss": 0.242,
+      "step": 7447
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.142528305272037e-06,
+      "loss": 0.2503,
+      "step": 7448
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.138692094459262e-06,
+      "loss": 0.2369,
+      "step": 7449
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.134856336046577e-06,
+      "loss": 0.2352,
+      "step": 7450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.131021030330036e-06,
+      "loss": 0.2344,
+      "step": 7451
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.127186177605665e-06,
+      "loss": 0.2084,
+      "step": 7452
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.123351778169457e-06,
+      "loss": 0.2594,
+      "step": 7453
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.119517832317364e-06,
+      "loss": 0.2311,
+      "step": 7454
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.115684340345303e-06,
+      "loss": 0.239,
+      "step": 7455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.111851302549167e-06,
+      "loss": 0.2862,
+      "step": 7456
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.108018719224804e-06,
+      "loss": 0.2487,
+      "step": 7457
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.104186590668017e-06,
+      "loss": 0.2426,
+      "step": 7458
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.100354917174603e-06,
+      "loss": 0.2362,
+      "step": 7459
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.096523699040289e-06,
+      "loss": 0.1994,
+      "step": 7460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.092692936560804e-06,
+      "loss": 0.2988,
+      "step": 7461
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.088862630031808e-06,
+      "loss": 0.235,
+      "step": 7462
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.085032779748937e-06,
+      "loss": 0.2192,
+      "step": 7463
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.081203386007809e-06,
+      "loss": 0.2448,
+      "step": 7464
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.077374449103981e-06,
+      "loss": 0.2438,
+      "step": 7465
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.07354596933299e-06,
+      "loss": 0.2608,
+      "step": 7466
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.069717946990335e-06,
+      "loss": 0.2235,
+      "step": 7467
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.065890382371474e-06,
+      "loss": 0.2256,
+      "step": 7468
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.062063275771847e-06,
+      "loss": 0.1932,
+      "step": 7469
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.058236627486831e-06,
+      "loss": 0.2485,
+      "step": 7470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.054410437811784e-06,
+      "loss": 0.2204,
+      "step": 7471
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.05058470704204e-06,
+      "loss": 0.2457,
+      "step": 7472
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.046759435472875e-06,
+      "loss": 0.2545,
+      "step": 7473
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.042934623399536e-06,
+      "loss": 0.2268,
+      "step": 7474
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.039110271117247e-06,
+      "loss": 0.2121,
+      "step": 7475
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.035286378921183e-06,
+      "loss": 0.236,
+      "step": 7476
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.03146294710648e-06,
+      "loss": 0.2369,
+      "step": 7477
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.027639975968268e-06,
+      "loss": 0.2519,
+      "step": 7478
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.023817465801596e-06,
+      "loss": 0.2476,
+      "step": 7479
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.019995416901515e-06,
+      "loss": 0.2449,
+      "step": 7480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.016173829563023e-06,
+      "loss": 0.2415,
+      "step": 7481
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.012352704081085e-06,
+      "loss": 0.2692,
+      "step": 7482
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.008532040750634e-06,
+      "loss": 0.2429,
+      "step": 7483
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.004711839866564e-06,
+      "loss": 0.2253,
+      "step": 7484
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.000892101723728e-06,
+      "loss": 0.2437,
+      "step": 7485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.99707282661696e-06,
+      "loss": 0.2515,
+      "step": 7486
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.99325401484104e-06,
+      "loss": 0.2287,
+      "step": 7487
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.989435666690717e-06,
+      "loss": 0.2276,
+      "step": 7488
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.985617782460714e-06,
+      "loss": 0.2632,
+      "step": 7489
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.981800362445709e-06,
+      "loss": 0.2573,
+      "step": 7490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.977983406940348e-06,
+      "loss": 0.2782,
+      "step": 7491
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.974166916239237e-06,
+      "loss": 0.2486,
+      "step": 7492
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.970350890636941e-06,
+      "loss": 0.2439,
+      "step": 7493
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.966535330428016e-06,
+      "loss": 0.2381,
+      "step": 7494
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.96272023590695e-06,
+      "loss": 0.2617,
+      "step": 7495
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.9589056073682e-06,
+      "loss": 0.2581,
+      "step": 7496
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.955091445106212e-06,
+      "loss": 0.2509,
+      "step": 7497
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.951277749415367e-06,
+      "loss": 0.2649,
+      "step": 7498
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.94746452059003e-06,
+      "loss": 0.2187,
+      "step": 7499
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.943651758924519e-06,
+      "loss": 0.2127,
+      "step": 7500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.939839464713109e-06,
+      "loss": 0.219,
+      "step": 7501
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.936027638250064e-06,
+      "loss": 0.2103,
+      "step": 7502
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.932216279829593e-06,
+      "loss": 0.2412,
+      "step": 7503
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.928405389745862e-06,
+      "loss": 0.2687,
+      "step": 7504
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.924594968293023e-06,
+      "loss": 0.2369,
+      "step": 7505
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.920785015765175e-06,
+      "loss": 0.2521,
+      "step": 7506
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.916975532456384e-06,
+      "loss": 0.2312,
+      "step": 7507
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.913166518660688e-06,
+      "loss": 0.2495,
+      "step": 7508
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.909357974672073e-06,
+      "loss": 0.2571,
+      "step": 7509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.905549900784512e-06,
+      "loss": 0.2443,
+      "step": 7510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.901742297291922e-06,
+      "loss": 0.2211,
+      "step": 7511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.897935164488176e-06,
+      "loss": 0.2597,
+      "step": 7512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.894128502667147e-06,
+      "loss": 0.2242,
+      "step": 7513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.890322312122636e-06,
+      "loss": 0.2333,
+      "step": 7514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.886516593148419e-06,
+      "loss": 0.2099,
+      "step": 7515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.882711346038245e-06,
+      "loss": 0.2064,
+      "step": 7516
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.878906571085814e-06,
+      "loss": 0.2718,
+      "step": 7517
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.875102268584789e-06,
+      "loss": 0.2308,
+      "step": 7518
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.871298438828816e-06,
+      "loss": 0.2556,
+      "step": 7519
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.867495082111476e-06,
+      "loss": 0.2432,
+      "step": 7520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.863692198726338e-06,
+      "loss": 0.2598,
+      "step": 7521
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.859889788966922e-06,
+      "loss": 0.2386,
+      "step": 7522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.856087853126708e-06,
+      "loss": 0.2521,
+      "step": 7523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.852286391499153e-06,
+      "loss": 0.2483,
+      "step": 7524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.848485404377668e-06,
+      "loss": 0.2104,
+      "step": 7525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.844684892055618e-06,
+      "loss": 0.3394,
+      "step": 7526
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.840884854826358e-06,
+      "loss": 0.2838,
+      "step": 7527
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.837085292983187e-06,
+      "loss": 0.2665,
+      "step": 7528
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.83328620681936e-06,
+      "loss": 0.2069,
+      "step": 7529
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.82948759662812e-06,
+      "loss": 0.2564,
+      "step": 7530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.825689462702651e-06,
+      "loss": 0.2361,
+      "step": 7531
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.821891805336114e-06,
+      "loss": 0.245,
+      "step": 7532
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.818094624821624e-06,
+      "loss": 0.2139,
+      "step": 7533
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.81429792145226e-06,
+      "loss": 0.2327,
+      "step": 7534
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.810501695521078e-06,
+      "loss": 0.2468,
+      "step": 7535
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.80670594732108e-06,
+      "loss": 0.2738,
+      "step": 7536
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.802910677145235e-06,
+      "loss": 0.2623,
+      "step": 7537
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.799115885286484e-06,
+      "loss": 0.2435,
+      "step": 7538
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.795321572037719e-06,
+      "loss": 0.246,
+      "step": 7539
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.791527737691806e-06,
+      "loss": 0.2246,
+      "step": 7540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.787734382541567e-06,
+      "loss": 0.2452,
+      "step": 7541
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.783941506879782e-06,
+      "loss": 0.2257,
+      "step": 7542
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.780149110999212e-06,
+      "loss": 0.2395,
+      "step": 7543
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.776357195192566e-06,
+      "loss": 0.2432,
+      "step": 7544
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.772565759752515e-06,
+      "loss": 0.2555,
+      "step": 7545
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.768774804971705e-06,
+      "loss": 0.2227,
+      "step": 7546
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.764984331142734e-06,
+      "loss": 0.2348,
+      "step": 7547
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.761194338558165e-06,
+      "loss": 0.2539,
+      "step": 7548
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.75740482751053e-06,
+      "loss": 0.2597,
+      "step": 7549
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.753615798292306e-06,
+      "loss": 0.2339,
+      "step": 7550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.749827251195966e-06,
+      "loss": 0.2126,
+      "step": 7551
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.746039186513915e-06,
+      "loss": 0.2064,
+      "step": 7552
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.742251604538528e-06,
+      "loss": 0.2323,
+      "step": 7553
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.738464505562161e-06,
+      "loss": 0.281,
+      "step": 7554
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.734677889877104e-06,
+      "loss": 0.1963,
+      "step": 7555
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.730891757775619e-06,
+      "loss": 0.2608,
+      "step": 7556
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.727106109549952e-06,
+      "loss": 0.2267,
+      "step": 7557
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.723320945492285e-06,
+      "loss": 0.2353,
+      "step": 7558
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.71953626589477e-06,
+      "loss": 0.2364,
+      "step": 7559
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.715752071049536e-06,
+      "loss": 0.2736,
+      "step": 7560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.711968361248649e-06,
+      "loss": 0.2499,
+      "step": 7561
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.708185136784166e-06,
+      "loss": 0.2245,
+      "step": 7562
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.70440239794809e-06,
+      "loss": 0.2859,
+      "step": 7563
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.700620145032367e-06,
+      "loss": 0.2797,
+      "step": 7564
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.696838378328953e-06,
+      "loss": 0.2457,
+      "step": 7565
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.69305709812973e-06,
+      "loss": 0.2607,
+      "step": 7566
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.689276304726546e-06,
+      "loss": 0.2288,
+      "step": 7567
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.685495998411233e-06,
+      "loss": 0.2492,
+      "step": 7568
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.681716179475554e-06,
+      "loss": 0.2583,
+      "step": 7569
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.677936848211269e-06,
+      "loss": 0.2371,
+      "step": 7570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.67415800491008e-06,
+      "loss": 0.2514,
+      "step": 7571
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.670379649863635e-06,
+      "loss": 0.2441,
+      "step": 7572
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.666601783363582e-06,
+      "loss": 0.2476,
+      "step": 7573
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.662824405701507e-06,
+      "loss": 0.2305,
+      "step": 7574
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.659047517168956e-06,
+      "loss": 0.2138,
+      "step": 7575
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.655271118057457e-06,
+      "loss": 0.2156,
+      "step": 7576
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.651495208658484e-06,
+      "loss": 0.1977,
+      "step": 7577
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.647719789263473e-06,
+      "loss": 0.2234,
+      "step": 7578
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.643944860163834e-06,
+      "loss": 0.2358,
+      "step": 7579
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.640170421650926e-06,
+      "loss": 0.2319,
+      "step": 7580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.63639647401608e-06,
+      "loss": 0.2418,
+      "step": 7581
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.63262301755058e-06,
+      "loss": 0.2373,
+      "step": 7582
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.628850052545677e-06,
+      "loss": 0.2189,
+      "step": 7583
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.625077579292589e-06,
+      "loss": 0.2302,
+      "step": 7584
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.621305598082492e-06,
+      "loss": 0.2551,
+      "step": 7585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.617534109206511e-06,
+      "loss": 0.256,
+      "step": 7586
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.613763112955763e-06,
+      "loss": 0.2385,
+      "step": 7587
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.609992609621303e-06,
+      "loss": 0.2396,
+      "step": 7588
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.606222599494142e-06,
+      "loss": 0.2338,
+      "step": 7589
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.60245308286528e-06,
+      "loss": 0.2078,
+      "step": 7590
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.598684060025653e-06,
+      "loss": 0.2222,
+      "step": 7591
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.594915531266179e-06,
+      "loss": 0.2094,
+      "step": 7592
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.591147496877728e-06,
+      "loss": 0.2288,
+      "step": 7593
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.587379957151121e-06,
+      "loss": 0.2289,
+      "step": 7594
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.583612912377167e-06,
+      "loss": 0.2365,
+      "step": 7595
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.579846362846624e-06,
+      "loss": 0.2404,
+      "step": 7596
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.576080308850189e-06,
+      "loss": 0.2222,
+      "step": 7597
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.57231475067856e-06,
+      "loss": 0.201,
+      "step": 7598
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.568549688622365e-06,
+      "loss": 0.2451,
+      "step": 7599
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.564785122972225e-06,
+      "loss": 0.2978,
+      "step": 7600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.561021054018692e-06,
+      "loss": 0.2174,
+      "step": 7601
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.55725748205229e-06,
+      "loss": 0.2576,
+      "step": 7602
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.55349440736352e-06,
+      "loss": 0.29,
+      "step": 7603
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.549731830242823e-06,
+      "loss": 0.2507,
+      "step": 7604
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.545969750980613e-06,
+      "loss": 0.2506,
+      "step": 7605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.542208169867263e-06,
+      "loss": 0.2319,
+      "step": 7606
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.538447087193104e-06,
+      "loss": 0.2367,
+      "step": 7607
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.534686503248428e-06,
+      "loss": 0.2479,
+      "step": 7608
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.530926418323504e-06,
+      "loss": 0.2204,
+      "step": 7609
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.527166832708538e-06,
+      "loss": 0.2086,
+      "step": 7610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.523407746693727e-06,
+      "loss": 0.2279,
+      "step": 7611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.519649160569201e-06,
+      "loss": 0.2472,
+      "step": 7612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.515891074625068e-06,
+      "loss": 0.2623,
+      "step": 7613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.51213348915139e-06,
+      "loss": 0.2493,
+      "step": 7614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.508376404438194e-06,
+      "loss": 0.2552,
+      "step": 7615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.504619820775461e-06,
+      "loss": 0.2571,
+      "step": 7616
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.500863738453151e-06,
+      "loss": 0.2496,
+      "step": 7617
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.49710815776117e-06,
+      "loss": 0.2627,
+      "step": 7618
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.493353078989382e-06,
+      "loss": 0.2924,
+      "step": 7619
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.48959850242763e-06,
+      "loss": 0.2731,
+      "step": 7620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.485844428365706e-06,
+      "loss": 0.2329,
+      "step": 7621
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.482090857093363e-06,
+      "loss": 0.2444,
+      "step": 7622
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.478337788900315e-06,
+      "loss": 0.2344,
+      "step": 7623
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.474585224076236e-06,
+      "loss": 0.2391,
+      "step": 7624
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.470833162910776e-06,
+      "loss": 0.2264,
+      "step": 7625
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.467081605693528e-06,
+      "loss": 0.2956,
+      "step": 7626
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.463330552714048e-06,
+      "loss": 0.2522,
+      "step": 7627
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.459580004261866e-06,
+      "loss": 0.2345,
+      "step": 7628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.455829960626466e-06,
+      "loss": 0.2155,
+      "step": 7629
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.452080422097287e-06,
+      "loss": 0.2236,
+      "step": 7630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.448331388963733e-06,
+      "loss": 0.2261,
+      "step": 7631
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.444582861515167e-06,
+      "loss": 0.2349,
+      "step": 7632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.440834840040928e-06,
+      "loss": 0.2798,
+      "step": 7633
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.437087324830293e-06,
+      "loss": 0.2681,
+      "step": 7634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.433340316172514e-06,
+      "loss": 0.2217,
+      "step": 7635
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.4295938143568e-06,
+      "loss": 0.225,
+      "step": 7636
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.425847819672328e-06,
+      "loss": 0.1994,
+      "step": 7637
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.422102332408223e-06,
+      "loss": 0.2546,
+      "step": 7638
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.418357352853578e-06,
+      "loss": 0.2575,
+      "step": 7639
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.41461288129744e-06,
+      "loss": 0.2495,
+      "step": 7640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.410868918028835e-06,
+      "loss": 0.2303,
+      "step": 7641
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.407125463336734e-06,
+      "loss": 0.235,
+      "step": 7642
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.40338251751006e-06,
+      "loss": 0.2097,
+      "step": 7643
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.399640080837729e-06,
+      "loss": 0.231,
+      "step": 7644
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.395898153608588e-06,
+      "loss": 0.2637,
+      "step": 7645
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.392156736111452e-06,
+      "loss": 0.2464,
+      "step": 7646
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.388415828635104e-06,
+      "loss": 0.2324,
+      "step": 7647
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.384675431468281e-06,
+      "loss": 0.2639,
+      "step": 7648
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.380935544899675e-06,
+      "loss": 0.2278,
+      "step": 7649
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.37719616921796e-06,
+      "loss": 0.2417,
+      "step": 7650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.373457304711742e-06,
+      "loss": 0.2343,
+      "step": 7651
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.369718951669615e-06,
+      "loss": 0.241,
+      "step": 7652
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.365981110380117e-06,
+      "loss": 0.2412,
+      "step": 7653
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.362243781131748e-06,
+      "loss": 0.2511,
+      "step": 7654
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.35850696421297e-06,
+      "loss": 0.2684,
+      "step": 7655
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.354770659912208e-06,
+      "loss": 0.2669,
+      "step": 7656
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.351034868517838e-06,
+      "loss": 0.2438,
+      "step": 7657
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.347299590318218e-06,
+      "loss": 0.2056,
+      "step": 7658
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.343564825601635e-06,
+      "loss": 0.2219,
+      "step": 7659
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.339830574656372e-06,
+      "loss": 0.2814,
+      "step": 7660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.336096837770644e-06,
+      "loss": 0.2443,
+      "step": 7661
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.33236361523264e-06,
+      "loss": 0.268,
+      "step": 7662
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.3286309073305e-06,
+      "loss": 0.2232,
+      "step": 7663
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.324898714352334e-06,
+      "loss": 0.2337,
+      "step": 7664
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.321167036586203e-06,
+      "loss": 0.2434,
+      "step": 7665
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.317435874320141e-06,
+      "loss": 0.2346,
+      "step": 7666
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.313705227842132e-06,
+      "loss": 0.2327,
+      "step": 7667
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.309975097440117e-06,
+      "loss": 0.2308,
+      "step": 7668
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.306245483402015e-06,
+      "loss": 0.2915,
+      "step": 7669
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.302516386015683e-06,
+      "loss": 0.2683,
+      "step": 7670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.298787805568952e-06,
+      "loss": 0.2592,
+      "step": 7671
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.29505974234961e-06,
+      "loss": 0.2424,
+      "step": 7672
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.291332196645398e-06,
+      "loss": 0.2626,
+      "step": 7673
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.287605168744034e-06,
+      "loss": 0.2304,
+      "step": 7674
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.283878658933178e-06,
+      "loss": 0.2248,
+      "step": 7675
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.280152667500454e-06,
+      "loss": 0.2882,
+      "step": 7676
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.276427194733462e-06,
+      "loss": 0.2573,
+      "step": 7677
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.272702240919743e-06,
+      "loss": 0.2293,
+      "step": 7678
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.268977806346803e-06,
+      "loss": 0.2227,
+      "step": 7679
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.265253891302108e-06,
+      "loss": 0.2014,
+      "step": 7680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.261530496073083e-06,
+      "loss": 0.2317,
+      "step": 7681
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.257807620947128e-06,
+      "loss": 0.2174,
+      "step": 7682
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.254085266211578e-06,
+      "loss": 0.2372,
+      "step": 7683
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.250363432153738e-06,
+      "loss": 0.2463,
+      "step": 7684
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.246642119060887e-06,
+      "loss": 0.2141,
+      "step": 7685
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.242921327220245e-06,
+      "loss": 0.2283,
+      "step": 7686
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.239201056918996e-06,
+      "loss": 0.2408,
+      "step": 7687
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.235481308444287e-06,
+      "loss": 0.25,
+      "step": 7688
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.231762082083224e-06,
+      "loss": 0.2554,
+      "step": 7689
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.22804337812287e-06,
+      "loss": 0.2285,
+      "step": 7690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.224325196850254e-06,
+      "loss": 0.2569,
+      "step": 7691
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.220607538552353e-06,
+      "loss": 0.2129,
+      "step": 7692
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.216890403516127e-06,
+      "loss": 0.2702,
+      "step": 7693
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.213173792028466e-06,
+      "loss": 0.2669,
+      "step": 7694
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.209457704376242e-06,
+      "loss": 0.2315,
+      "step": 7695
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.20574214084627e-06,
+      "loss": 0.2113,
+      "step": 7696
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.20202710172534e-06,
+      "loss": 0.2471,
+      "step": 7697
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.19831258730018e-06,
+      "loss": 0.2746,
+      "step": 7698
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.194598597857512e-06,
+      "loss": 0.2517,
+      "step": 7699
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.19088513368398e-06,
+      "loss": 0.2198,
+      "step": 7700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.18717219506622e-06,
+      "loss": 0.2189,
+      "step": 7701
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.183459782290802e-06,
+      "loss": 0.249,
+      "step": 7702
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.179747895644267e-06,
+      "loss": 0.2516,
+      "step": 7703
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.176036535413113e-06,
+      "loss": 0.2377,
+      "step": 7704
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.172325701883802e-06,
+      "loss": 0.2605,
+      "step": 7705
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.168615395342744e-06,
+      "loss": 0.2437,
+      "step": 7706
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.164905616076325e-06,
+      "loss": 0.2083,
+      "step": 7707
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.161196364370878e-06,
+      "loss": 0.2445,
+      "step": 7708
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.157487640512692e-06,
+      "loss": 0.2507,
+      "step": 7709
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.153779444788031e-06,
+      "loss": 0.2495,
+      "step": 7710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.150071777483105e-06,
+      "loss": 0.3038,
+      "step": 7711
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.14636463888409e-06,
+      "loss": 0.2247,
+      "step": 7712
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.142658029277114e-06,
+      "loss": 0.256,
+      "step": 7713
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.138951948948262e-06,
+      "loss": 0.2209,
+      "step": 7714
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.135246398183604e-06,
+      "loss": 0.2794,
+      "step": 7715
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.131541377269134e-06,
+      "loss": 0.2482,
+      "step": 7716
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.127836886490822e-06,
+      "loss": 0.2315,
+      "step": 7717
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.124132926134605e-06,
+      "loss": 0.2558,
+      "step": 7718
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.120429496486365e-06,
+      "loss": 0.2499,
+      "step": 7719
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.116726597831947e-06,
+      "loss": 0.219,
+      "step": 7720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.113024230457157e-06,
+      "loss": 0.2651,
+      "step": 7721
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.109322394647752e-06,
+      "loss": 0.2334,
+      "step": 7722
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.10562109068947e-06,
+      "loss": 0.2195,
+      "step": 7723
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.101920318867981e-06,
+      "loss": 0.2411,
+      "step": 7724
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.098220079468929e-06,
+      "loss": 0.2455,
+      "step": 7725
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.094520372777916e-06,
+      "loss": 0.2876,
+      "step": 7726
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.090821199080503e-06,
+      "loss": 0.2698,
+      "step": 7727
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.087122558662203e-06,
+      "loss": 0.2625,
+      "step": 7728
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.083424451808494e-06,
+      "loss": 0.2924,
+      "step": 7729
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.079726878804803e-06,
+      "loss": 0.2473,
+      "step": 7730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.076029839936539e-06,
+      "loss": 0.2171,
+      "step": 7731
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.072333335489045e-06,
+      "loss": 0.2377,
+      "step": 7732
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.068637365747631e-06,
+      "loss": 0.21,
+      "step": 7733
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.064941930997579e-06,
+      "loss": 0.2366,
+      "step": 7734
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.061247031524108e-06,
+      "loss": 0.2465,
+      "step": 7735
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.05755266761241e-06,
+      "loss": 0.245,
+      "step": 7736
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.053858839547628e-06,
+      "loss": 0.2777,
+      "step": 7737
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.05016554761487e-06,
+      "loss": 0.2413,
+      "step": 7738
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.046472792099191e-06,
+      "loss": 0.2211,
+      "step": 7739
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.042780573285625e-06,
+      "loss": 0.2319,
+      "step": 7740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.039088891459144e-06,
+      "loss": 0.2248,
+      "step": 7741
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.035397746904698e-06,
+      "loss": 0.1991,
+      "step": 7742
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.031707139907175e-06,
+      "loss": 0.254,
+      "step": 7743
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.028017070751437e-06,
+      "loss": 0.2675,
+      "step": 7744
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.024327539722295e-06,
+      "loss": 0.2946,
+      "step": 7745
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.020638547104525e-06,
+      "loss": 0.2564,
+      "step": 7746
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.016950093182851e-06,
+      "loss": 0.2383,
+      "step": 7747
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.013262178241978e-06,
+      "loss": 0.2174,
+      "step": 7748
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.009574802566542e-06,
+      "loss": 0.2678,
+      "step": 7749
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.005887966441152e-06,
+      "loss": 0.2334,
+      "step": 7750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.002201670150381e-06,
+      "loss": 0.2257,
+      "step": 7751
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.998515913978746e-06,
+      "loss": 0.2237,
+      "step": 7752
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.994830698210731e-06,
+      "loss": 0.2375,
+      "step": 7753
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.991146023130777e-06,
+      "loss": 0.235,
+      "step": 7754
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.987461889023272e-06,
+      "loss": 0.2317,
+      "step": 7755
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.98377829617259e-06,
+      "loss": 0.2217,
+      "step": 7756
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.980095244863037e-06,
+      "loss": 0.2342,
+      "step": 7757
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.976412735378882e-06,
+      "loss": 0.2574,
+      "step": 7758
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.972730768004368e-06,
+      "loss": 0.2551,
+      "step": 7759
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.969049343023675e-06,
+      "loss": 0.2445,
+      "step": 7760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.965368460720955e-06,
+      "loss": 0.2357,
+      "step": 7761
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.961688121380313e-06,
+      "loss": 0.2373,
+      "step": 7762
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.958008325285806e-06,
+      "loss": 0.2421,
+      "step": 7763
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.954329072721467e-06,
+      "loss": 0.2674,
+      "step": 7764
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.950650363971272e-06,
+      "loss": 0.2452,
+      "step": 7765
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.946972199319153e-06,
+      "loss": 0.2376,
+      "step": 7766
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.943294579049017e-06,
+      "loss": 0.2428,
+      "step": 7767
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.939617503444712e-06,
+      "loss": 0.2691,
+      "step": 7768
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.93594097279005e-06,
+      "loss": 0.2367,
+      "step": 7769
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.9322649873688e-06,
+      "loss": 0.2553,
+      "step": 7770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.928589547464686e-06,
+      "loss": 0.2598,
+      "step": 7771
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.924914653361406e-06,
+      "loss": 0.2644,
+      "step": 7772
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.921240305342595e-06,
+      "loss": 0.248,
+      "step": 7773
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.91756650369185e-06,
+      "loss": 0.2197,
+      "step": 7774
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.91389324869274e-06,
+      "loss": 0.2137,
+      "step": 7775
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.910220540628781e-06,
+      "loss": 0.2765,
+      "step": 7776
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.906548379783444e-06,
+      "loss": 0.2361,
+      "step": 7777
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.90287676644016e-06,
+      "loss": 0.269,
+      "step": 7778
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.899205700882326e-06,
+      "loss": 0.2476,
+      "step": 7779
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.895535183393277e-06,
+      "loss": 0.2556,
+      "step": 7780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.891865214256334e-06,
+      "loss": 0.2786,
+      "step": 7781
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.888195793754749e-06,
+      "loss": 0.2302,
+      "step": 7782
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.884526922171754e-06,
+      "loss": 0.2419,
+      "step": 7783
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.880858599790523e-06,
+      "loss": 0.216,
+      "step": 7784
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.87719082689419e-06,
+      "loss": 0.2271,
+      "step": 7785
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.87352360376585e-06,
+      "loss": 0.2602,
+      "step": 7786
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.869856930688557e-06,
+      "loss": 0.2481,
+      "step": 7787
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.866190807945311e-06,
+      "loss": 0.2544,
+      "step": 7788
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.862525235819092e-06,
+      "loss": 0.2137,
+      "step": 7789
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.858860214592811e-06,
+      "loss": 0.2367,
+      "step": 7790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.85519574454936e-06,
+      "loss": 0.2263,
+      "step": 7791
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.851531825971575e-06,
+      "loss": 0.2086,
+      "step": 7792
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.847868459142255e-06,
+      "loss": 0.2143,
+      "step": 7793
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.844205644344149e-06,
+      "loss": 0.2174,
+      "step": 7794
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.840543381859969e-06,
+      "loss": 0.2717,
+      "step": 7795
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.836881671972376e-06,
+      "loss": 0.2259,
+      "step": 7796
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.833220514964015e-06,
+      "loss": 0.2116,
+      "step": 7797
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.829559911117456e-06,
+      "loss": 0.2417,
+      "step": 7798
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.825899860715236e-06,
+      "loss": 0.294,
+      "step": 7799
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.822240364039867e-06,
+      "loss": 0.2368,
+      "step": 7800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.818581421373788e-06,
+      "loss": 0.2317,
+      "step": 7801
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.814923032999433e-06,
+      "loss": 0.2345,
+      "step": 7802
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.811265199199154e-06,
+      "loss": 0.2226,
+      "step": 7803
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.807607920255272e-06,
+      "loss": 0.2578,
+      "step": 7804
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.803951196450091e-06,
+      "loss": 0.2351,
+      "step": 7805
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.80029502806584e-06,
+      "loss": 0.2113,
+      "step": 7806
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.796639415384714e-06,
+      "loss": 0.2359,
+      "step": 7807
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.792984358688877e-06,
+      "loss": 0.2137,
+      "step": 7808
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.789329858260445e-06,
+      "loss": 0.2453,
+      "step": 7809
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.785675914381476e-06,
+      "loss": 0.2966,
+      "step": 7810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.782022527334001e-06,
+      "loss": 0.2812,
+      "step": 7811
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.7783696974e-06,
+      "loss": 0.2089,
+      "step": 7812
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.774717424861421e-06,
+      "loss": 0.2343,
+      "step": 7813
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.771065710000162e-06,
+      "loss": 0.2366,
+      "step": 7814
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.767414553098065e-06,
+      "loss": 0.2493,
+      "step": 7815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.763763954436956e-06,
+      "loss": 0.2329,
+      "step": 7816
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.7601139142986e-06,
+      "loss": 0.256,
+      "step": 7817
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.756464432964715e-06,
+      "loss": 0.2375,
+      "step": 7818
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.752815510716992e-06,
+      "loss": 0.2236,
+      "step": 7819
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.749167147837058e-06,
+      "loss": 0.2306,
+      "step": 7820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.745519344606523e-06,
+      "loss": 0.2259,
+      "step": 7821
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.741872101306934e-06,
+      "loss": 0.2476,
+      "step": 7822
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.73822541821979e-06,
+      "loss": 0.2292,
+      "step": 7823
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.734579295626577e-06,
+      "loss": 0.2389,
+      "step": 7824
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.730933733808707e-06,
+      "loss": 0.2583,
+      "step": 7825
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.727288733047553e-06,
+      "loss": 0.2899,
+      "step": 7826
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.723644293624471e-06,
+      "loss": 0.2069,
+      "step": 7827
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.720000415820734e-06,
+      "loss": 0.2399,
+      "step": 7828
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.716357099917596e-06,
+      "loss": 0.2447,
+      "step": 7829
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.71271434619627e-06,
+      "loss": 0.2456,
+      "step": 7830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.709072154937912e-06,
+      "loss": 0.2502,
+      "step": 7831
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.705430526423648e-06,
+      "loss": 0.216,
+      "step": 7832
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.70178946093455e-06,
+      "loss": 0.2719,
+      "step": 7833
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.698148958751648e-06,
+      "loss": 0.2048,
+      "step": 7834
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.694509020155944e-06,
+      "loss": 0.2735,
+      "step": 7835
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.690869645428369e-06,
+      "loss": 0.2746,
+      "step": 7836
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.687230834849824e-06,
+      "loss": 0.263,
+      "step": 7837
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.68359258870118e-06,
+      "loss": 0.2157,
+      "step": 7838
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.679954907263243e-06,
+      "loss": 0.2235,
+      "step": 7839
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.676317790816783e-06,
+      "loss": 0.1929,
+      "step": 7840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.672681239642536e-06,
+      "loss": 0.2486,
+      "step": 7841
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.669045254021178e-06,
+      "loss": 0.249,
+      "step": 7842
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.665409834233357e-06,
+      "loss": 0.2173,
+      "step": 7843
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.661774980559672e-06,
+      "loss": 0.2415,
+      "step": 7844
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.658140693280658e-06,
+      "loss": 0.2319,
+      "step": 7845
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.654506972676843e-06,
+      "loss": 0.2731,
+      "step": 7846
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.650873819028689e-06,
+      "loss": 0.2436,
+      "step": 7847
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.647241232616605e-06,
+      "loss": 0.2357,
+      "step": 7848
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.643609213720985e-06,
+      "loss": 0.2287,
+      "step": 7849
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.639977762622159e-06,
+      "loss": 0.2355,
+      "step": 7850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.63634687960041e-06,
+      "loss": 0.2414,
+      "step": 7851
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.632716564936e-06,
+      "loss": 0.2368,
+      "step": 7852
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.629086818909113e-06,
+      "loss": 0.25,
+      "step": 7853
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.625457641799922e-06,
+      "loss": 0.2317,
+      "step": 7854
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.621829033888537e-06,
+      "loss": 0.263,
+      "step": 7855
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.618200995455023e-06,
+      "loss": 0.2329,
+      "step": 7856
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.614573526779419e-06,
+      "loss": 0.2124,
+      "step": 7857
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.610946628141701e-06,
+      "loss": 0.2166,
+      "step": 7858
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.6073202998218075e-06,
+      "loss": 0.2493,
+      "step": 7859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.603694542099639e-06,
+      "loss": 0.2884,
+      "step": 7860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.600069355255042e-06,
+      "loss": 0.2502,
+      "step": 7861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.596444739567826e-06,
+      "loss": 0.2453,
+      "step": 7862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5928206953177515e-06,
+      "loss": 0.2738,
+      "step": 7863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.589197222784531e-06,
+      "loss": 0.2582,
+      "step": 7864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.585574322247853e-06,
+      "loss": 0.2601,
+      "step": 7865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5819519939873425e-06,
+      "loss": 0.2351,
+      "step": 7866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.578330238282578e-06,
+      "loss": 0.2581,
+      "step": 7867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.574709055413112e-06,
+      "loss": 0.257,
+      "step": 7868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.571088445658449e-06,
+      "loss": 0.2432,
+      "step": 7869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.567468409298018e-06,
+      "loss": 0.2468,
+      "step": 7870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.563848946611249e-06,
+      "loss": 0.2708,
+      "step": 7871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.560230057877495e-06,
+      "loss": 0.249,
+      "step": 7872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.556611743376089e-06,
+      "loss": 0.2346,
+      "step": 7873
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.552994003386302e-06,
+      "loss": 0.2954,
+      "step": 7874
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.549376838187358e-06,
+      "loss": 0.276,
+      "step": 7875
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5457602480584584e-06,
+      "loss": 0.2098,
+      "step": 7876
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.542144233278749e-06,
+      "loss": 0.2387,
+      "step": 7877
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5385287941273065e-06,
+      "loss": 0.2099,
+      "step": 7878
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.534913930883206e-06,
+      "loss": 0.2379,
+      "step": 7879
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.531299643825452e-06,
+      "loss": 0.2368,
+      "step": 7880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.527685933233004e-06,
+      "loss": 0.2484,
+      "step": 7881
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.524072799384792e-06,
+      "loss": 0.2498,
+      "step": 7882
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5204602425596845e-06,
+      "loss": 0.2148,
+      "step": 7883
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.516848263036524e-06,
+      "loss": 0.2289,
+      "step": 7884
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.513236861094094e-06,
+      "loss": 0.2349,
+      "step": 7885
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5096260370111355e-06,
+      "loss": 0.2566,
+      "step": 7886
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.506015791066347e-06,
+      "loss": 0.2461,
+      "step": 7887
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.5024061235383824e-06,
+      "loss": 0.2416,
+      "step": 7888
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.498797034705844e-06,
+      "loss": 0.2528,
+      "step": 7889
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.495188524847311e-06,
+      "loss": 0.2278,
+      "step": 7890
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.4915805942412884e-06,
+      "loss": 0.3122,
+      "step": 7891
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.487973243166265e-06,
+      "loss": 0.2099,
+      "step": 7892
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.4843664719006644e-06,
+      "loss": 0.2518,
+      "step": 7893
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.480760280722872e-06,
+      "loss": 0.2254,
+      "step": 7894
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.4771546699112295e-06,
+      "loss": 0.2687,
+      "step": 7895
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.473549639744032e-06,
+      "loss": 0.263,
+      "step": 7896
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.469945190499525e-06,
+      "loss": 0.2422,
+      "step": 7897
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.466341322455928e-06,
+      "loss": 0.2637,
+      "step": 7898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4627380358913935e-06,
+      "loss": 0.2174,
+      "step": 7899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.459135331084036e-06,
+      "loss": 0.2434,
+      "step": 7900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.455533208311935e-06,
+      "loss": 0.2142,
+      "step": 7901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.451931667853115e-06,
+      "loss": 0.2352,
+      "step": 7902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.448330709985556e-06,
+      "loss": 0.2377,
+      "step": 7903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.444730334987195e-06,
+      "loss": 0.2609,
+      "step": 7904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.441130543135919e-06,
+      "loss": 0.2606,
+      "step": 7905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.437531334709586e-06,
+      "loss": 0.218,
+      "step": 7906
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.43393270998599e-06,
+      "loss": 0.2388,
+      "step": 7907
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.430334669242887e-06,
+      "loss": 0.2213,
+      "step": 7908
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.426737212757995e-06,
+      "loss": 0.2641,
+      "step": 7909
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.423140340808978e-06,
+      "loss": 0.2332,
+      "step": 7910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.419544053673455e-06,
+      "loss": 0.2555,
+      "step": 7911
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.415948351629004e-06,
+      "loss": 0.2503,
+      "step": 7912
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4123532349531526e-06,
+      "loss": 0.2749,
+      "step": 7913
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.4087587039233954e-06,
+      "loss": 0.2653,
+      "step": 7914
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.405164758817167e-06,
+      "loss": 0.2091,
+      "step": 7915
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.40157139991186e-06,
+      "loss": 0.2531,
+      "step": 7916
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.397978627484834e-06,
+      "loss": 0.2345,
+      "step": 7917
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.39438644181339e-06,
+      "loss": 0.2451,
+      "step": 7918
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.390794843174786e-06,
+      "loss": 0.2157,
+      "step": 7919
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.387203831846238e-06,
+      "loss": 0.2301,
+      "step": 7920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.383613408104907e-06,
+      "loss": 0.1973,
+      "step": 7921
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3800235722279315e-06,
+      "loss": 0.249,
+      "step": 7922
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3764343244923834e-06,
+      "loss": 0.2863,
+      "step": 7923
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.372845665175289e-06,
+      "loss": 0.2216,
+      "step": 7924
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.369257594553647e-06,
+      "loss": 0.2283,
+      "step": 7925
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.365670112904396e-06,
+      "loss": 0.2354,
+      "step": 7926
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.362083220504432e-06,
+      "loss": 0.2052,
+      "step": 7927
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.358496917630603e-06,
+      "loss": 0.232,
+      "step": 7928
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.354911204559719e-06,
+      "loss": 0.2593,
+      "step": 7929
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.351326081568534e-06,
+      "loss": 0.2209,
+      "step": 7930
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.347741548933772e-06,
+      "loss": 0.2461,
+      "step": 7931
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.344157606932091e-06,
+      "loss": 0.2773,
+      "step": 7932
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.340574255840128e-06,
+      "loss": 0.206,
+      "step": 7933
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.336991495934453e-06,
+      "loss": 0.2637,
+      "step": 7934
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3334093274916006e-06,
+      "loss": 0.2214,
+      "step": 7935
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.329827750788054e-06,
+      "loss": 0.2453,
+      "step": 7936
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.326246766100258e-06,
+      "loss": 0.2129,
+      "step": 7937
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.322666373704599e-06,
+      "loss": 0.2756,
+      "step": 7938
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.31908657387744e-06,
+      "loss": 0.2488,
+      "step": 7939
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.315507366895078e-06,
+      "loss": 0.2756,
+      "step": 7940
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.311928753033766e-06,
+      "loss": 0.2332,
+      "step": 7941
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.3083507325697285e-06,
+      "loss": 0.2118,
+      "step": 7942
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.304773305779125e-06,
+      "loss": 0.2698,
+      "step": 7943
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.301196472938074e-06,
+      "loss": 0.2199,
+      "step": 7944
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.297620234322656e-06,
+      "loss": 0.2448,
+      "step": 7945
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.2940445902088895e-06,
+      "loss": 0.2226,
+      "step": 7946
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.290469540872769e-06,
+      "loss": 0.202,
+      "step": 7947
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.286895086590231e-06,
+      "loss": 0.2205,
+      "step": 7948
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.2833212276371545e-06,
+      "loss": 0.2462,
+      "step": 7949
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.2797479642894e-06,
+      "loss": 0.2341,
+      "step": 7950
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.2761752968227604e-06,
+      "loss": 0.2297,
+      "step": 7951
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.2726032255129875e-06,
+      "loss": 0.2544,
+      "step": 7952
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.269031750635791e-06,
+      "loss": 0.2093,
+      "step": 7953
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.265460872466826e-06,
+      "loss": 0.2562,
+      "step": 7954
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.261890591281718e-06,
+      "loss": 0.2419,
+      "step": 7955
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.25832090735603e-06,
+      "loss": 0.2361,
+      "step": 7956
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.254751820965283e-06,
+      "loss": 0.2356,
+      "step": 7957
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.251183332384959e-06,
+      "loss": 0.2508,
+      "step": 7958
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.2476154418904886e-06,
+      "loss": 0.2739,
+      "step": 7959
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.244048149757255e-06,
+      "loss": 0.2062,
+      "step": 7960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.240481456260595e-06,
+      "loss": 0.2282,
+      "step": 7961
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.236915361675796e-06,
+      "loss": 0.246,
+      "step": 7962
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.233349866278115e-06,
+      "loss": 0.2408,
+      "step": 7963
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.229784970342749e-06,
+      "loss": 0.2235,
+      "step": 7964
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.226220674144841e-06,
+      "loss": 0.2304,
+      "step": 7965
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.222656977959513e-06,
+      "loss": 0.2361,
+      "step": 7966
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.219093882061819e-06,
+      "loss": 0.2353,
+      "step": 7967
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.2155313867267745e-06,
+      "loss": 0.2412,
+      "step": 7968
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.211969492229346e-06,
+      "loss": 0.2772,
+      "step": 7969
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.208408198844455e-06,
+      "loss": 0.2393,
+      "step": 7970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.204847506846975e-06,
+      "loss": 0.2512,
+      "step": 7971
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.20128741651174e-06,
+      "loss": 0.24,
+      "step": 7972
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.197727928113528e-06,
+      "loss": 0.2382,
+      "step": 7973
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.194169041927083e-06,
+      "loss": 0.2048,
+      "step": 7974
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.190610758227089e-06,
+      "loss": 0.2309,
+      "step": 7975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1870530772881895e-06,
+      "loss": 0.2869,
+      "step": 7976
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.183495999384981e-06,
+      "loss": 0.238,
+      "step": 7977
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.179939524792014e-06,
+      "loss": 0.2603,
+      "step": 7978
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.176383653783786e-06,
+      "loss": 0.2239,
+      "step": 7979
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.172828386634764e-06,
+      "loss": 0.2323,
+      "step": 7980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1692737236193486e-06,
+      "loss": 0.2439,
+      "step": 7981
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1657196650119156e-06,
+      "loss": 0.2301,
+      "step": 7982
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.162166211086773e-06,
+      "loss": 0.2571,
+      "step": 7983
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.158613362118194e-06,
+      "loss": 0.2372,
+      "step": 7984
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.155061118380402e-06,
+      "loss": 0.2287,
+      "step": 7985
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.151509480147572e-06,
+      "loss": 0.2465,
+      "step": 7986
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1479584476938315e-06,
+      "loss": 0.2424,
+      "step": 7987
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1444080212932715e-06,
+      "loss": 0.2414,
+      "step": 7988
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.140858201219927e-06,
+      "loss": 0.2309,
+      "step": 7989
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.13730898774778e-06,
+      "loss": 0.2389,
+      "step": 7990
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.133760381150785e-06,
+      "loss": 0.2586,
+      "step": 7991
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.130212381702832e-06,
+      "loss": 0.204,
+      "step": 7992
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.126664989677771e-06,
+      "loss": 0.2399,
+      "step": 7993
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.123118205349404e-06,
+      "loss": 0.2337,
+      "step": 7994
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.119572028991482e-06,
+      "loss": 0.2698,
+      "step": 7995
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.116026460877725e-06,
+      "loss": 0.2465,
+      "step": 7996
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.112481501281787e-06,
+      "loss": 0.2343,
+      "step": 7997
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.108937150477279e-06,
+      "loss": 0.2384,
+      "step": 7998
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.1053934087377805e-06,
+      "loss": 0.2493,
+      "step": 7999
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.101850276336803e-06,
+      "loss": 0.2684,
+      "step": 8000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.0983077535478236e-06,
+      "loss": 0.2184,
+      "step": 8001
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.094765840644269e-06,
+      "loss": 0.2326,
+      "step": 8002
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.09122453789951e-06,
+      "loss": 0.2634,
+      "step": 8003
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.087683845586894e-06,
+      "loss": 0.2334,
+      "step": 8004
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.084143763979699e-06,
+      "loss": 0.2454,
+      "step": 8005
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.080604293351157e-06,
+      "loss": 0.2745,
+      "step": 8006
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.077065433974471e-06,
+      "loss": 0.2331,
+      "step": 8007
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.07352718612278e-06,
+      "loss": 0.209,
+      "step": 8008
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.0699895500691785e-06,
+      "loss": 0.2266,
+      "step": 8009
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.0664525260867175e-06,
+      "loss": 0.2334,
+      "step": 8010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.0629161144484e-06,
+      "loss": 0.2352,
+      "step": 8011
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.059380315427173e-06,
+      "loss": 0.2616,
+      "step": 8012
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.055845129295957e-06,
+      "loss": 0.2805,
+      "step": 8013
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.052310556327602e-06,
+      "loss": 0.2205,
+      "step": 8014
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.04877659679493e-06,
+      "loss": 0.2309,
+      "step": 8015
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.045243250970701e-06,
+      "loss": 0.1931,
+      "step": 8016
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.041710519127636e-06,
+      "loss": 0.2388,
+      "step": 8017
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0381784015384035e-06,
+      "loss": 0.2495,
+      "step": 8018
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0346468984756285e-06,
+      "loss": 0.2358,
+      "step": 8019
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.031116010211881e-06,
+      "loss": 0.266,
+      "step": 8020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.027585737019699e-06,
+      "loss": 0.2492,
+      "step": 8021
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0240560791715545e-06,
+      "loss": 0.2693,
+      "step": 8022
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.020527036939893e-06,
+      "loss": 0.222,
+      "step": 8023
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.016998610597093e-06,
+      "loss": 0.2384,
+      "step": 8024
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.013470800415495e-06,
+      "loss": 0.2431,
+      "step": 8025
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.009943606667387e-06,
+      "loss": 0.2644,
+      "step": 8026
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.006417029625017e-06,
+      "loss": 0.2269,
+      "step": 8027
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.0028910695605715e-06,
+      "loss": 0.2122,
+      "step": 8028
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.999365726746211e-06,
+      "loss": 0.2355,
+      "step": 8029
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.995841001454032e-06,
+      "loss": 0.2427,
+      "step": 8030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.992316893956081e-06,
+      "loss": 0.2407,
+      "step": 8031
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.988793404524371e-06,
+      "loss": 0.2324,
+      "step": 8032
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.985270533430859e-06,
+      "loss": 0.2177,
+      "step": 8033
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.981748280947454e-06,
+      "loss": 0.2804,
+      "step": 8034
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.978226647346016e-06,
+      "loss": 0.2249,
+      "step": 8035
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.974705632898354e-06,
+      "loss": 0.23,
+      "step": 8036
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.971185237876246e-06,
+      "loss": 0.2553,
+      "step": 8037
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.967665462551405e-06,
+      "loss": 0.2393,
+      "step": 8038
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.964146307195498e-06,
+      "loss": 0.2653,
+      "step": 8039
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.9606277720801575e-06,
+      "loss": 0.2373,
+      "step": 8040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.957109857476954e-06,
+      "loss": 0.2233,
+      "step": 8041
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.953592563657413e-06,
+      "loss": 0.2152,
+      "step": 8042
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.950075890893016e-06,
+      "loss": 0.2328,
+      "step": 8043
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.946559839455187e-06,
+      "loss": 0.1749,
+      "step": 8044
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.943044409615324e-06,
+      "loss": 0.2396,
+      "step": 8045
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.939529601644751e-06,
+      "loss": 0.2485,
+      "step": 8046
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.936015415814756e-06,
+      "loss": 0.2261,
+      "step": 8047
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.9325018523965855e-06,
+      "loss": 0.2471,
+      "step": 8048
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.928988911661426e-06,
+      "loss": 0.2144,
+      "step": 8049
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.925476593880422e-06,
+      "loss": 0.2426,
+      "step": 8050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.921964899324669e-06,
+      "loss": 0.2287,
+      "step": 8051
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.918453828265209e-06,
+      "loss": 0.2569,
+      "step": 8052
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.914943380973049e-06,
+      "loss": 0.2224,
+      "step": 8053
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.9114335577191395e-06,
+      "loss": 0.2196,
+      "step": 8054
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.907924358774372e-06,
+      "loss": 0.219,
+      "step": 8055
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.904415784409616e-06,
+      "loss": 0.2361,
+      "step": 8056
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.900907834895671e-06,
+      "loss": 0.2434,
+      "step": 8057
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.897400510503294e-06,
+      "loss": 0.2773,
+      "step": 8058
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.8938938115031976e-06,
+      "loss": 0.2162,
+      "step": 8059
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.890387738166042e-06,
+      "loss": 0.2399,
+      "step": 8060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.886882290762434e-06,
+      "loss": 0.2632,
+      "step": 8061
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.883377469562951e-06,
+      "loss": 0.2225,
+      "step": 8062
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.879873274838098e-06,
+      "loss": 0.2135,
+      "step": 8063
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.8763697068583535e-06,
+      "loss": 0.2764,
+      "step": 8064
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.872866765894134e-06,
+      "loss": 0.2219,
+      "step": 8065
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.86936445221581e-06,
+      "loss": 0.2167,
+      "step": 8066
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.865862766093704e-06,
+      "loss": 0.2512,
+      "step": 8067
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.862361707798092e-06,
+      "loss": 0.2072,
+      "step": 8068
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.8588612775991925e-06,
+      "loss": 0.2275,
+      "step": 8069
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.8553614757671964e-06,
+      "loss": 0.2227,
+      "step": 8070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.851862302572227e-06,
+      "loss": 0.2391,
+      "step": 8071
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.848363758284358e-06,
+      "loss": 0.2608,
+      "step": 8072
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.844865843173636e-06,
+      "loss": 0.1972,
+      "step": 8073
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.841368557510035e-06,
+      "loss": 0.221,
+      "step": 8074
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.837871901563492e-06,
+      "loss": 0.2231,
+      "step": 8075
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.834375875603894e-06,
+      "loss": 0.2324,
+      "step": 8076
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.830880479901073e-06,
+      "loss": 0.2326,
+      "step": 8077
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.827385714724828e-06,
+      "loss": 0.2131,
+      "step": 8078
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.823891580344896e-06,
+      "loss": 0.2405,
+      "step": 8079
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.82039807703096e-06,
+      "loss": 0.2507,
+      "step": 8080
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.816905205052679e-06,
+      "loss": 0.2333,
+      "step": 8081
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.813412964679632e-06,
+      "loss": 0.2779,
+      "step": 8082
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.8099213561813826e-06,
+      "loss": 0.207,
+      "step": 8083
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.806430379827411e-06,
+      "loss": 0.2372,
+      "step": 8084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.8029400358871656e-06,
+      "loss": 0.2423,
+      "step": 8085
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.799450324630057e-06,
+      "loss": 0.2321,
+      "step": 8086
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.79596124632543e-06,
+      "loss": 0.2088,
+      "step": 8087
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.792472801242579e-06,
+      "loss": 0.2122,
+      "step": 8088
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.78898498965077e-06,
+      "loss": 0.2898,
+      "step": 8089
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.785497811819201e-06,
+      "loss": 0.2087,
+      "step": 8090
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.782011268017024e-06,
+      "loss": 0.2738,
+      "step": 8091
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.778525358513349e-06,
+      "loss": 0.2476,
+      "step": 8092
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.775040083577224e-06,
+      "loss": 0.2198,
+      "step": 8093
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7715554434776715e-06,
+      "loss": 0.235,
+      "step": 8094
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.768071438483641e-06,
+      "loss": 0.2317,
+      "step": 8095
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.76458806886404e-06,
+      "loss": 0.247,
+      "step": 8096
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.761105334887739e-06,
+      "loss": 0.2201,
+      "step": 8097
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.757623236823545e-06,
+      "loss": 0.244,
+      "step": 8098
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7541417749402154e-06,
+      "loss": 0.2411,
+      "step": 8099
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.750660949506478e-06,
+      "loss": 0.2293,
+      "step": 8100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.747180760790985e-06,
+      "loss": 0.2577,
+      "step": 8101
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.74370120906235e-06,
+      "loss": 0.2896,
+      "step": 8102
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.740222294589149e-06,
+      "loss": 0.2094,
+      "step": 8103
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.736744017639889e-06,
+      "loss": 0.2507,
+      "step": 8104
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7332663784830495e-06,
+      "loss": 0.2397,
+      "step": 8105
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.729789377387043e-06,
+      "loss": 0.2547,
+      "step": 8106
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.726313014620234e-06,
+      "loss": 0.2475,
+      "step": 8107
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.7228372904509584e-06,
+      "loss": 0.225,
+      "step": 8108
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.71936220514747e-06,
+      "loss": 0.2319,
+      "step": 8109
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.715887758977991e-06,
+      "loss": 0.2623,
+      "step": 8110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.712413952210705e-06,
+      "loss": 0.2317,
+      "step": 8111
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.708940785113723e-06,
+      "loss": 0.2237,
+      "step": 8112
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.70546825795513e-06,
+      "loss": 0.2428,
+      "step": 8113
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.701996371002945e-06,
+      "loss": 0.2413,
+      "step": 8114
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.698525124525135e-06,
+      "loss": 0.2125,
+      "step": 8115
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.695054518789645e-06,
+      "loss": 0.2592,
+      "step": 8116
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.691584554064333e-06,
+      "loss": 0.2478,
+      "step": 8117
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.6881152306170245e-06,
+      "loss": 0.2215,
+      "step": 8118
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.684646548715506e-06,
+      "loss": 0.2554,
+      "step": 8119
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.681178508627503e-06,
+      "loss": 0.2203,
+      "step": 8120
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.677711110620684e-06,
+      "loss": 0.2599,
+      "step": 8121
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.674244354962692e-06,
+      "loss": 0.2333,
+      "step": 8122
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.67077824192109e-06,
+      "loss": 0.2151,
+      "step": 8123
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.667312771763423e-06,
+      "loss": 0.2338,
+      "step": 8124
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.663847944757165e-06,
+      "loss": 0.2758,
+      "step": 8125
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.660383761169735e-06,
+      "loss": 0.2532,
+      "step": 8126
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.656920221268526e-06,
+      "loss": 0.2347,
+      "step": 8127
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.653457325320861e-06,
+      "loss": 0.2225,
+      "step": 8128
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.649995073594019e-06,
+      "loss": 0.2443,
+      "step": 8129
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.646533466355241e-06,
+      "loss": 0.2636,
+      "step": 8130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.643072503871704e-06,
+      "loss": 0.223,
+      "step": 8131
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.639612186410531e-06,
+      "loss": 0.2416,
+      "step": 8132
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.636152514238822e-06,
+      "loss": 0.2053,
+      "step": 8133
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.6326934876235855e-06,
+      "loss": 0.2596,
+      "step": 8134
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.629235106831821e-06,
+      "loss": 0.219,
+      "step": 8135
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.625777372130456e-06,
+      "loss": 0.2585,
+      "step": 8136
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.622320283786366e-06,
+      "loss": 0.2468,
+      "step": 8137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.618863842066396e-06,
+      "loss": 0.2589,
+      "step": 8138
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.61540804723732e-06,
+      "loss": 0.2394,
+      "step": 8139
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.611952899565869e-06,
+      "loss": 0.2281,
+      "step": 8140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.608498399318733e-06,
+      "loss": 0.2568,
+      "step": 8141
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.605044546762542e-06,
+      "loss": 0.2367,
+      "step": 8142
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.601591342163877e-06,
+      "loss": 0.2312,
+      "step": 8143
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.59813878578927e-06,
+      "loss": 0.2012,
+      "step": 8144
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.594686877905202e-06,
+      "loss": 0.2272,
+      "step": 8145
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.591235618778112e-06,
+      "loss": 0.2447,
+      "step": 8146
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.58778500867438e-06,
+      "loss": 0.2128,
+      "step": 8147
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.584335047860332e-06,
+      "loss": 0.2539,
+      "step": 8148
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5808857366022615e-06,
+      "loss": 0.2239,
+      "step": 8149
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.577437075166398e-06,
+      "loss": 0.2341,
+      "step": 8150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.573989063818911e-06,
+      "loss": 0.2551,
+      "step": 8151
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5705417028259485e-06,
+      "loss": 0.299,
+      "step": 8152
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5670949924535775e-06,
+      "loss": 0.2192,
+      "step": 8153
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.563648932967844e-06,
+      "loss": 0.2128,
+      "step": 8154
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.560203524634722e-06,
+      "loss": 0.23,
+      "step": 8155
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.556758767720138e-06,
+      "loss": 0.2517,
+      "step": 8156
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.553314662489982e-06,
+      "loss": 0.2567,
+      "step": 8157
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.549871209210085e-06,
+      "loss": 0.2197,
+      "step": 8158
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.546428408146213e-06,
+      "loss": 0.2239,
+      "step": 8159
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.542986259564108e-06,
+      "loss": 0.2163,
+      "step": 8160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.5395447637294475e-06,
+      "loss": 0.2408,
+      "step": 8161
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.536103920907852e-06,
+      "loss": 0.2105,
+      "step": 8162
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.532663731364913e-06,
+      "loss": 0.2539,
+      "step": 8163
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.529224195366148e-06,
+      "loss": 0.2634,
+      "step": 8164
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.525785313177043e-06,
+      "loss": 0.2335,
+      "step": 8165
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.522347085063022e-06,
+      "loss": 0.2738,
+      "step": 8166
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.518909511289461e-06,
+      "loss": 0.272,
+      "step": 8167
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.515472592121687e-06,
+      "loss": 0.2451,
+      "step": 8168
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.512036327824974e-06,
+      "loss": 0.241,
+      "step": 8169
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.508600718664543e-06,
+      "loss": 0.2576,
+      "step": 8170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.505165764905578e-06,
+      "loss": 0.2618,
+      "step": 8171
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.501731466813202e-06,
+      "loss": 0.2359,
+      "step": 8172
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.498297824652477e-06,
+      "loss": 0.2645,
+      "step": 8173
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4948648386884396e-06,
+      "loss": 0.2329,
+      "step": 8174
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.491432509186057e-06,
+      "loss": 0.2219,
+      "step": 8175
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.48800083641025e-06,
+      "loss": 0.2431,
+      "step": 8176
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.484569820625888e-06,
+      "loss": 0.3161,
+      "step": 8177
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.481139462097787e-06,
+      "loss": 0.2262,
+      "step": 8178
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.477709761090728e-06,
+      "loss": 0.2356,
+      "step": 8179
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.474280717869421e-06,
+      "loss": 0.2367,
+      "step": 8180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.470852332698533e-06,
+      "loss": 0.2264,
+      "step": 8181
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.467424605842688e-06,
+      "loss": 0.2358,
+      "step": 8182
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.463997537566449e-06,
+      "loss": 0.2341,
+      "step": 8183
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.460571128134331e-06,
+      "loss": 0.2267,
+      "step": 8184
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.457145377810796e-06,
+      "loss": 0.2483,
+      "step": 8185
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.453720286860255e-06,
+      "loss": 0.2472,
+      "step": 8186
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.450295855547082e-06,
+      "loss": 0.2214,
+      "step": 8187
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.44687208413558e-06,
+      "loss": 0.289,
+      "step": 8188
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.443448972890009e-06,
+      "loss": 0.2539,
+      "step": 8189
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.440026522074587e-06,
+      "loss": 0.2533,
+      "step": 8190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.436604731953467e-06,
+      "loss": 0.2358,
+      "step": 8191
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.433183602790759e-06,
+      "loss": 0.2399,
+      "step": 8192
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.429763134850519e-06,
+      "loss": 0.2276,
+      "step": 8193
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.426343328396746e-06,
+      "loss": 0.2258,
+      "step": 8194
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.4229241836934075e-06,
+      "loss": 0.2624,
+      "step": 8195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.419505701004403e-06,
+      "loss": 0.2368,
+      "step": 8196
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.4160878805935765e-06,
+      "loss": 0.2221,
+      "step": 8197
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.412670722724744e-06,
+      "loss": 0.2582,
+      "step": 8198
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.409254227661648e-06,
+      "loss": 0.1976,
+      "step": 8199
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.405838395667988e-06,
+      "loss": 0.2483,
+      "step": 8200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.402423227007413e-06,
+      "loss": 0.2424,
+      "step": 8201
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.399008721943519e-06,
+      "loss": 0.2793,
+      "step": 8202
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.395594880739849e-06,
+      "loss": 0.2284,
+      "step": 8203
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.392181703659902e-06,
+      "loss": 0.2485,
+      "step": 8204
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3887691909671185e-06,
+      "loss": 0.255,
+      "step": 8205
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.385357342924895e-06,
+      "loss": 0.2666,
+      "step": 8206
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.381946159796569e-06,
+      "loss": 0.2106,
+      "step": 8207
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.378535641845429e-06,
+      "loss": 0.2797,
+      "step": 8208
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.375125789334714e-06,
+      "loss": 0.2573,
+      "step": 8209
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.37171660252761e-06,
+      "loss": 0.2277,
+      "step": 8210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.368308081687248e-06,
+      "loss": 0.2784,
+      "step": 8211
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.364900227076721e-06,
+      "loss": 0.2142,
+      "step": 8212
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.36149303895905e-06,
+      "loss": 0.2461,
+      "step": 8213
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.358086517597229e-06,
+      "loss": 0.2682,
+      "step": 8214
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.354680663254181e-06,
+      "loss": 0.2282,
+      "step": 8215
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.351275476192784e-06,
+      "loss": 0.2002,
+      "step": 8216
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.347870956675863e-06,
+      "loss": 0.2578,
+      "step": 8217
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.344467104966195e-06,
+      "loss": 0.2448,
+      "step": 8218
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.341063921326498e-06,
+      "loss": 0.2665,
+      "step": 8219
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.337661406019453e-06,
+      "loss": 0.2327,
+      "step": 8220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.334259559307675e-06,
+      "loss": 0.2409,
+      "step": 8221
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.330858381453729e-06,
+      "loss": 0.2486,
+      "step": 8222
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.327457872720143e-06,
+      "loss": 0.2202,
+      "step": 8223
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.324058033369375e-06,
+      "loss": 0.2199,
+      "step": 8224
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.320658863663839e-06,
+      "loss": 0.2572,
+      "step": 8225
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.317260363865898e-06,
+      "loss": 0.2039,
+      "step": 8226
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.3138625342378575e-06,
+      "loss": 0.24,
+      "step": 8227
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.310465375041985e-06,
+      "loss": 0.2087,
+      "step": 8228
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.307068886540483e-06,
+      "loss": 0.2682,
+      "step": 8229
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.303673068995502e-06,
+      "loss": 0.2484,
+      "step": 8230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.300277922669154e-06,
+      "loss": 0.2747,
+      "step": 8231
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.296883447823486e-06,
+      "loss": 0.2466,
+      "step": 8232
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.293489644720499e-06,
+      "loss": 0.2174,
+      "step": 8233
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2900965136221394e-06,
+      "loss": 0.2372,
+      "step": 8234
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.286704054790299e-06,
+      "loss": 0.2727,
+      "step": 8235
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.283312268486831e-06,
+      "loss": 0.2162,
+      "step": 8236
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.279921154973522e-06,
+      "loss": 0.2124,
+      "step": 8237
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.276530714512109e-06,
+      "loss": 0.2289,
+      "step": 8238
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.27314094736429e-06,
+      "loss": 0.2052,
+      "step": 8239
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.269751853791696e-06,
+      "loss": 0.2493,
+      "step": 8240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.266363434055909e-06,
+      "loss": 0.2094,
+      "step": 8241
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2629756884184655e-06,
+      "loss": 0.2564,
+      "step": 8242
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.259588617140838e-06,
+      "loss": 0.2543,
+      "step": 8243
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.256202220484465e-06,
+      "loss": 0.2439,
+      "step": 8244
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.252816498710719e-06,
+      "loss": 0.2701,
+      "step": 8245
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.249431452080917e-06,
+      "loss": 0.2342,
+      "step": 8246
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.246047080856343e-06,
+      "loss": 0.2266,
+      "step": 8247
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.2426633852982115e-06,
+      "loss": 0.216,
+      "step": 8248
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.239280365667688e-06,
+      "loss": 0.226,
+      "step": 8249
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.235898022225889e-06,
+      "loss": 0.2161,
+      "step": 8250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.23251635523388e-06,
+      "loss": 0.2286,
+      "step": 8251
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.229135364952666e-06,
+      "loss": 0.2244,
+      "step": 8252
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.225755051643214e-06,
+      "loss": 0.2682,
+      "step": 8253
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.222375415566421e-06,
+      "loss": 0.2421,
+      "step": 8254
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.2189964569831545e-06,
+      "loss": 0.2359,
+      "step": 8255
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.215618176154209e-06,
+      "loss": 0.2254,
+      "step": 8256
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.2122405733403346e-06,
+      "loss": 0.2424,
+      "step": 8257
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.2088636488022285e-06,
+      "loss": 0.2232,
+      "step": 8258
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.205487402800536e-06,
+      "loss": 0.2486,
+      "step": 8259
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.202111835595844e-06,
+      "loss": 0.229,
+      "step": 8260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1987369474487055e-06,
+      "loss": 0.2257,
+      "step": 8261
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1953627386196005e-06,
+      "loss": 0.2331,
+      "step": 8262
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1919892093689585e-06,
+      "loss": 0.2846,
+      "step": 8263
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.188616359957178e-06,
+      "loss": 0.262,
+      "step": 8264
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.185244190644579e-06,
+      "loss": 0.2526,
+      "step": 8265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1818727016914415e-06,
+      "loss": 0.2207,
+      "step": 8266
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.178501893357991e-06,
+      "loss": 0.2215,
+      "step": 8267
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.175131765904395e-06,
+      "loss": 0.2323,
+      "step": 8268
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.171762319590786e-06,
+      "loss": 0.2154,
+      "step": 8269
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.168393554677224e-06,
+      "loss": 0.2492,
+      "step": 8270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.165025471423721e-06,
+      "loss": 0.2345,
+      "step": 8271
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.16165807009025e-06,
+      "loss": 0.2391,
+      "step": 8272
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.158291350936713e-06,
+      "loss": 0.2713,
+      "step": 8273
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.154925314222972e-06,
+      "loss": 0.2096,
+      "step": 8274
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1515599602088285e-06,
+      "loss": 0.1969,
+      "step": 8275
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.148195289154029e-06,
+      "loss": 0.2494,
+      "step": 8276
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.144831301318285e-06,
+      "loss": 0.2455,
+      "step": 8277
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.1414679969612375e-06,
+      "loss": 0.2422,
+      "step": 8278
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.138105376342474e-06,
+      "loss": 0.2697,
+      "step": 8279
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.134743439721549e-06,
+      "loss": 0.2473,
+      "step": 8280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.131382187357941e-06,
+      "loss": 0.2178,
+      "step": 8281
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.128021619511088e-06,
+      "loss": 0.2008,
+      "step": 8282
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.124661736440371e-06,
+      "loss": 0.2339,
+      "step": 8283
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.121302538405117e-06,
+      "loss": 0.2274,
+      "step": 8284
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.117944025664613e-06,
+      "loss": 0.2492,
+      "step": 8285
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.114586198478075e-06,
+      "loss": 0.2227,
+      "step": 8286
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.11122905710467e-06,
+      "loss": 0.2784,
+      "step": 8287
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.107872601803528e-06,
+      "loss": 0.2366,
+      "step": 8288
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.104516832833707e-06,
+      "loss": 0.2147,
+      "step": 8289
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.101161750454224e-06,
+      "loss": 0.2866,
+      "step": 8290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.09780735492403e-06,
+      "loss": 0.2452,
+      "step": 8291
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.094453646502039e-06,
+      "loss": 0.2433,
+      "step": 8292
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.091100625447094e-06,
+      "loss": 0.2525,
+      "step": 8293
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.087748292018007e-06,
+      "loss": 0.2406,
+      "step": 8294
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.084396646473514e-06,
+      "loss": 0.2449,
+      "step": 8295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.0810456890723204e-06,
+      "loss": 0.207,
+      "step": 8296
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.07769542007306e-06,
+      "loss": 0.2396,
+      "step": 8297
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.074345839734325e-06,
+      "loss": 0.2401,
+      "step": 8298
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.070996948314645e-06,
+      "loss": 0.232,
+      "step": 8299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.067648746072504e-06,
+      "loss": 0.2319,
+      "step": 8300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.064301233266323e-06,
+      "loss": 0.2373,
+      "step": 8301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.060954410154488e-06,
+      "loss": 0.2545,
+      "step": 8302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.057608276995312e-06,
+      "loss": 0.2247,
+      "step": 8303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.054262834047073e-06,
+      "loss": 0.2199,
+      "step": 8304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.050918081567979e-06,
+      "loss": 0.2379,
+      "step": 8305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.047574019816195e-06,
+      "loss": 0.212,
+      "step": 8306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.044230649049827e-06,
+      "loss": 0.23,
+      "step": 8307
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.040887969526931e-06,
+      "loss": 0.2282,
+      "step": 8308
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.037545981505504e-06,
+      "loss": 0.2361,
+      "step": 8309
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0342046852435064e-06,
+      "loss": 0.2559,
+      "step": 8310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.030864080998825e-06,
+      "loss": 0.2274,
+      "step": 8311
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.027524169029297e-06,
+      "loss": 0.2553,
+      "step": 8312
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.024184949592723e-06,
+      "loss": 0.2513,
+      "step": 8313
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.020846422946834e-06,
+      "loss": 0.2234,
+      "step": 8314
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0175085893493055e-06,
+      "loss": 0.2233,
+      "step": 8315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.01417144905777e-06,
+      "loss": 0.2597,
+      "step": 8316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0108350023297955e-06,
+      "loss": 0.2225,
+      "step": 8317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.007499249422914e-06,
+      "loss": 0.239,
+      "step": 8318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.004164190594586e-06,
+      "loss": 0.2259,
+      "step": 8319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.0008298261022206e-06,
+      "loss": 0.2135,
+      "step": 8320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.997496156203189e-06,
+      "loss": 0.2388,
+      "step": 8321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.994163181154791e-06,
+      "loss": 0.2201,
+      "step": 8322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.990830901214281e-06,
+      "loss": 0.215,
+      "step": 8323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.987499316638857e-06,
+      "loss": 0.2543,
+      "step": 8324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.984168427685661e-06,
+      "loss": 0.1865,
+      "step": 8325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.980838234611793e-06,
+      "loss": 0.2461,
+      "step": 8326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.977508737674289e-06,
+      "loss": 0.2237,
+      "step": 8327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9741799371301245e-06,
+      "loss": 0.2425,
+      "step": 8328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9708518332362425e-06,
+      "loss": 0.259,
+      "step": 8329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.967524426249513e-06,
+      "loss": 0.2585,
+      "step": 8330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.964197716426763e-06,
+      "loss": 0.2617,
+      "step": 8331
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9608717040247594e-06,
+      "loss": 0.2627,
+      "step": 8332
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.957546389300216e-06,
+      "loss": 0.211,
+      "step": 8333
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.954221772509794e-06,
+      "loss": 0.2473,
+      "step": 8334
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9508978539101045e-06,
+      "loss": 0.2235,
+      "step": 8335
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.947574633757697e-06,
+      "loss": 0.2394,
+      "step": 8336
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.94425211230908e-06,
+      "loss": 0.2399,
+      "step": 8337
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.940930289820691e-06,
+      "loss": 0.2723,
+      "step": 8338
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.937609166548928e-06,
+      "loss": 0.2112,
+      "step": 8339
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.934288742750126e-06,
+      "loss": 0.2368,
+      "step": 8340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.930969018680568e-06,
+      "loss": 0.2105,
+      "step": 8341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.92764999459648e-06,
+      "loss": 0.2444,
+      "step": 8342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.924331670754048e-06,
+      "loss": 0.231,
+      "step": 8343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.921014047409387e-06,
+      "loss": 0.2438,
+      "step": 8344
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9176971248185695e-06,
+      "loss": 0.2806,
+      "step": 8345
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9143809032376075e-06,
+      "loss": 0.2408,
+      "step": 8346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.911065382922461e-06,
+      "loss": 0.2386,
+      "step": 8347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.9077505641290355e-06,
+      "loss": 0.2462,
+      "step": 8348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.904436447113183e-06,
+      "loss": 0.2031,
+      "step": 8349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.901123032130693e-06,
+      "loss": 0.2393,
+      "step": 8350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.897810319437322e-06,
+      "loss": 0.2534,
+      "step": 8351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.894498309288751e-06,
+      "loss": 0.2351,
+      "step": 8352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.89118700194061e-06,
+      "loss": 0.2444,
+      "step": 8353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8878763976484925e-06,
+      "loss": 0.2398,
+      "step": 8354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.884566496667917e-06,
+      "loss": 0.2281,
+      "step": 8355
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.881257299254358e-06,
+      "loss": 0.2428,
+      "step": 8356
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.877948805663229e-06,
+      "loss": 0.2663,
+      "step": 8357
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.874641016149891e-06,
+      "loss": 0.2526,
+      "step": 8358
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.871333930969662e-06,
+      "loss": 0.2431,
+      "step": 8359
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8680275503777925e-06,
+      "loss": 0.2176,
+      "step": 8360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.864721874629478e-06,
+      "loss": 0.2513,
+      "step": 8361
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.861416903979873e-06,
+      "loss": 0.2312,
+      "step": 8362
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.858112638684065e-06,
+      "loss": 0.2841,
+      "step": 8363
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.854809078997088e-06,
+      "loss": 0.1984,
+      "step": 8364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.851506225173928e-06,
+      "loss": 0.2462,
+      "step": 8365
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8482040774695074e-06,
+      "loss": 0.2362,
+      "step": 8366
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.844902636138708e-06,
+      "loss": 0.2613,
+      "step": 8367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.841601901436345e-06,
+      "loss": 0.2472,
+      "step": 8368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.838301873617179e-06,
+      "loss": 0.2664,
+      "step": 8369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8350025529359265e-06,
+      "loss": 0.2257,
+      "step": 8370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.83170393964724e-06,
+      "loss": 0.2692,
+      "step": 8371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8284060340057205e-06,
+      "loss": 0.2301,
+      "step": 8372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.825108836265913e-06,
+      "loss": 0.2577,
+      "step": 8373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.821812346682303e-06,
+      "loss": 0.2269,
+      "step": 8374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.81851656550934e-06,
+      "loss": 0.2467,
+      "step": 8375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.815221493001399e-06,
+      "loss": 0.2553,
+      "step": 8376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.811927129412804e-06,
+      "loss": 0.2723,
+      "step": 8377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.808633474997836e-06,
+      "loss": 0.245,
+      "step": 8378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.805340530010709e-06,
+      "loss": 0.2385,
+      "step": 8379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.80204829470558e-06,
+      "loss": 0.2206,
+      "step": 8380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.798756769336574e-06,
+      "loss": 0.2386,
+      "step": 8381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7954659541577275e-06,
+      "loss": 0.2629,
+      "step": 8382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7921758494230415e-06,
+      "loss": 0.2484,
+      "step": 8383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.788886455386469e-06,
+      "loss": 0.2653,
+      "step": 8384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7855977723018886e-06,
+      "loss": 0.2362,
+      "step": 8385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.782309800423146e-06,
+      "loss": 0.2868,
+      "step": 8386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.779022540004015e-06,
+      "loss": 0.2856,
+      "step": 8387
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7757359912982134e-06,
+      "loss": 0.2396,
+      "step": 8388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.772450154559427e-06,
+      "loss": 0.2434,
+      "step": 8389
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.769165030041255e-06,
+      "loss": 0.2287,
+      "step": 8390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.765880617997256e-06,
+      "loss": 0.1931,
+      "step": 8391
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.762596918680947e-06,
+      "loss": 0.247,
+      "step": 8392
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.759313932345771e-06,
+      "loss": 0.2909,
+      "step": 8393
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.756031659245118e-06,
+      "loss": 0.2121,
+      "step": 8394
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.752750099632337e-06,
+      "loss": 0.2565,
+      "step": 8395
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.749469253760702e-06,
+      "loss": 0.2547,
+      "step": 8396
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.746189121883459e-06,
+      "loss": 0.2233,
+      "step": 8397
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.742909704253765e-06,
+      "loss": 0.2406,
+      "step": 8398
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.739631001124742e-06,
+      "loss": 0.2026,
+      "step": 8399
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.73635301274946e-06,
+      "loss": 0.2321,
+      "step": 8400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.733075739380927e-06,
+      "loss": 0.2364,
+      "step": 8401
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7297991812720865e-06,
+      "loss": 0.2453,
+      "step": 8402
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7265233386758526e-06,
+      "loss": 0.2158,
+      "step": 8403
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.723248211845054e-06,
+      "loss": 0.2439,
+      "step": 8404
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.719973801032491e-06,
+      "loss": 0.2124,
+      "step": 8405
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.716700106490895e-06,
+      "loss": 0.2632,
+      "step": 8406
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.713427128472929e-06,
+      "loss": 0.2245,
+      "step": 8407
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.71015486723123e-06,
+      "loss": 0.2062,
+      "step": 8408
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7068833230183596e-06,
+      "loss": 0.219,
+      "step": 8409
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.703612496086824e-06,
+      "loss": 0.2343,
+      "step": 8410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.700342386689089e-06,
+      "loss": 0.2797,
+      "step": 8411
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.697072995077551e-06,
+      "loss": 0.2112,
+      "step": 8412
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.693804321504549e-06,
+      "loss": 0.2293,
+      "step": 8413
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.690536366222389e-06,
+      "loss": 0.2535,
+      "step": 8414
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6872691294832855e-06,
+      "loss": 0.2384,
+      "step": 8415
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.684002611539432e-06,
+      "loss": 0.2298,
+      "step": 8416
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.680736812642944e-06,
+      "loss": 0.3054,
+      "step": 8417
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.677471733045891e-06,
+      "loss": 0.233,
+      "step": 8418
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.674207373000288e-06,
+      "loss": 0.2279,
+      "step": 8419
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.670943732758093e-06,
+      "loss": 0.2335,
+      "step": 8420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6676808125711994e-06,
+      "loss": 0.2502,
+      "step": 8421
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6644186126914615e-06,
+      "loss": 0.2119,
+      "step": 8422
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.661157133370673e-06,
+      "loss": 0.2438,
+      "step": 8423
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.657896374860552e-06,
+      "loss": 0.2398,
+      "step": 8424
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.654636337412791e-06,
+      "loss": 0.2602,
+      "step": 8425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6513770212790035e-06,
+      "loss": 0.2439,
+      "step": 8426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.648118426710769e-06,
+      "loss": 0.2179,
+      "step": 8427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.644860553959593e-06,
+      "loss": 0.229,
+      "step": 8428
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.641603403276926e-06,
+      "loss": 0.2232,
+      "step": 8429
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.638346974914181e-06,
+      "loss": 0.2663,
+      "step": 8430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6350912691227e-06,
+      "loss": 0.2515,
+      "step": 8431
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6318362861537565e-06,
+      "loss": 0.2183,
+      "step": 8432
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.628582026258602e-06,
+      "loss": 0.2143,
+      "step": 8433
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6253284896884e-06,
+      "loss": 0.2581,
+      "step": 8434
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.622075676694285e-06,
+      "loss": 0.2604,
+      "step": 8435
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.618823587527314e-06,
+      "loss": 0.2087,
+      "step": 8436
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.615572222438496e-06,
+      "loss": 0.2008,
+      "step": 8437
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.612321581678793e-06,
+      "loss": 0.2341,
+      "step": 8438
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.609071665499102e-06,
+      "loss": 0.2169,
+      "step": 8439
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.6058224741502515e-06,
+      "loss": 0.2337,
+      "step": 8440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.602574007883041e-06,
+      "loss": 0.2356,
+      "step": 8441
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.599326266948197e-06,
+      "loss": 0.2122,
+      "step": 8442
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.596079251596389e-06,
+      "loss": 0.2606,
+      "step": 8443
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.592832962078243e-06,
+      "loss": 0.2563,
+      "step": 8444
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.589587398644313e-06,
+      "loss": 0.2487,
+      "step": 8445
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.586342561545115e-06,
+      "loss": 0.2006,
+      "step": 8446
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.583098451031092e-06,
+      "loss": 0.2196,
+      "step": 8447
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.57985506735264e-06,
+      "loss": 0.2293,
+      "step": 8448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.576612410760098e-06,
+      "loss": 0.245,
+      "step": 8449
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.573370481503745e-06,
+      "loss": 0.2031,
+      "step": 8450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.570129279833801e-06,
+      "loss": 0.2422,
+      "step": 8451
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.566888806000448e-06,
+      "loss": 0.2421,
+      "step": 8452
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.563649060253794e-06,
+      "loss": 0.2106,
+      "step": 8453
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.560410042843888e-06,
+      "loss": 0.2052,
+      "step": 8454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.557171754020745e-06,
+      "loss": 0.2223,
+      "step": 8455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.553934194034301e-06,
+      "loss": 0.2173,
+      "step": 8456
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.550697363134447e-06,
+      "loss": 0.2543,
+      "step": 8457
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.547461261571014e-06,
+      "loss": 0.2263,
+      "step": 8458
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.544225889593772e-06,
+      "loss": 0.2331,
+      "step": 8459
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.540991247452451e-06,
+      "loss": 0.2526,
+      "step": 8460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.53775733539671e-06,
+      "loss": 0.2783,
+      "step": 8461
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.534524153676149e-06,
+      "loss": 0.2116,
+      "step": 8462
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.531291702540331e-06,
+      "loss": 0.31,
+      "step": 8463
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.528059982238743e-06,
+      "loss": 0.2853,
+      "step": 8464
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.524828993020822e-06,
+      "loss": 0.2215,
+      "step": 8465
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5215987351359504e-06,
+      "loss": 0.2639,
+      "step": 8466
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.518369208833448e-06,
+      "loss": 0.2232,
+      "step": 8467
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.515140414362593e-06,
+      "loss": 0.2491,
+      "step": 8468
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.511912351972591e-06,
+      "loss": 0.2268,
+      "step": 8469
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.508685021912596e-06,
+      "loss": 0.275,
+      "step": 8470
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.50545842443171e-06,
+      "loss": 0.2263,
+      "step": 8471
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.502232559778976e-06,
+      "loss": 0.1987,
+      "step": 8472
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.499007428203378e-06,
+      "loss": 0.2017,
+      "step": 8473
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.495783029953845e-06,
+      "loss": 0.2301,
+      "step": 8474
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4925593652792435e-06,
+      "loss": 0.2159,
+      "step": 8475
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4893364344283995e-06,
+      "loss": 0.2332,
+      "step": 8476
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.48611423765007e-06,
+      "loss": 0.2617,
+      "step": 8477
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4828927751929475e-06,
+      "loss": 0.2342,
+      "step": 8478
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.4796720473056935e-06,
+      "loss": 0.2345,
+      "step": 8479
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.476452054236887e-06,
+      "loss": 0.2922,
+      "step": 8480
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.473232796235062e-06,
+      "loss": 0.2562,
+      "step": 8481
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.470014273548697e-06,
+      "loss": 0.2312,
+      "step": 8482
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.466796486426209e-06,
+      "loss": 0.2519,
+      "step": 8483
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.463579435115954e-06,
+      "loss": 0.2098,
+      "step": 8484
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.460363119866249e-06,
+      "loss": 0.2266,
+      "step": 8485
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.457147540925331e-06,
+      "loss": 0.3314,
+      "step": 8486
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.453932698541404e-06,
+      "loss": 0.2774,
+      "step": 8487
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.450718592962595e-06,
+      "loss": 0.2304,
+      "step": 8488
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.447505224436986e-06,
+      "loss": 0.2014,
+      "step": 8489
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.444292593212594e-06,
+      "loss": 0.246,
+      "step": 8490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.441080699537386e-06,
+      "loss": 0.2111,
+      "step": 8491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.437869543659262e-06,
+      "loss": 0.2812,
+      "step": 8492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4346591258260855e-06,
+      "loss": 0.2965,
+      "step": 8493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.431449446285635e-06,
+      "loss": 0.1967,
+      "step": 8494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.428240505285662e-06,
+      "loss": 0.2228,
+      "step": 8495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.42503230307384e-06,
+      "loss": 0.2747,
+      "step": 8496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.421824839897788e-06,
+      "loss": 0.2374,
+      "step": 8497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.418618116005075e-06,
+      "loss": 0.2481,
+      "step": 8498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.415412131643206e-06,
+      "loss": 0.2402,
+      "step": 8499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.412206887059631e-06,
+      "loss": 0.2292,
+      "step": 8500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.40900238250175e-06,
+      "loss": 0.2471,
+      "step": 8501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.405798618216899e-06,
+      "loss": 0.2076,
+      "step": 8502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.402595594452348e-06,
+      "loss": 0.2138,
+      "step": 8503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.399393311455335e-06,
+      "loss": 0.2076,
+      "step": 8504
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.396191769473018e-06,
+      "loss": 0.2468,
+      "step": 8505
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.392990968752504e-06,
+      "loss": 0.2358,
+      "step": 8506
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.389790909540846e-06,
+      "loss": 0.2541,
+      "step": 8507
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.386591592085033e-06,
+      "loss": 0.2499,
+      "step": 8508
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.38339301663201e-06,
+      "loss": 0.2468,
+      "step": 8509
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3801951834286535e-06,
+      "loss": 0.2137,
+      "step": 8510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.376998092721777e-06,
+      "loss": 0.2222,
+      "step": 8511
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.373801744758159e-06,
+      "loss": 0.2219,
+      "step": 8512
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.370606139784501e-06,
+      "loss": 0.2973,
+      "step": 8513
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.367411278047452e-06,
+      "loss": 0.235,
+      "step": 8514
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.364217159793605e-06,
+      "loss": 0.2586,
+      "step": 8515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.36102378526949e-06,
+      "loss": 0.1913,
+      "step": 8516
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.357831154721596e-06,
+      "loss": 0.2308,
+      "step": 8517
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.354639268396337e-06,
+      "loss": 0.2467,
+      "step": 8518
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.351448126540073e-06,
+      "loss": 0.233,
+      "step": 8519
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.348257729399118e-06,
+      "loss": 0.2657,
+      "step": 8520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.345068077219716e-06,
+      "loss": 0.2363,
+      "step": 8521
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.341879170248057e-06,
+      "loss": 0.2513,
+      "step": 8522
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.338691008730274e-06,
+      "loss": 0.2289,
+      "step": 8523
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.335503592912443e-06,
+      "loss": 0.2083,
+      "step": 8524
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.3323169230405765e-06,
+      "loss": 0.2253,
+      "step": 8525
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.329130999360647e-06,
+      "loss": 0.256,
+      "step": 8526
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.325945822118544e-06,
+      "loss": 0.2905,
+      "step": 8527
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.322761391560124e-06,
+      "loss": 0.2586,
+      "step": 8528
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.31957770793117e-06,
+      "loss": 0.2489,
+      "step": 8529
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.316394771477412e-06,
+      "loss": 0.2478,
+      "step": 8530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.313212582444522e-06,
+      "loss": 0.2293,
+      "step": 8531
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.310031141078116e-06,
+      "loss": 0.2108,
+      "step": 8532
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.306850447623743e-06,
+      "loss": 0.2492,
+      "step": 8533
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.303670502326913e-06,
+      "loss": 0.1936,
+      "step": 8534
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.30049130543306e-06,
+      "loss": 0.2371,
+      "step": 8535
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.297312857187574e-06,
+      "loss": 0.2272,
+      "step": 8536
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.294135157835779e-06,
+      "loss": 0.2288,
+      "step": 8537
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.29095820762294e-06,
+      "loss": 0.2569,
+      "step": 8538
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.287782006794268e-06,
+      "loss": 0.2008,
+      "step": 8539
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.2846065555949165e-06,
+      "loss": 0.2326,
+      "step": 8540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.281431854269976e-06,
+      "loss": 0.2205,
+      "step": 8541
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.27825790306449e-06,
+      "loss": 0.2333,
+      "step": 8542
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.275084702223436e-06,
+      "loss": 0.201,
+      "step": 8543
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.271912251991725e-06,
+      "loss": 0.2081,
+      "step": 8544
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.268740552614234e-06,
+      "loss": 0.235,
+      "step": 8545
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.265569604335762e-06,
+      "loss": 0.2371,
+      "step": 8546
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.262399407401055e-06,
+      "loss": 0.2506,
+      "step": 8547
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.259229962054803e-06,
+      "loss": 0.2143,
+      "step": 8548
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.256061268541631e-06,
+      "loss": 0.2065,
+      "step": 8549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2528933271061215e-06,
+      "loss": 0.2398,
+      "step": 8550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.249726137992786e-06,
+      "loss": 0.241,
+      "step": 8551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.246559701446077e-06,
+      "loss": 0.23,
+      "step": 8552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.243394017710399e-06,
+      "loss": 0.2034,
+      "step": 8553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.240229087030092e-06,
+      "loss": 0.2417,
+      "step": 8554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.237064909649437e-06,
+      "loss": 0.1911,
+      "step": 8555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.233901485812659e-06,
+      "loss": 0.2587,
+      "step": 8556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.230738815763917e-06,
+      "loss": 0.2575,
+      "step": 8557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.22757689974733e-06,
+      "loss": 0.235,
+      "step": 8558
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2244157380069444e-06,
+      "loss": 0.2319,
+      "step": 8559
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.221255330786745e-06,
+      "loss": 0.283,
+      "step": 8560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.2180956783306764e-06,
+      "loss": 0.2602,
+      "step": 8561
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.214936780882607e-06,
+      "loss": 0.3123,
+      "step": 8562
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.211778638686356e-06,
+      "loss": 0.2265,
+      "step": 8563
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.208621251985679e-06,
+      "loss": 0.215,
+      "step": 8564
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.205464621024274e-06,
+      "loss": 0.2595,
+      "step": 8565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.202308746045793e-06,
+      "loss": 0.2259,
+      "step": 8566
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.199153627293812e-06,
+      "loss": 0.244,
+      "step": 8567
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.195999265011852e-06,
+      "loss": 0.2598,
+      "step": 8568
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.192845659443391e-06,
+      "loss": 0.2378,
+      "step": 8569
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.18969281083183e-06,
+      "loss": 0.2503,
+      "step": 8570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1865407194205225e-06,
+      "loss": 0.2538,
+      "step": 8571
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.183389385452757e-06,
+      "loss": 0.2251,
+      "step": 8572
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.180238809171769e-06,
+      "loss": 0.2269,
+      "step": 8573
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.177088990820725e-06,
+      "loss": 0.2423,
+      "step": 8574
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1739399306427535e-06,
+      "loss": 0.256,
+      "step": 8575
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.170791628880901e-06,
+      "loss": 0.2046,
+      "step": 8576
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.167644085778176e-06,
+      "loss": 0.2392,
+      "step": 8577
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.164497301577515e-06,
+      "loss": 0.2296,
+      "step": 8578
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.161351276521802e-06,
+      "loss": 0.2231,
+      "step": 8579
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.158206010853856e-06,
+      "loss": 0.2584,
+      "step": 8580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.155061504816443e-06,
+      "loss": 0.2304,
+      "step": 8581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.151917758652265e-06,
+      "loss": 0.2411,
+      "step": 8582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1487747726039795e-06,
+      "loss": 0.2307,
+      "step": 8583
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.145632546914169e-06,
+      "loss": 0.2164,
+      "step": 8584
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1424910818253595e-06,
+      "loss": 0.2269,
+      "step": 8585
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.139350377580033e-06,
+      "loss": 0.2745,
+      "step": 8586
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.136210434420595e-06,
+      "loss": 0.2387,
+      "step": 8587
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1330712525894e-06,
+      "loss": 0.2353,
+      "step": 8588
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.129932832328745e-06,
+      "loss": 0.271,
+      "step": 8589
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.126795173880859e-06,
+      "loss": 0.2364,
+      "step": 8590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.12365827748793e-06,
+      "loss": 0.208,
+      "step": 8591
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.120522143392072e-06,
+      "loss": 0.2297,
+      "step": 8592
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.117386771835342e-06,
+      "loss": 0.2619,
+      "step": 8593
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.114252163059748e-06,
+      "loss": 0.2674,
+      "step": 8594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.111118317307227e-06,
+      "loss": 0.2305,
+      "step": 8595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.107985234819664e-06,
+      "loss": 0.2573,
+      "step": 8596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1048529158388815e-06,
+      "loss": 0.2143,
+      "step": 8597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.101721360606643e-06,
+      "loss": 0.2277,
+      "step": 8598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.098590569364661e-06,
+      "loss": 0.2321,
+      "step": 8599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.095460542354581e-06,
+      "loss": 0.2587,
+      "step": 8600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.092331279817985e-06,
+      "loss": 0.213,
+      "step": 8601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.089202781996414e-06,
+      "loss": 0.2335,
+      "step": 8602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.086075049131331e-06,
+      "loss": 0.2567,
+      "step": 8603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.082948081464149e-06,
+      "loss": 0.232,
+      "step": 8604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.07982187923622e-06,
+      "loss": 0.231,
+      "step": 8605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.076696442688833e-06,
+      "loss": 0.2671,
+      "step": 8606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0735717720632315e-06,
+      "loss": 0.2065,
+      "step": 8607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.070447867600585e-06,
+      "loss": 0.2477,
+      "step": 8608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0673247295420045e-06,
+      "loss": 0.259,
+      "step": 8609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.064202358128559e-06,
+      "loss": 0.2327,
+      "step": 8610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.061080753601238e-06,
+      "loss": 0.2194,
+      "step": 8611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.057959916200981e-06,
+      "loss": 0.2401,
+      "step": 8612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.054839846168668e-06,
+      "loss": 0.2113,
+      "step": 8613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.051720543745118e-06,
+      "loss": 0.2599,
+      "step": 8614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.048602009171089e-06,
+      "loss": 0.2019,
+      "step": 8615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.04548424268729e-06,
+      "loss": 0.2385,
+      "step": 8616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.042367244534352e-06,
+      "loss": 0.2252,
+      "step": 8617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0392510149528694e-06,
+      "loss": 0.2539,
+      "step": 8618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.036135554183363e-06,
+      "loss": 0.2336,
+      "step": 8619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.033020862466293e-06,
+      "loss": 0.2931,
+      "step": 8620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.029906940042068e-06,
+      "loss": 0.2444,
+      "step": 8621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.026793787151031e-06,
+      "loss": 0.2372,
+      "step": 8622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.023681404033464e-06,
+      "loss": 0.265,
+      "step": 8623
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.020569790929602e-06,
+      "loss": 0.2372,
+      "step": 8624
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.017458948079605e-06,
+      "loss": 0.2135,
+      "step": 8625
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.01434887572359e-06,
+      "loss": 0.2443,
+      "step": 8626
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.011239574101598e-06,
+      "loss": 0.2423,
+      "step": 8627
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.00813104345362e-06,
+      "loss": 0.2306,
+      "step": 8628
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.005023284019586e-06,
+      "loss": 0.2219,
+      "step": 8629
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.001916296039363e-06,
+      "loss": 0.2531,
+      "step": 8630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.998810079752758e-06,
+      "loss": 0.2493,
+      "step": 8631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.995704635399531e-06,
+      "loss": 0.2072,
+      "step": 8632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.99259996321937e-06,
+      "loss": 0.2506,
+      "step": 8633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9894960634519e-06,
+      "loss": 0.2359,
+      "step": 8634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.986392936336702e-06,
+      "loss": 0.225,
+      "step": 8635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.983290582113286e-06,
+      "loss": 0.2347,
+      "step": 8636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9801890010211e-06,
+      "loss": 0.2203,
+      "step": 8637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.977088193299541e-06,
+      "loss": 0.2285,
+      "step": 8638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.973988159187936e-06,
+      "loss": 0.2353,
+      "step": 8639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9708888989255685e-06,
+      "loss": 0.2522,
+      "step": 8640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.967790412751647e-06,
+      "loss": 0.1902,
+      "step": 8641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.964692700905321e-06,
+      "loss": 0.2513,
+      "step": 8642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.961595763625694e-06,
+      "loss": 0.2122,
+      "step": 8643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.958499601151797e-06,
+      "loss": 0.238,
+      "step": 8644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.955404213722602e-06,
+      "loss": 0.2616,
+      "step": 8645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.952309601577026e-06,
+      "loss": 0.2388,
+      "step": 8646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.94921576495392e-06,
+      "loss": 0.2063,
+      "step": 8647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.946122704092087e-06,
+      "loss": 0.23,
+      "step": 8648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.943030419230257e-06,
+      "loss": 0.2203,
+      "step": 8649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.939938910607101e-06,
+      "loss": 0.265,
+      "step": 8650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.936848178461247e-06,
+      "loss": 0.2402,
+      "step": 8651
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.933758223031241e-06,
+      "loss": 0.2182,
+      "step": 8652
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.930669044555582e-06,
+      "loss": 0.2524,
+      "step": 8653
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.927580643272705e-06,
+      "loss": 0.2554,
+      "step": 8654
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.924493019420979e-06,
+      "loss": 0.2537,
+      "step": 8655
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.921406173238732e-06,
+      "loss": 0.2359,
+      "step": 8656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.918320104964213e-06,
+      "loss": 0.2275,
+      "step": 8657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.915234814835613e-06,
+      "loss": 0.2018,
+      "step": 8658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.912150303091077e-06,
+      "loss": 0.2341,
+      "step": 8659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.9090665699686765e-06,
+      "loss": 0.2321,
+      "step": 8660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.905983615706423e-06,
+      "loss": 0.2075,
+      "step": 8661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.902901440542283e-06,
+      "loss": 0.2147,
+      "step": 8662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.899820044714139e-06,
+      "loss": 0.2145,
+      "step": 8663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.896739428459824e-06,
+      "loss": 0.2251,
+      "step": 8664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8936595920171255e-06,
+      "loss": 0.2381,
+      "step": 8665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.890580535623748e-06,
+      "loss": 0.2401,
+      "step": 8666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.887502259517353e-06,
+      "loss": 0.1994,
+      "step": 8667
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.884424763935531e-06,
+      "loss": 0.2317,
+      "step": 8668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.881348049115813e-06,
+      "loss": 0.2378,
+      "step": 8669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.878272115295686e-06,
+      "loss": 0.1752,
+      "step": 8670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.875196962712546e-06,
+      "loss": 0.2235,
+      "step": 8671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.87212259160375e-06,
+      "loss": 0.2052,
+      "step": 8672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.869049002206599e-06,
+      "loss": 0.2105,
+      "step": 8673
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.865976194758319e-06,
+      "loss": 0.216,
+      "step": 8674
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.862904169496078e-06,
+      "loss": 0.2422,
+      "step": 8675
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.859832926657001e-06,
+      "loss": 0.2197,
+      "step": 8676
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8567624664781245e-06,
+      "loss": 0.2545,
+      "step": 8677
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8536927891964555e-06,
+      "loss": 0.2599,
+      "step": 8678
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.850623895048911e-06,
+      "loss": 0.2651,
+      "step": 8679
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8475557842723606e-06,
+      "loss": 0.2167,
+      "step": 8680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8444884571036225e-06,
+      "loss": 0.2615,
+      "step": 8681
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8414219137794415e-06,
+      "loss": 0.2485,
+      "step": 8682
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.838356154536503e-06,
+      "loss": 0.2069,
+      "step": 8683
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.835291179611441e-06,
+      "loss": 0.236,
+      "step": 8684
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.832226989240821e-06,
+      "loss": 0.2383,
+      "step": 8685
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.829163583661145e-06,
+      "loss": 0.1939,
+      "step": 8686
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.826100963108873e-06,
+      "loss": 0.2518,
+      "step": 8687
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.823039127820372e-06,
+      "loss": 0.2545,
+      "step": 8688
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.819978078031981e-06,
+      "loss": 0.2716,
+      "step": 8689
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.81691781397996e-06,
+      "loss": 0.1855,
+      "step": 8690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.813858335900509e-06,
+      "loss": 0.1977,
+      "step": 8691
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.810799644029779e-06,
+      "loss": 0.2497,
+      "step": 8692
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.807741738603848e-06,
+      "loss": 0.2226,
+      "step": 8693
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.804684619858735e-06,
+      "loss": 0.2517,
+      "step": 8694
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.8016282880304144e-06,
+      "loss": 0.2438,
+      "step": 8695
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.798572743354768e-06,
+      "loss": 0.2383,
+      "step": 8696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7955179860676465e-06,
+      "loss": 0.2348,
+      "step": 8697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7924640164048286e-06,
+      "loss": 0.2285,
+      "step": 8698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.789410834602026e-06,
+      "loss": 0.226,
+      "step": 8699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.786358440894903e-06,
+      "loss": 0.2163,
+      "step": 8700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.783306835519054e-06,
+      "loss": 0.2209,
+      "step": 8701
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.78025601871001e-06,
+      "loss": 0.2385,
+      "step": 8702
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.777205990703253e-06,
+      "loss": 0.1952,
+      "step": 8703
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7741567517342e-06,
+      "loss": 0.2448,
+      "step": 8704
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7711083020381866e-06,
+      "loss": 0.2439,
+      "step": 8705
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.768060641850521e-06,
+      "loss": 0.251,
+      "step": 8706
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7650137714064236e-06,
+      "loss": 0.2361,
+      "step": 8707
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.761967690941074e-06,
+      "loss": 0.2564,
+      "step": 8708
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.758922400689578e-06,
+      "loss": 0.2173,
+      "step": 8709
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.755877900886978e-06,
+      "loss": 0.2339,
+      "step": 8710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7528341917682696e-06,
+      "loss": 0.2405,
+      "step": 8711
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.749791273568381e-06,
+      "loss": 0.2518,
+      "step": 8712
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.746749146522162e-06,
+      "loss": 0.2667,
+      "step": 8713
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.74370781086443e-06,
+      "loss": 0.2633,
+      "step": 8714
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.740667266829926e-06,
+      "loss": 0.2331,
+      "step": 8715
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.737627514653322e-06,
+      "loss": 0.2166,
+      "step": 8716
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.734588554569254e-06,
+      "loss": 0.2595,
+      "step": 8717
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.73155038681227e-06,
+      "loss": 0.2505,
+      "step": 8718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.728513011616874e-06,
+      "loss": 0.2094,
+      "step": 8719
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.72547642921751e-06,
+      "loss": 0.2231,
+      "step": 8720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7224406398485335e-06,
+      "loss": 0.2396,
+      "step": 8721
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.719405643744278e-06,
+      "loss": 0.2256,
+      "step": 8722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.71637144113899e-06,
+      "loss": 0.2256,
+      "step": 8723
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.713338032266859e-06,
+      "loss": 0.253,
+      "step": 8724
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.710305417362024e-06,
+      "loss": 0.2364,
+      "step": 8725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.707273596658544e-06,
+      "loss": 0.2098,
+      "step": 8726
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7042425703904405e-06,
+      "loss": 0.187,
+      "step": 8727
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.7012123387916536e-06,
+      "loss": 0.2468,
+      "step": 8728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.698182902096069e-06,
+      "loss": 0.2686,
+      "step": 8729
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6951542605375145e-06,
+      "loss": 0.2692,
+      "step": 8730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.692126414349749e-06,
+      "loss": 0.2528,
+      "step": 8731
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.689099363766472e-06,
+      "loss": 0.2456,
+      "step": 8732
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6860731090213305e-06,
+      "loss": 0.2287,
+      "step": 8733
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.683047650347903e-06,
+      "loss": 0.2554,
+      "step": 8734
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.680022987979699e-06,
+      "loss": 0.2699,
+      "step": 8735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.676999122150185e-06,
+      "loss": 0.2153,
+      "step": 8736
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.673976053092749e-06,
+      "loss": 0.2181,
+      "step": 8737
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6709537810407285e-06,
+      "loss": 0.2371,
+      "step": 8738
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.667932306227392e-06,
+      "loss": 0.2367,
+      "step": 8739
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6649116288859425e-06,
+      "loss": 0.2797,
+      "step": 8740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6618917492495425e-06,
+      "loss": 0.2259,
+      "step": 8741
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6588726675512715e-06,
+      "loss": 0.2119,
+      "step": 8742
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.65585438402415e-06,
+      "loss": 0.2462,
+      "step": 8743
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.652836898901152e-06,
+      "loss": 0.2476,
+      "step": 8744
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.649820212415175e-06,
+      "loss": 0.2327,
+      "step": 8745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.646804324799057e-06,
+      "loss": 0.2228,
+      "step": 8746
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.64378923628558e-06,
+      "loss": 0.255,
+      "step": 8747
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.640774947107455e-06,
+      "loss": 0.2257,
+      "step": 8748
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6377614574973465e-06,
+      "loss": 0.2313,
+      "step": 8749
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6347487676878435e-06,
+      "loss": 0.2089,
+      "step": 8750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.631736877911473e-06,
+      "loss": 0.2594,
+      "step": 8751
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.6287257884007145e-06,
+      "loss": 0.2232,
+      "step": 8752
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.625715499387973e-06,
+      "loss": 0.2255,
+      "step": 8753
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.622706011105595e-06,
+      "loss": 0.2567,
+      "step": 8754
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.619697323785863e-06,
+      "loss": 0.2121,
+      "step": 8755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.616689437660997e-06,
+      "loss": 0.2268,
+      "step": 8756
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.613682352963168e-06,
+      "loss": 0.2242,
+      "step": 8757
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.610676069924469e-06,
+      "loss": 0.2138,
+      "step": 8758
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.607670588776933e-06,
+      "loss": 0.2347,
+      "step": 8759
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.604665909752547e-06,
+      "loss": 0.2598,
+      "step": 8760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.601662033083219e-06,
+      "loss": 0.2532,
+      "step": 8761
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.598658959000799e-06,
+      "loss": 0.2319,
+      "step": 8762
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.595656687737077e-06,
+      "loss": 0.2631,
+      "step": 8763
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.592655219523783e-06,
+      "loss": 0.2237,
+      "step": 8764
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.589654554592575e-06,
+      "loss": 0.2619,
+      "step": 8765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.586654693175069e-06,
+      "loss": 0.2125,
+      "step": 8766
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.583655635502797e-06,
+      "loss": 0.2396,
+      "step": 8767
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.5806573818072465e-06,
+      "loss": 0.2596,
+      "step": 8768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.577659932319831e-06,
+      "loss": 0.2072,
+      "step": 8769
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.574663287271906e-06,
+      "loss": 0.2392,
+      "step": 8770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.571667446894768e-06,
+      "loss": 0.2662,
+      "step": 8771
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.568672411419643e-06,
+      "loss": 0.2341,
+      "step": 8772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.565678181077699e-06,
+      "loss": 0.2197,
+      "step": 8773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.562684756100053e-06,
+      "loss": 0.2218,
+      "step": 8774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5596921367177435e-06,
+      "loss": 0.2293,
+      "step": 8775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.55670032316175e-06,
+      "loss": 0.1962,
+      "step": 8776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.553709315663001e-06,
+      "loss": 0.2632,
+      "step": 8777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.550719114452351e-06,
+      "loss": 0.2724,
+      "step": 8778
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.547729719760596e-06,
+      "loss": 0.2444,
+      "step": 8779
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.54474113181847e-06,
+      "loss": 0.2287,
+      "step": 8780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5417533508566415e-06,
+      "loss": 0.2123,
+      "step": 8781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.538766377105727e-06,
+      "loss": 0.1919,
+      "step": 8782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.535780210796268e-06,
+      "loss": 0.2636,
+      "step": 8783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.532794852158748e-06,
+      "loss": 0.2267,
+      "step": 8784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5298103014235955e-06,
+      "loss": 0.2276,
+      "step": 8785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.5268265588211675e-06,
+      "loss": 0.265,
+      "step": 8786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.523843624581762e-06,
+      "loss": 0.2236,
+      "step": 8787
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.520861498935613e-06,
+      "loss": 0.2587,
+      "step": 8788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.517880182112888e-06,
+      "loss": 0.2857,
+      "step": 8789
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.51489967434371e-06,
+      "loss": 0.2325,
+      "step": 8790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.51191997585812e-06,
+      "loss": 0.2296,
+      "step": 8791
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.508941086886098e-06,
+      "loss": 0.2288,
+      "step": 8792
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.505963007657578e-06,
+      "loss": 0.2288,
+      "step": 8793
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.502985738402416e-06,
+      "loss": 0.2377,
+      "step": 8794
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.500009279350409e-06,
+      "loss": 0.25,
+      "step": 8795
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.497033630731292e-06,
+      "loss": 0.2086,
+      "step": 8796
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.494058792774735e-06,
+      "loss": 0.2374,
+      "step": 8797
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.491084765710356e-06,
+      "loss": 0.2642,
+      "step": 8798
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.488111549767699e-06,
+      "loss": 0.1906,
+      "step": 8799
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4851391451762435e-06,
+      "loss": 0.2759,
+      "step": 8800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.482167552165423e-06,
+      "loss": 0.2617,
+      "step": 8801
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.479196770964594e-06,
+      "loss": 0.2035,
+      "step": 8802
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.476226801803048e-06,
+      "loss": 0.2065,
+      "step": 8803
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.473257644910025e-06,
+      "loss": 0.2224,
+      "step": 8804
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.470289300514695e-06,
+      "loss": 0.2338,
+      "step": 8805
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.467321768846164e-06,
+      "loss": 0.2194,
+      "step": 8806
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.464355050133483e-06,
+      "loss": 0.2493,
+      "step": 8807
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4613891446056326e-06,
+      "loss": 0.2369,
+      "step": 8808
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.458424052491538e-06,
+      "loss": 0.2457,
+      "step": 8809
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.455459774020057e-06,
+      "loss": 0.2712,
+      "step": 8810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.452496309419981e-06,
+      "loss": 0.1823,
+      "step": 8811
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.449533658920044e-06,
+      "loss": 0.2184,
+      "step": 8812
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4465718227489164e-06,
+      "loss": 0.2162,
+      "step": 8813
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.443610801135199e-06,
+      "loss": 0.2017,
+      "step": 8814
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4406505943074465e-06,
+      "loss": 0.2273,
+      "step": 8815
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.43769120249413e-06,
+      "loss": 0.2456,
+      "step": 8816
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.434732625923677e-06,
+      "loss": 0.2235,
+      "step": 8817
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4317748648244374e-06,
+      "loss": 0.2377,
+      "step": 8818
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.428817919424705e-06,
+      "loss": 0.2368,
+      "step": 8819
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.425861789952707e-06,
+      "loss": 0.2563,
+      "step": 8820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.422906476636612e-06,
+      "loss": 0.2423,
+      "step": 8821
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4199519797045175e-06,
+      "loss": 0.2564,
+      "step": 8822
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4169982993844734e-06,
+      "loss": 0.2691,
+      "step": 8823
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.414045435904453e-06,
+      "loss": 0.2076,
+      "step": 8824
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.411093389492363e-06,
+      "loss": 0.2678,
+      "step": 8825
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.408142160376069e-06,
+      "loss": 0.2164,
+      "step": 8826
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4051917487833495e-06,
+      "loss": 0.2431,
+      "step": 8827
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.402242154941931e-06,
+      "loss": 0.2316,
+      "step": 8828
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.399293379079479e-06,
+      "loss": 0.2216,
+      "step": 8829
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.396345421423582e-06,
+      "loss": 0.2615,
+      "step": 8830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.393398282201788e-06,
+      "loss": 0.2469,
+      "step": 8831
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.390451961641565e-06,
+      "loss": 0.2208,
+      "step": 8832
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.387506459970317e-06,
+      "loss": 0.2081,
+      "step": 8833
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.384561777415399e-06,
+      "loss": 0.2409,
+      "step": 8834
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.381617914204088e-06,
+      "loss": 0.2226,
+      "step": 8835
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.378674870563607e-06,
+      "loss": 0.2333,
+      "step": 8836
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.375732646721109e-06,
+      "loss": 0.2116,
+      "step": 8837
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.372791242903685e-06,
+      "loss": 0.2861,
+      "step": 8838
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.369850659338373e-06,
+      "loss": 0.2607,
+      "step": 8839
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.366910896252134e-06,
+      "loss": 0.2341,
+      "step": 8840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.363971953871866e-06,
+      "loss": 0.2197,
+      "step": 8841
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.361033832424421e-06,
+      "loss": 0.2072,
+      "step": 8842
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.358096532136569e-06,
+      "loss": 0.241,
+      "step": 8843
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.355160053235022e-06,
+      "loss": 0.2272,
+      "step": 8844
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3522243959464315e-06,
+      "loss": 0.2404,
+      "step": 8845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.349289560497382e-06,
+      "loss": 0.204,
+      "step": 8846
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.346355547114393e-06,
+      "loss": 0.2362,
+      "step": 8847
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.343422356023933e-06,
+      "loss": 0.2506,
+      "step": 8848
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.340489987452387e-06,
+      "loss": 0.2866,
+      "step": 8849
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.337558441626099e-06,
+      "loss": 0.1936,
+      "step": 8850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.334627718771331e-06,
+      "loss": 0.2318,
+      "step": 8851
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.33169781911429e-06,
+      "loss": 0.2359,
+      "step": 8852
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.328768742881119e-06,
+      "loss": 0.2654,
+      "step": 8853
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.325840490297892e-06,
+      "loss": 0.2469,
+      "step": 8854
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.322913061590622e-06,
+      "loss": 0.2015,
+      "step": 8855
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3199864569852706e-06,
+      "loss": 0.2252,
+      "step": 8856
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3170606767077145e-06,
+      "loss": 0.236,
+      "step": 8857
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.314135720983785e-06,
+      "loss": 0.2128,
+      "step": 8858
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.31121159003924e-06,
+      "loss": 0.2111,
+      "step": 8859
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.3082882840997756e-06,
+      "loss": 0.2428,
+      "step": 8860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.305365803391025e-06,
+      "loss": 0.2253,
+      "step": 8861
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.302444148138557e-06,
+      "loss": 0.2609,
+      "step": 8862
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.299523318567873e-06,
+      "loss": 0.2819,
+      "step": 8863
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.296603314904423e-06,
+      "loss": 0.2592,
+      "step": 8864
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.293684137373581e-06,
+      "loss": 0.2214,
+      "step": 8865
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.290765786200657e-06,
+      "loss": 0.2793,
+      "step": 8866
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2878482616109084e-06,
+      "loss": 0.2418,
+      "step": 8867
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.28493156382952e-06,
+      "loss": 0.2304,
+      "step": 8868
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.282015693081616e-06,
+      "loss": 0.2509,
+      "step": 8869
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.279100649592251e-06,
+      "loss": 0.2022,
+      "step": 8870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.276186433586418e-06,
+      "loss": 0.202,
+      "step": 8871
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2732730452890565e-06,
+      "loss": 0.1956,
+      "step": 8872
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.27036048492503e-06,
+      "loss": 0.2151,
+      "step": 8873
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.267448752719139e-06,
+      "loss": 0.1919,
+      "step": 8874
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.264537848896129e-06,
+      "loss": 0.1995,
+      "step": 8875
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.261627773680673e-06,
+      "loss": 0.2296,
+      "step": 8876
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2587185272973835e-06,
+      "loss": 0.2614,
+      "step": 8877
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.255810109970807e-06,
+      "loss": 0.1914,
+      "step": 8878
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2529025219254205e-06,
+      "loss": 0.2124,
+      "step": 8879
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2499957633856574e-06,
+      "loss": 0.2076,
+      "step": 8880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.247089834575867e-06,
+      "loss": 0.2215,
+      "step": 8881
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.244184735720335e-06,
+      "loss": 0.2144,
+      "step": 8882
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241280467043299e-06,
+      "loss": 0.2103,
+      "step": 8883
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.238377028768918e-06,
+      "loss": 0.2234,
+      "step": 8884
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2354744211212915e-06,
+      "loss": 0.2505,
+      "step": 8885
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.232572644324456e-06,
+      "loss": 0.1992,
+      "step": 8886
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.229671698602375e-06,
+      "loss": 0.1968,
+      "step": 8887
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.226771584178968e-06,
+      "loss": 0.2557,
+      "step": 8888
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.223872301278071e-06,
+      "loss": 0.234,
+      "step": 8889
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.220973850123459e-06,
+      "loss": 0.2392,
+      "step": 8890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.218076230938856e-06,
+      "loss": 0.1904,
+      "step": 8891
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.215179443947908e-06,
+      "loss": 0.2549,
+      "step": 8892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2122834893742e-06,
+      "loss": 0.2098,
+      "step": 8893
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.209388367441254e-06,
+      "loss": 0.2474,
+      "step": 8894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.206494078372528e-06,
+      "loss": 0.2132,
+      "step": 8895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.203600622391411e-06,
+      "loss": 0.212,
+      "step": 8896
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2007079997212386e-06,
+      "loss": 0.223,
+      "step": 8897
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.19781621058527e-06,
+      "loss": 0.2554,
+      "step": 8898
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.194925255206712e-06,
+      "loss": 0.2066,
+      "step": 8899
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.192035133808697e-06,
+      "loss": 0.2306,
+      "step": 8900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.189145846614295e-06,
+      "loss": 0.2407,
+      "step": 8901
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.186257393846517e-06,
+      "loss": 0.2234,
+      "step": 8902
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.183369775728303e-06,
+      "loss": 0.23,
+      "step": 8903
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.180482992482527e-06,
+      "loss": 0.2599,
+      "step": 8904
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.177597044332012e-06,
+      "loss": 0.2084,
+      "step": 8905
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1747119314995015e-06,
+      "loss": 0.2815,
+      "step": 8906
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.171827654207679e-06,
+      "loss": 0.2347,
+      "step": 8907
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.168944212679173e-06,
+      "loss": 0.2166,
+      "step": 8908
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.166061607136533e-06,
+      "loss": 0.2376,
+      "step": 8909
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.163179837802254e-06,
+      "loss": 0.2333,
+      "step": 8910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1602989048987605e-06,
+      "loss": 0.2094,
+      "step": 8911
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.15741880864841e-06,
+      "loss": 0.2377,
+      "step": 8912
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.15453954927351e-06,
+      "loss": 0.2245,
+      "step": 8913
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.15166112699629e-06,
+      "loss": 0.2548,
+      "step": 8914
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1487835420389125e-06,
+      "loss": 0.2721,
+      "step": 8915
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.145906794623493e-06,
+      "loss": 0.2138,
+      "step": 8916
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.143030884972063e-06,
+      "loss": 0.2593,
+      "step": 8917
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.140155813306601e-06,
+      "loss": 0.2071,
+      "step": 8918
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.137281579849013e-06,
+      "loss": 0.273,
+      "step": 8919
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1344081848211415e-06,
+      "loss": 0.2242,
+      "step": 8920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.131535628444777e-06,
+      "loss": 0.2513,
+      "step": 8921
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.128663910941629e-06,
+      "loss": 0.2297,
+      "step": 8922
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.125793032533347e-06,
+      "loss": 0.2275,
+      "step": 8923
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1229229934415225e-06,
+      "loss": 0.2054,
+      "step": 8924
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1200537938876754e-06,
+      "loss": 0.2437,
+      "step": 8925
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.1171854340932605e-06,
+      "loss": 0.2288,
+      "step": 8926
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.114317914279672e-06,
+      "loss": 0.2413,
+      "step": 8927
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.11145123466823e-06,
+      "loss": 0.223,
+      "step": 8928
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.108585395480208e-06,
+      "loss": 0.253,
+      "step": 8929
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.105720396936797e-06,
+      "loss": 0.2304,
+      "step": 8930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.102856239259125e-06,
+      "loss": 0.2093,
+      "step": 8931
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.099992922668269e-06,
+      "loss": 0.2547,
+      "step": 8932
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.097130447385229e-06,
+      "loss": 0.2195,
+      "step": 8933
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.094268813630942e-06,
+      "loss": 0.242,
+      "step": 8934
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.091408021626278e-06,
+      "loss": 0.2063,
+      "step": 8935
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.088548071592048e-06,
+      "loss": 0.2381,
+      "step": 8936
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0856889637489885e-06,
+      "loss": 0.2301,
+      "step": 8937
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.082830698317788e-06,
+      "loss": 0.2454,
+      "step": 8938
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.079973275519049e-06,
+      "loss": 0.2489,
+      "step": 8939
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0771166955733265e-06,
+      "loss": 0.2376,
+      "step": 8940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.074260958701103e-06,
+      "loss": 0.2349,
+      "step": 8941
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.071406065122788e-06,
+      "loss": 0.2407,
+      "step": 8942
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06855201505875e-06,
+      "loss": 0.2317,
+      "step": 8943
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065698808729261e-06,
+      "loss": 0.2524,
+      "step": 8944
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062846446354545e-06,
+      "loss": 0.2522,
+      "step": 8945
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.059994928154767e-06,
+      "loss": 0.2377,
+      "step": 8946
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0571442543500096e-06,
+      "loss": 0.2059,
+      "step": 8947
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.05429442516031e-06,
+      "loss": 0.2524,
+      "step": 8948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0514454408056244e-06,
+      "loss": 0.232,
+      "step": 8949
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.048597301505846e-06,
+      "loss": 0.2649,
+      "step": 8950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0457500074808165e-06,
+      "loss": 0.2591,
+      "step": 8951
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.042903558950293e-06,
+      "loss": 0.2479,
+      "step": 8952
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.040057956133972e-06,
+      "loss": 0.2405,
+      "step": 8953
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0372131992514995e-06,
+      "loss": 0.2417,
+      "step": 8954
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.034369288522443e-06,
+      "loss": 0.2759,
+      "step": 8955
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.031526224166301e-06,
+      "loss": 0.247,
+      "step": 8956
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.028684006402521e-06,
+      "loss": 0.1978,
+      "step": 8957
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0258426354504685e-06,
+      "loss": 0.2127,
+      "step": 8958
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.023002111529468e-06,
+      "loss": 0.2257,
+      "step": 8959
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.020162434858747e-06,
+      "loss": 0.2308,
+      "step": 8960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.017323605657486e-06,
+      "loss": 0.238,
+      "step": 8961
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.014485624144804e-06,
+      "loss": 0.2194,
+      "step": 8962
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.011648490539746e-06,
+      "loss": 0.2079,
+      "step": 8963
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.008812205061287e-06,
+      "loss": 0.2136,
+      "step": 8964
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.005976767928355e-06,
+      "loss": 0.2547,
+      "step": 8965
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.003142179359793e-06,
+      "loss": 0.2571,
+      "step": 8966
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.000308439574385e-06,
+      "loss": 0.2505,
+      "step": 8967
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.997475548790862e-06,
+      "loss": 0.2395,
+      "step": 8968
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.994643507227862e-06,
+      "loss": 0.2,
+      "step": 8969
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9918123151039855e-06,
+      "loss": 0.2265,
+      "step": 8970
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.988981972637753e-06,
+      "loss": 0.2814,
+      "step": 8971
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.986152480047616e-06,
+      "loss": 0.2258,
+      "step": 8972
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.983323837551977e-06,
+      "loss": 0.2209,
+      "step": 8973
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.980496045369155e-06,
+      "loss": 0.2321,
+      "step": 8974
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.977669103717411e-06,
+      "loss": 0.203,
+      "step": 8975
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.974843012814948e-06,
+      "loss": 0.2327,
+      "step": 8976
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.972017772879883e-06,
+      "loss": 0.2629,
+      "step": 8977
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.969193384130289e-06,
+      "loss": 0.1994,
+      "step": 8978
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.966369846784161e-06,
+      "loss": 0.2236,
+      "step": 8979
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9635471610594275e-06,
+      "loss": 0.2444,
+      "step": 8980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.960725327173963e-06,
+      "loss": 0.2477,
+      "step": 8981
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.957904345345565e-06,
+      "loss": 0.2304,
+      "step": 8982
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9550842157919644e-06,
+      "loss": 0.2496,
+      "step": 8983
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.952264938730837e-06,
+      "loss": 0.24,
+      "step": 8984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.949446514379789e-06,
+      "loss": 0.2427,
+      "step": 8985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.946628942956342e-06,
+      "loss": 0.2634,
+      "step": 8986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.943812224677984e-06,
+      "loss": 0.2298,
+      "step": 8987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.940996359762111e-06,
+      "loss": 0.2251,
+      "step": 8988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.938181348426072e-06,
+      "loss": 0.23,
+      "step": 8989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.935367190887136e-06,
+      "loss": 0.2754,
+      "step": 8990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9325538873625085e-06,
+      "loss": 0.2249,
+      "step": 8991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.929741438069341e-06,
+      "loss": 0.2657,
+      "step": 8992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.926929843224708e-06,
+      "loss": 0.2164,
+      "step": 8993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9241191030456066e-06,
+      "loss": 0.2354,
+      "step": 8994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.921309217748997e-06,
+      "loss": 0.253,
+      "step": 8995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.918500187551754e-06,
+      "loss": 0.2326,
+      "step": 8996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.915692012670682e-06,
+      "loss": 0.2442,
+      "step": 8997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.91288469332254e-06,
+      "loss": 0.2341,
+      "step": 8998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.910078229723996e-06,
+      "loss": 0.2459,
+      "step": 8999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.907272622091678e-06,
+      "loss": 0.2278,
+      "step": 9000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9044678706421325e-06,
+      "loss": 0.2327,
+      "step": 9001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.901663975591827e-06,
+      "loss": 0.2569,
+      "step": 9002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.898860937157191e-06,
+      "loss": 0.2616,
+      "step": 9003
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.896058755554572e-06,
+      "loss": 0.2356,
+      "step": 9004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8932574310002515e-06,
+      "loss": 0.246,
+      "step": 9005
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.890456963710451e-06,
+      "loss": 0.2174,
+      "step": 9006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8876573539013225e-06,
+      "loss": 0.1809,
+      "step": 9007
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.884858601788943e-06,
+      "loss": 0.1857,
+      "step": 9008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.882060707589345e-06,
+      "loss": 0.2246,
+      "step": 9009
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.879263671518473e-06,
+      "loss": 0.2426,
+      "step": 9010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.876467493792215e-06,
+      "loss": 0.2224,
+      "step": 9011
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.873672174626396e-06,
+      "loss": 0.2722,
+      "step": 9012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.87087771423676e-06,
+      "loss": 0.2163,
+      "step": 9013
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.868084112839006e-06,
+      "loss": 0.2136,
+      "step": 9014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.865291370648753e-06,
+      "loss": 0.2697,
+      "step": 9015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.86249948788155e-06,
+      "loss": 0.2177,
+      "step": 9016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.859708464752894e-06,
+      "loss": 0.225,
+      "step": 9017
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.856918301478207e-06,
+      "loss": 0.2332,
+      "step": 9018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8541289982728425e-06,
+      "loss": 0.2538,
+      "step": 9019
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.851340555352091e-06,
+      "loss": 0.2382,
+      "step": 9020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.848552972931172e-06,
+      "loss": 0.2559,
+      "step": 9021
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.845766251225252e-06,
+      "loss": 0.2235,
+      "step": 9022
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.842980390449417e-06,
+      "loss": 0.2391,
+      "step": 9023
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.840195390818687e-06,
+      "loss": 0.2319,
+      "step": 9024
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8374112525480265e-06,
+      "loss": 0.2491,
+      "step": 9025
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.834627975852324e-06,
+      "loss": 0.2464,
+      "step": 9026
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.831845560946407e-06,
+      "loss": 0.2049,
+      "step": 9027
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.829064008045031e-06,
+      "loss": 0.218,
+      "step": 9028
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.826283317362884e-06,
+      "loss": 0.2417,
+      "step": 9029
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8235034891145995e-06,
+      "loss": 0.2249,
+      "step": 9030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.820724523514731e-06,
+      "loss": 0.2173,
+      "step": 9031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.817946420777768e-06,
+      "loss": 0.2595,
+      "step": 9032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.815169181118146e-06,
+      "loss": 0.2078,
+      "step": 9033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8123928047502165e-06,
+      "loss": 0.2092,
+      "step": 9034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8096172918882738e-06,
+      "loss": 0.2019,
+      "step": 9035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.806842642746543e-06,
+      "loss": 0.2378,
+      "step": 9036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8040688575391815e-06,
+      "loss": 0.2168,
+      "step": 9037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.801295936480279e-06,
+      "loss": 0.2139,
+      "step": 9038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7985238797838685e-06,
+      "loss": 0.2171,
+      "step": 9039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7957526876639016e-06,
+      "loss": 0.2391,
+      "step": 9040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.792982360334279e-06,
+      "loss": 0.2096,
+      "step": 9041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7902128980088194e-06,
+      "loss": 0.266,
+      "step": 9042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7874443009012843e-06,
+      "loss": 0.2954,
+      "step": 9043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7846765692253654e-06,
+      "loss": 0.2207,
+      "step": 9044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7819097031946853e-06,
+      "loss": 0.2178,
+      "step": 9045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7791437030227998e-06,
+      "loss": 0.2315,
+      "step": 9046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7763785689232085e-06,
+      "loss": 0.245,
+      "step": 9047
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7736143011093264e-06,
+      "loss": 0.2297,
+      "step": 9048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7708508997945208e-06,
+      "loss": 0.2248,
+      "step": 9049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7680883651920783e-06,
+      "loss": 0.2165,
+      "step": 9050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.765326697515222e-06,
+      "loss": 0.246,
+      "step": 9051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7625658969771103e-06,
+      "loss": 0.2277,
+      "step": 9052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.759805963790832e-06,
+      "loss": 0.217,
+      "step": 9053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7570468981694056e-06,
+      "loss": 0.211,
+      "step": 9054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.754288700325797e-06,
+      "loss": 0.229,
+      "step": 9055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.751531370472891e-06,
+      "loss": 0.2384,
+      "step": 9056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.748774908823504e-06,
+      "loss": 0.2212,
+      "step": 9057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.746019315590401e-06,
+      "loss": 0.2526,
+      "step": 9058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.743264590986265e-06,
+      "loss": 0.2733,
+      "step": 9059
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7405107352237176e-06,
+      "loss": 0.214,
+      "step": 9060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.737757748515314e-06,
+      "loss": 0.2656,
+      "step": 9061
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.735005631073536e-06,
+      "loss": 0.2494,
+      "step": 9062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7322543831108114e-06,
+      "loss": 0.2252,
+      "step": 9063
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7295040048394873e-06,
+      "loss": 0.2369,
+      "step": 9064
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.726754496471849e-06,
+      "loss": 0.2307,
+      "step": 9065
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7240058582201203e-06,
+      "loss": 0.2464,
+      "step": 9066
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7212580902964493e-06,
+      "loss": 0.24,
+      "step": 9067
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7185111929129203e-06,
+      "loss": 0.2225,
+      "step": 9068
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7157651662815496e-06,
+      "loss": 0.2244,
+      "step": 9069
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.713020010614283e-06,
+      "loss": 0.2242,
+      "step": 9070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7102757261230113e-06,
+      "loss": 0.2236,
+      "step": 9071
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7075323130195464e-06,
+      "loss": 0.2454,
+      "step": 9072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7047897715156325e-06,
+      "loss": 0.237,
+      "step": 9073
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.7020481018229567e-06,
+      "loss": 0.2149,
+      "step": 9074
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.69930730415313e-06,
+      "loss": 0.2157,
+      "step": 9075
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6965673787176977e-06,
+      "loss": 0.2314,
+      "step": 9076
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.693828325728138e-06,
+      "loss": 0.2266,
+      "step": 9077
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6910901453958594e-06,
+      "loss": 0.2194,
+      "step": 9078
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6883528379322153e-06,
+      "loss": 0.2247,
+      "step": 9079
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6856164035484757e-06,
+      "loss": 0.2429,
+      "step": 9080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6828808424558474e-06,
+      "loss": 0.2177,
+      "step": 9081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6801461548654797e-06,
+      "loss": 0.2246,
+      "step": 9082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.677412340988445e-06,
+      "loss": 0.2513,
+      "step": 9083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.674679401035749e-06,
+      "loss": 0.266,
+      "step": 9084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6719473352183298e-06,
+      "loss": 0.2238,
+      "step": 9085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.669216143747063e-06,
+      "loss": 0.2344,
+      "step": 9086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.666485826832746e-06,
+      "loss": 0.2664,
+      "step": 9087
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6637563846861278e-06,
+      "loss": 0.2272,
+      "step": 9088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6610278175178655e-06,
+      "loss": 0.2208,
+      "step": 9089
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6583001255385733e-06,
+      "loss": 0.1953,
+      "step": 9090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.655573308958779e-06,
+      "loss": 0.2262,
+      "step": 9091
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6528473679889508e-06,
+      "loss": 0.252,
+      "step": 9092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6501223028394886e-06,
+      "loss": 0.225,
+      "step": 9093
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.647398113720723e-06,
+      "loss": 0.228,
+      "step": 9094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.644674800842915e-06,
+      "loss": 0.2179,
+      "step": 9095
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6419523644162693e-06,
+      "loss": 0.1962,
+      "step": 9096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6392308046509103e-06,
+      "loss": 0.2098,
+      "step": 9097
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.636510121756895e-06,
+      "loss": 0.2448,
+      "step": 9098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.633790315944228e-06,
+      "loss": 0.2573,
+      "step": 9099
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6310713874228265e-06,
+      "loss": 0.25,
+      "step": 9100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6283533364025532e-06,
+      "loss": 0.2107,
+      "step": 9101
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6256361630931944e-06,
+      "loss": 0.2551,
+      "step": 9102
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6229198677044716e-06,
+      "loss": 0.2109,
+      "step": 9103
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6202044504460475e-06,
+      "loss": 0.2168,
+      "step": 9104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6174899115275038e-06,
+      "loss": 0.2315,
+      "step": 9105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.614776251158357e-06,
+      "loss": 0.2225,
+      "step": 9106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6120634695480676e-06,
+      "loss": 0.2643,
+      "step": 9107
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6093515669060127e-06,
+      "loss": 0.2302,
+      "step": 9108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6066405434415122e-06,
+      "loss": 0.2157,
+      "step": 9109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.60393039936381e-06,
+      "loss": 0.2199,
+      "step": 9110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.6012211348820837e-06,
+      "loss": 0.2327,
+      "step": 9111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.598512750205455e-06,
+      "loss": 0.2089,
+      "step": 9112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5958052455429623e-06,
+      "loss": 0.2237,
+      "step": 9113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5930986211035794e-06,
+      "loss": 0.2529,
+      "step": 9114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.590392877096222e-06,
+      "loss": 0.2367,
+      "step": 9115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5876880137297278e-06,
+      "loss": 0.2429,
+      "step": 9116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.584984031212867e-06,
+      "loss": 0.1921,
+      "step": 9117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.582280929754348e-06,
+      "loss": 0.21,
+      "step": 9118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5795787095627998e-06,
+      "loss": 0.2035,
+      "step": 9119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5768773708468007e-06,
+      "loss": 0.2331,
+      "step": 9120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5741769138148474e-06,
+      "loss": 0.2264,
+      "step": 9121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.5714773386753662e-06,
+      "loss": 0.2085,
+      "step": 9122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.568778645636733e-06,
+      "loss": 0.2125,
+      "step": 9123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5660808349072364e-06,
+      "loss": 0.2376,
+      "step": 9124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5633839066951094e-06,
+      "loss": 0.2315,
+      "step": 9125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.560687861208507e-06,
+      "loss": 0.2131,
+      "step": 9126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5579926986555256e-06,
+      "loss": 0.2232,
+      "step": 9127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5552984192441805e-06,
+      "loss": 0.2228,
+      "step": 9128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5526050231824392e-06,
+      "loss": 0.2483,
+      "step": 9129
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.54991251067818e-06,
+      "loss": 0.2544,
+      "step": 9130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.54722088193923e-06,
+      "loss": 0.2149,
+      "step": 9131
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5445301371733365e-06,
+      "loss": 0.2092,
+      "step": 9132
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5418402765881815e-06,
+      "loss": 0.1916,
+      "step": 9133
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.539151300391379e-06,
+      "loss": 0.213,
+      "step": 9134
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.536463208790478e-06,
+      "loss": 0.2431,
+      "step": 9135
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.533776001992951e-06,
+      "loss": 0.1937,
+      "step": 9136
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.531089680206217e-06,
+      "loss": 0.1782,
+      "step": 9137
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.528404243637607e-06,
+      "loss": 0.2135,
+      "step": 9138
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5257196924944045e-06,
+      "loss": 0.219,
+      "step": 9139
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5230360269838097e-06,
+      "loss": 0.2167,
+      "step": 9140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.520353247312958e-06,
+      "loss": 0.2486,
+      "step": 9141
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.517671353688919e-06,
+      "loss": 0.1965,
+      "step": 9142
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.51499034631869e-06,
+      "loss": 0.219,
+      "step": 9143
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.512310225409201e-06,
+      "loss": 0.2363,
+      "step": 9144
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5096309911673213e-06,
+      "loss": 0.213,
+      "step": 9145
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.506952643799842e-06,
+      "loss": 0.2435,
+      "step": 9146
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5042751835134853e-06,
+      "loss": 0.2283,
+      "step": 9147
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.5015986105149163e-06,
+      "loss": 0.2336,
+      "step": 9148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4989229250107185e-06,
+      "loss": 0.2313,
+      "step": 9149
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4962481272074154e-06,
+      "loss": 0.2276,
+      "step": 9150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4935742173114564e-06,
+      "loss": 0.2145,
+      "step": 9151
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4909011955292237e-06,
+      "loss": 0.2201,
+      "step": 9152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4882290620670364e-06,
+      "loss": 0.2169,
+      "step": 9153
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4855578171311405e-06,
+      "loss": 0.2285,
+      "step": 9154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4828874609277086e-06,
+      "loss": 0.203,
+      "step": 9155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4802179936628576e-06,
+      "loss": 0.2947,
+      "step": 9156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.477549415542623e-06,
+      "loss": 0.2475,
+      "step": 9157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.474881726772978e-06,
+      "loss": 0.233,
+      "step": 9158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.472214927559827e-06,
+      "loss": 0.2233,
+      "step": 9159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.469549018108998e-06,
+      "loss": 0.2479,
+      "step": 9160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.466883998626266e-06,
+      "loss": 0.2053,
+      "step": 9161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.464219869317326e-06,
+      "loss": 0.1947,
+      "step": 9162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.461556630387801e-06,
+      "loss": 0.2229,
+      "step": 9163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4588942820432582e-06,
+      "loss": 0.2039,
+      "step": 9164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4562328244891853e-06,
+      "loss": 0.2269,
+      "step": 9165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4535722579310064e-06,
+      "loss": 0.2444,
+      "step": 9166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.450912582574072e-06,
+      "loss": 0.2497,
+      "step": 9167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.448253798623669e-06,
+      "loss": 0.1968,
+      "step": 9168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.445595906285009e-06,
+      "loss": 0.2648,
+      "step": 9169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.442938905763246e-06,
+      "loss": 0.2381,
+      "step": 9170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4402827972634527e-06,
+      "loss": 0.1962,
+      "step": 9171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.437627580990643e-06,
+      "loss": 0.2546,
+      "step": 9172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.434973257149759e-06,
+      "loss": 0.2445,
+      "step": 9173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4323198259456674e-06,
+      "loss": 0.2077,
+      "step": 9174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4296672875831716e-06,
+      "loss": 0.2434,
+      "step": 9175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.427015642267009e-06,
+      "loss": 0.2393,
+      "step": 9176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4243648902018366e-06,
+      "loss": 0.2667,
+      "step": 9177
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.42171503159226e-06,
+      "loss": 0.2335,
+      "step": 9178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4190660666427997e-06,
+      "loss": 0.2114,
+      "step": 9179
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.41641799555792e-06,
+      "loss": 0.2388,
+      "step": 9180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4137708185420075e-06,
+      "loss": 0.239,
+      "step": 9181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.41112453579938e-06,
+      "loss": 0.2116,
+      "step": 9182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.40847914753429e-06,
+      "loss": 0.2327,
+      "step": 9183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4058346539509206e-06,
+      "loss": 0.2372,
+      "step": 9184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4031910552533795e-06,
+      "loss": 0.2527,
+      "step": 9185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.400548351645719e-06,
+      "loss": 0.2732,
+      "step": 9186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3979065433319097e-06,
+      "loss": 0.2349,
+      "step": 9187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3952656305158523e-06,
+      "loss": 0.2329,
+      "step": 9188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3926256134013932e-06,
+      "loss": 0.2296,
+      "step": 9189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.389986492192296e-06,
+      "loss": 0.2137,
+      "step": 9190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3873482670922584e-06,
+      "loss": 0.2387,
+      "step": 9191
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3847109383049086e-06,
+      "loss": 0.229,
+      "step": 9192
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3820745060338032e-06,
+      "loss": 0.2432,
+      "step": 9193
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.379438970482443e-06,
+      "loss": 0.2224,
+      "step": 9194
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3768043318542418e-06,
+      "loss": 0.2654,
+      "step": 9195
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3741705903525504e-06,
+      "loss": 0.1975,
+      "step": 9196
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3715377461806613e-06,
+      "loss": 0.2144,
+      "step": 9197
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.36890579954178e-06,
+      "loss": 0.2551,
+      "step": 9198
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.366274750639056e-06,
+      "loss": 0.2275,
+      "step": 9199
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.363644599675561e-06,
+      "loss": 0.2298,
+      "step": 9200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.361015346854297e-06,
+      "loss": 0.2454,
+      "step": 9201
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3583869923782114e-06,
+      "loss": 0.2604,
+      "step": 9202
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3557595364501666e-06,
+      "loss": 0.2221,
+      "step": 9203
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.353132979272954e-06,
+      "loss": 0.216,
+      "step": 9204
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3505073210493127e-06,
+      "loss": 0.2486,
+      "step": 9205
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3478825619818974e-06,
+      "loss": 0.2059,
+      "step": 9206
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3452587022732984e-06,
+      "loss": 0.2136,
+      "step": 9207
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3426357421260345e-06,
+      "loss": 0.247,
+      "step": 9208
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.340013681742554e-06,
+      "loss": 0.2559,
+      "step": 9209
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3373925213252453e-06,
+      "loss": 0.2272,
+      "step": 9210
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.334772261076418e-06,
+      "loss": 0.2215,
+      "step": 9211
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3321529011983105e-06,
+      "loss": 0.2309,
+      "step": 9212
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.329534441893102e-06,
+      "loss": 0.2333,
+      "step": 9213
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3269168833628954e-06,
+      "loss": 0.2176,
+      "step": 9214
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.32430022580972e-06,
+      "loss": 0.2465,
+      "step": 9215
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.321684469435544e-06,
+      "loss": 0.2017,
+      "step": 9216
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3190696144422615e-06,
+      "loss": 0.2313,
+      "step": 9217
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.316455661031694e-06,
+      "loss": 0.2272,
+      "step": 9218
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.313842609405604e-06,
+      "loss": 0.2039,
+      "step": 9219
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.311230459765671e-06,
+      "loss": 0.251,
+      "step": 9220
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3086192123135193e-06,
+      "loss": 0.2331,
+      "step": 9221
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.306008867250693e-06,
+      "loss": 0.2378,
+      "step": 9222
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3033994247786613e-06,
+      "loss": 0.2141,
+      "step": 9223
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3007908850988476e-06,
+      "loss": 0.2251,
+      "step": 9224
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2981832484125773e-06,
+      "loss": 0.2064,
+      "step": 9225
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2955765149211174e-06,
+      "loss": 0.1925,
+      "step": 9226
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.292970684825676e-06,
+      "loss": 0.2342,
+      "step": 9227
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2903657583273778e-06,
+      "loss": 0.2528,
+      "step": 9228
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.287761735627276e-06,
+      "loss": 0.2471,
+      "step": 9229
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2851586169263683e-06,
+      "loss": 0.2329,
+      "step": 9230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2825564024255665e-06,
+      "loss": 0.2512,
+      "step": 9231
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2799550923257342e-06,
+      "loss": 0.2388,
+      "step": 9232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.277354686827637e-06,
+      "loss": 0.214,
+      "step": 9233
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.274755186131986e-06,
+      "loss": 0.2725,
+      "step": 9234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2721565904394297e-06,
+      "loss": 0.2494,
+      "step": 9235
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2695588999505333e-06,
+      "loss": 0.2515,
+      "step": 9236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2669621148657954e-06,
+      "loss": 0.2146,
+      "step": 9237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.2643662353856516e-06,
+      "loss": 0.2153,
+      "step": 9238
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.261771261710458e-06,
+      "loss": 0.2397,
+      "step": 9239
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2591771940405146e-06,
+      "loss": 0.21,
+      "step": 9240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.256584032576032e-06,
+      "loss": 0.1991,
+      "step": 9241
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2539917775171617e-06,
+      "loss": 0.2435,
+      "step": 9242
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2514004290639905e-06,
+      "loss": 0.2032,
+      "step": 9243
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2488099874165276e-06,
+      "loss": 0.244,
+      "step": 9244
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.246220452774709e-06,
+      "loss": 0.2248,
+      "step": 9245
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.243631825338413e-06,
+      "loss": 0.2006,
+      "step": 9246
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2410441053074393e-06,
+      "loss": 0.2305,
+      "step": 9247
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2384572928815115e-06,
+      "loss": 0.2437,
+      "step": 9248
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2358713882603036e-06,
+      "loss": 0.2395,
+      "step": 9249
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2332863916433926e-06,
+      "loss": 0.2602,
+      "step": 9250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2307023032303085e-06,
+      "loss": 0.2165,
+      "step": 9251
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2281191232205005e-06,
+      "loss": 0.2174,
+      "step": 9252
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.225536851813342e-06,
+      "loss": 0.2297,
+      "step": 9253
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2229554892081542e-06,
+      "loss": 0.2242,
+      "step": 9254
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2203750356041724e-06,
+      "loss": 0.2163,
+      "step": 9255
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.217795491200562e-06,
+      "loss": 0.2081,
+      "step": 9256
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2152168561964357e-06,
+      "loss": 0.208,
+      "step": 9257
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2126391307908126e-06,
+      "loss": 0.2463,
+      "step": 9258
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.21006231518265e-06,
+      "loss": 0.2133,
+      "step": 9259
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2074864095708463e-06,
+      "loss": 0.2485,
+      "step": 9260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.2049114141542114e-06,
+      "loss": 0.2175,
+      "step": 9261
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.202337329131505e-06,
+      "loss": 0.2109,
+      "step": 9262
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1997641547013984e-06,
+      "loss": 0.2105,
+      "step": 9263
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1971918910624975e-06,
+      "loss": 0.2631,
+      "step": 9264
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1946205384133487e-06,
+      "loss": 0.2257,
+      "step": 9265
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.192050096952418e-06,
+      "loss": 0.2325,
+      "step": 9266
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1894805668780934e-06,
+      "loss": 0.2498,
+      "step": 9267
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1869119483887115e-06,
+      "loss": 0.1966,
+      "step": 9268
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.184344241682522e-06,
+      "loss": 0.2051,
+      "step": 9269
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1817774469577176e-06,
+      "loss": 0.2292,
+      "step": 9270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1792115644124136e-06,
+      "loss": 0.2612,
+      "step": 9271
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.176646594244648e-06,
+      "loss": 0.2244,
+      "step": 9272
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1740825366524064e-06,
+      "loss": 0.2401,
+      "step": 9273
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.171519391833591e-06,
+      "loss": 0.2214,
+      "step": 9274
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1689571599860245e-06,
+      "loss": 0.286,
+      "step": 9275
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.166395841307485e-06,
+      "loss": 0.2472,
+      "step": 9276
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1638354359956597e-06,
+      "loss": 0.2073,
+      "step": 9277
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.161275944248168e-06,
+      "loss": 0.2303,
+      "step": 9278
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.158717366262568e-06,
+      "loss": 0.221,
+      "step": 9279
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1561597022363375e-06,
+      "loss": 0.2498,
+      "step": 9280
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.153602952366891e-06,
+      "loss": 0.2389,
+      "step": 9281
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1510471168515733e-06,
+      "loss": 0.2482,
+      "step": 9282
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.148492195887638e-06,
+      "loss": 0.2069,
+      "step": 9283
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1459381896723004e-06,
+      "loss": 0.2243,
+      "step": 9284
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.143385098402685e-06,
+      "loss": 0.205,
+      "step": 9285
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.140832922275843e-06,
+      "loss": 0.2577,
+      "step": 9286
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.138281661488773e-06,
+      "loss": 0.2193,
+      "step": 9287
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.135731316238385e-06,
+      "loss": 0.2534,
+      "step": 9288
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.133181886721524e-06,
+      "loss": 0.2139,
+      "step": 9289
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.130633373134973e-06,
+      "loss": 0.2637,
+      "step": 9290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.128085775675431e-06,
+      "loss": 0.2206,
+      "step": 9291
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1255390945395347e-06,
+      "loss": 0.2482,
+      "step": 9292
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.122993329923846e-06,
+      "loss": 0.2025,
+      "step": 9293
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1204484820248538e-06,
+      "loss": 0.1746,
+      "step": 9294
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.117904551038988e-06,
+      "loss": 0.2351,
+      "step": 9295
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.115361537162596e-06,
+      "loss": 0.2436,
+      "step": 9296
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.112819440591955e-06,
+      "loss": 0.2154,
+      "step": 9297
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.1102782615232804e-06,
+      "loss": 0.2665,
+      "step": 9298
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1077380001527083e-06,
+      "loss": 0.2271,
+      "step": 9299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.105198656676307e-06,
+      "loss": 0.2862,
+      "step": 9300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1026602312900726e-06,
+      "loss": 0.2917,
+      "step": 9301
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.1001227241899273e-06,
+      "loss": 0.2375,
+      "step": 9302
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0975861355717354e-06,
+      "loss": 0.2609,
+      "step": 9303
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.095050465631278e-06,
+      "loss": 0.2237,
+      "step": 9304
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0925157145642625e-06,
+      "loss": 0.238,
+      "step": 9305
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0899818825663396e-06,
+      "loss": 0.2551,
+      "step": 9306
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0874489698330793e-06,
+      "loss": 0.2183,
+      "step": 9307
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0849169765599803e-06,
+      "loss": 0.2217,
+      "step": 9308
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0823859029424735e-06,
+      "loss": 0.2484,
+      "step": 9309
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0798557491759133e-06,
+      "loss": 0.2228,
+      "step": 9310
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.077326515455594e-06,
+      "loss": 0.2329,
+      "step": 9311
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0747982019767314e-06,
+      "loss": 0.2537,
+      "step": 9312
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.072270808934465e-06,
+      "loss": 0.2356,
+      "step": 9313
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.069744336523877e-06,
+      "loss": 0.2858,
+      "step": 9314
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0672187849399686e-06,
+      "loss": 0.1974,
+      "step": 9315
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0646941543776723e-06,
+      "loss": 0.2349,
+      "step": 9316
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.062170445031849e-06,
+      "loss": 0.2698,
+      "step": 9317
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.059647657097289e-06,
+      "loss": 0.249,
+      "step": 9318
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.057125790768708e-06,
+      "loss": 0.2014,
+      "step": 9319
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.054604846240761e-06,
+      "loss": 0.2103,
+      "step": 9320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.052084823708018e-06,
+      "loss": 0.2168,
+      "step": 9321
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0495657233649944e-06,
+      "loss": 0.2354,
+      "step": 9322
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0470475454061165e-06,
+      "loss": 0.2622,
+      "step": 9323
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.044530290025752e-06,
+      "loss": 0.2779,
+      "step": 9324
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0420139574181895e-06,
+      "loss": 0.2594,
+      "step": 9325
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0394985477776527e-06,
+      "loss": 0.2511,
+      "step": 9326
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.036984061298287e-06,
+      "loss": 0.2423,
+      "step": 9327
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.034470498174178e-06,
+      "loss": 0.2445,
+      "step": 9328
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.031957858599327e-06,
+      "loss": 0.2366,
+      "step": 9329
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.029446142767674e-06,
+      "loss": 0.2714,
+      "step": 9330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0269353508730846e-06,
+      "loss": 0.2263,
+      "step": 9331
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.024425483109347e-06,
+      "loss": 0.2247,
+      "step": 9332
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0219165396701866e-06,
+      "loss": 0.2197,
+      "step": 9333
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0194085207492543e-06,
+      "loss": 0.2097,
+      "step": 9334
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.016901426540123e-06,
+      "loss": 0.23,
+      "step": 9335
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0143952572363103e-06,
+      "loss": 0.2266,
+      "step": 9336
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0118900130312487e-06,
+      "loss": 0.2563,
+      "step": 9337
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0093856941182982e-06,
+      "loss": 0.1942,
+      "step": 9338
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.006882300690763e-06,
+      "loss": 0.2361,
+      "step": 9339
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0043798329418596e-06,
+      "loss": 0.2349,
+      "step": 9340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0018782910647393e-06,
+      "loss": 0.2394,
+      "step": 9341
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.99937767525248e-06,
+      "loss": 0.2515,
+      "step": 9342
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.996877985698088e-06,
+      "loss": 0.1958,
+      "step": 9343
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9943792225945065e-06,
+      "loss": 0.198,
+      "step": 9344
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9918813861345958e-06,
+      "loss": 0.2155,
+      "step": 9345
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9893844765111456e-06,
+      "loss": 0.2246,
+      "step": 9346
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.986888493916887e-06,
+      "loss": 0.2352,
+      "step": 9347
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9843934385444667e-06,
+      "loss": 0.2027,
+      "step": 9348
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.981899310586461e-06,
+      "loss": 0.2751,
+      "step": 9349
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9794061102353787e-06,
+      "loss": 0.1988,
+      "step": 9350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9769138376836504e-06,
+      "loss": 0.2463,
+      "step": 9351
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9744224931236486e-06,
+      "loss": 0.2471,
+      "step": 9352
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9719320767476613e-06,
+      "loss": 0.2596,
+      "step": 9353
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.969442588747907e-06,
+      "loss": 0.2275,
+      "step": 9354
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9669540293165397e-06,
+      "loss": 0.2638,
+      "step": 9355
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9644663986456344e-06,
+      "loss": 0.2184,
+      "step": 9356
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9619796969271972e-06,
+      "loss": 0.2086,
+      "step": 9357
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9594939243531617e-06,
+      "loss": 0.2233,
+      "step": 9358
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.957009081115389e-06,
+      "loss": 0.2254,
+      "step": 9359
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.954525167405667e-06,
+      "loss": 0.2076,
+      "step": 9360
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9520421834157217e-06,
+      "loss": 0.2088,
+      "step": 9361
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9495601293371943e-06,
+      "loss": 0.2109,
+      "step": 9362
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.947079005361665e-06,
+      "loss": 0.201,
+      "step": 9363
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.944598811680633e-06,
+      "loss": 0.2667,
+      "step": 9364
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9421195484855336e-06,
+      "loss": 0.242,
+      "step": 9365
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9396412159677223e-06,
+      "loss": 0.208,
+      "step": 9366
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9371638143184914e-06,
+      "loss": 0.2268,
+      "step": 9367
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.93468734372905e-06,
+      "loss": 0.2331,
+      "step": 9368
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.93221180439055e-06,
+      "loss": 0.2281,
+      "step": 9369
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9297371964940586e-06,
+      "loss": 0.2299,
+      "step": 9370
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9272635202305824e-06,
+      "loss": 0.2386,
+      "step": 9371
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9247907757910462e-06,
+      "loss": 0.244,
+      "step": 9372
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9223189633663073e-06,
+      "loss": 0.2054,
+      "step": 9373
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.919848083147149e-06,
+      "loss": 0.2101,
+      "step": 9374
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9173781353242864e-06,
+      "loss": 0.2684,
+      "step": 9375
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9149091200883543e-06,
+      "loss": 0.2297,
+      "step": 9376
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.912441037629931e-06,
+      "loss": 0.2416,
+      "step": 9377
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.909973888139507e-06,
+      "loss": 0.2496,
+      "step": 9378
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.907507671807507e-06,
+      "loss": 0.2179,
+      "step": 9379
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.905042388824289e-06,
+      "loss": 0.2709,
+      "step": 9380
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9025780393801298e-06,
+      "loss": 0.2061,
+      "step": 9381
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9001146236652383e-06,
+      "loss": 0.2342,
+      "step": 9382
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8976521418697514e-06,
+      "loss": 0.2721,
+      "step": 9383
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8951905941837314e-06,
+      "loss": 0.2358,
+      "step": 9384
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.892729980797176e-06,
+      "loss": 0.2444,
+      "step": 9385
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.890270301900003e-06,
+      "loss": 0.2382,
+      "step": 9386
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8878115576820568e-06,
+      "loss": 0.2622,
+      "step": 9387
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8853537483331214e-06,
+      "loss": 0.2271,
+      "step": 9388
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8828968740428953e-06,
+      "loss": 0.2528,
+      "step": 9389
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8804409350010126e-06,
+      "loss": 0.2534,
+      "step": 9390
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8779859313970303e-06,
+      "loss": 0.2332,
+      "step": 9391
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8755318634204353e-06,
+      "loss": 0.278,
+      "step": 9392
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.873078731260646e-06,
+      "loss": 0.2295,
+      "step": 9393
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.870626535107006e-06,
+      "loss": 0.2742,
+      "step": 9394
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.868175275148779e-06,
+      "loss": 0.2178,
+      "step": 9395
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8657249515751717e-06,
+      "loss": 0.2335,
+      "step": 9396
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.863275564575308e-06,
+      "loss": 0.234,
+      "step": 9397
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.86082711433824e-06,
+      "loss": 0.2409,
+      "step": 9398
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.858379601052951e-06,
+      "loss": 0.2529,
+      "step": 9399
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.855933024908344e-06,
+      "loss": 0.2015,
+      "step": 9400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8534873860932647e-06,
+      "loss": 0.2148,
+      "step": 9401
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.851042684796475e-06,
+      "loss": 0.221,
+      "step": 9402
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.848598921206662e-06,
+      "loss": 0.2196,
+      "step": 9403
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8461560955124537e-06,
+      "loss": 0.2237,
+      "step": 9404
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8437142079023924e-06,
+      "loss": 0.2245,
+      "step": 9405
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8412732585649557e-06,
+      "loss": 0.2022,
+      "step": 9406
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.838833247688543e-06,
+      "loss": 0.2444,
+      "step": 9407
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8363941754614888e-06,
+      "loss": 0.2245,
+      "step": 9408
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8339560420720417e-06,
+      "loss": 0.2405,
+      "step": 9409
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.831518847708399e-06,
+      "loss": 0.2227,
+      "step": 9410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8290825925586624e-06,
+      "loss": 0.2272,
+      "step": 9411
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.826647276810884e-06,
+      "loss": 0.2128,
+      "step": 9412
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.824212900653023e-06,
+      "loss": 0.2452,
+      "step": 9413
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.821779464272977e-06,
+      "loss": 0.1876,
+      "step": 9414
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8193469678585704e-06,
+      "loss": 0.2649,
+      "step": 9415
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8169154115975506e-06,
+      "loss": 0.2417,
+      "step": 9416
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8144847956775937e-06,
+      "loss": 0.2031,
+      "step": 9417
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8120551202863102e-06,
+      "loss": 0.2313,
+      "step": 9418
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8096263856112288e-06,
+      "loss": 0.2452,
+      "step": 9419
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.807198591839807e-06,
+      "loss": 0.2368,
+      "step": 9420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.8047717391594403e-06,
+      "loss": 0.2068,
+      "step": 9421
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.802345827757437e-06,
+      "loss": 0.2464,
+      "step": 9422
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7999208578210394e-06,
+      "loss": 0.206,
+      "step": 9423
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7974968295374193e-06,
+      "loss": 0.2223,
+      "step": 9424
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.795073743093667e-06,
+      "loss": 0.2244,
+      "step": 9425
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.792651598676814e-06,
+      "loss": 0.2439,
+      "step": 9426
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7902303964738083e-06,
+      "loss": 0.2375,
+      "step": 9427
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7878101366715253e-06,
+      "loss": 0.2405,
+      "step": 9428
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.785390819456777e-06,
+      "loss": 0.233,
+      "step": 9429
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7829724450162937e-06,
+      "loss": 0.242,
+      "step": 9430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7805550135367346e-06,
+      "loss": 0.2084,
+      "step": 9431
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7781385252046876e-06,
+      "loss": 0.2277,
+      "step": 9432
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.775722980206665e-06,
+      "loss": 0.2094,
+      "step": 9433
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.773308378729112e-06,
+      "loss": 0.2429,
+      "step": 9434
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7708947209583986e-06,
+      "loss": 0.2153,
+      "step": 9435
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7684820070808136e-06,
+      "loss": 0.2316,
+      "step": 9436
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7660702372825884e-06,
+      "loss": 0.2188,
+      "step": 9437
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7636594117498725e-06,
+      "loss": 0.2847,
+      "step": 9438
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7612495306687404e-06,
+      "loss": 0.2219,
+      "step": 9439
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7588405942251975e-06,
+      "loss": 0.2469,
+      "step": 9440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7564326026051727e-06,
+      "loss": 0.2124,
+      "step": 9441
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7540255559945305e-06,
+      "loss": 0.1918,
+      "step": 9442
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7516194545790534e-06,
+      "loss": 0.2544,
+      "step": 9443
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7492142985444536e-06,
+      "loss": 0.2251,
+      "step": 9444
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7468100880763734e-06,
+      "loss": 0.1894,
+      "step": 9445
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.74440682336038e-06,
+      "loss": 0.2241,
+      "step": 9446
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.742004504581967e-06,
+      "loss": 0.2554,
+      "step": 9447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7396031319265537e-06,
+      "loss": 0.2143,
+      "step": 9448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7372027055794885e-06,
+      "loss": 0.2528,
+      "step": 9449
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7348032257260417e-06,
+      "loss": 0.2512,
+      "step": 9450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7324046925514256e-06,
+      "loss": 0.2137,
+      "step": 9451
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7300071062407584e-06,
+      "loss": 0.2488,
+      "step": 9452
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7276104669791053e-06,
+      "loss": 0.2173,
+      "step": 9453
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7252147749514435e-06,
+      "loss": 0.2295,
+      "step": 9454
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7228200303426833e-06,
+      "loss": 0.1892,
+      "step": 9455
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.72042623333766e-06,
+      "loss": 0.2386,
+      "step": 9456
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.718033384121138e-06,
+      "loss": 0.2176,
+      "step": 9457
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.715641482877804e-06,
+      "loss": 0.2337,
+      "step": 9458
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.713250529792281e-06,
+      "loss": 0.2023,
+      "step": 9459
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7108605250491076e-06,
+      "loss": 0.1991,
+      "step": 9460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7084714688327577e-06,
+      "loss": 0.2374,
+      "step": 9461
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7060833613276293e-06,
+      "loss": 0.2115,
+      "step": 9462
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.703696202718043e-06,
+      "loss": 0.2449,
+      "step": 9463
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7013099931882514e-06,
+      "loss": 0.2251,
+      "step": 9464
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6989247329224304e-06,
+      "loss": 0.2935,
+      "step": 9465
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.696540422104683e-06,
+      "loss": 0.2091,
+      "step": 9466
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.694157060919047e-06,
+      "loss": 0.2647,
+      "step": 9467
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6917746495494733e-06,
+      "loss": 0.2283,
+      "step": 9468
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6893931881798466e-06,
+      "loss": 0.2476,
+      "step": 9469
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6870126769939852e-06,
+      "loss": 0.2607,
+      "step": 9470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.68463311617562e-06,
+      "loss": 0.1991,
+      "step": 9471
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6822545059084193e-06,
+      "loss": 0.2298,
+      "step": 9472
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.679876846375969e-06,
+      "loss": 0.2487,
+      "step": 9473
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6775001377617886e-06,
+      "loss": 0.2294,
+      "step": 9474
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.675124380249327e-06,
+      "loss": 0.2617,
+      "step": 9475
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.672749574021951e-06,
+      "loss": 0.2288,
+      "step": 9476
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.670375719262954e-06,
+      "loss": 0.2188,
+      "step": 9477
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6680028161555696e-06,
+      "loss": 0.2138,
+      "step": 9478
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.665630864882943e-06,
+      "loss": 0.2134,
+      "step": 9479
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.663259865628151e-06,
+      "loss": 0.2146,
+      "step": 9480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.660889818574197e-06,
+      "loss": 0.2493,
+      "step": 9481
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.65852072390401e-06,
+      "loss": 0.2292,
+      "step": 9482
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.65615258180045e-06,
+      "loss": 0.2224,
+      "step": 9483
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6537853924462994e-06,
+      "loss": 0.2156,
+      "step": 9484
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.651419156024262e-06,
+      "loss": 0.2179,
+      "step": 9485
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.649053872716984e-06,
+      "loss": 0.1991,
+      "step": 9486
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6466895427070213e-06,
+      "loss": 0.2465,
+      "step": 9487
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6443261661768627e-06,
+      "loss": 0.2478,
+      "step": 9488
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.641963743308926e-06,
+      "loss": 0.2466,
+      "step": 9489
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6396022742855463e-06,
+      "loss": 0.2061,
+      "step": 9490
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6372417592890008e-06,
+      "loss": 0.219,
+      "step": 9491
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6348821985014786e-06,
+      "loss": 0.2105,
+      "step": 9492
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6325235921050975e-06,
+      "loss": 0.2031,
+      "step": 9493
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6301659402819124e-06,
+      "loss": 0.2179,
+      "step": 9494
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6278092432138934e-06,
+      "loss": 0.2419,
+      "step": 9495
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.625453501082937e-06,
+      "loss": 0.2354,
+      "step": 9496
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6230987140708735e-06,
+      "loss": 0.2414,
+      "step": 9497
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6207448823594534e-06,
+      "loss": 0.1995,
+      "step": 9498
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6183920061303497e-06,
+      "loss": 0.2066,
+      "step": 9499
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6160400855651773e-06,
+      "loss": 0.2326,
+      "step": 9500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6136891208454596e-06,
+      "loss": 0.2252,
+      "step": 9501
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6113391121526577e-06,
+      "loss": 0.2128,
+      "step": 9502
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.608990059668155e-06,
+      "loss": 0.2437,
+      "step": 9503
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6066419635732563e-06,
+      "loss": 0.2719,
+      "step": 9504
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.604294824049209e-06,
+      "loss": 0.2319,
+      "step": 9505
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.601948641277162e-06,
+      "loss": 0.2076,
+      "step": 9506
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5996034154382064e-06,
+      "loss": 0.2724,
+      "step": 9507
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.597259146713362e-06,
+      "loss": 0.2405,
+      "step": 9508
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.594915835283566e-06,
+      "loss": 0.265,
+      "step": 9509
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5925734813296803e-06,
+      "loss": 0.2587,
+      "step": 9510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5902320850325066e-06,
+      "loss": 0.198,
+      "step": 9511
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.587891646572754e-06,
+      "loss": 0.2285,
+      "step": 9512
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5855521661310815e-06,
+      "loss": 0.2115,
+      "step": 9513
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.583213643888046e-06,
+      "loss": 0.1958,
+      "step": 9514
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.580876080024145e-06,
+      "loss": 0.2202,
+      "step": 9515
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.578539474719809e-06,
+      "loss": 0.2482,
+      "step": 9516
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5762038281553834e-06,
+      "loss": 0.2762,
+      "step": 9517
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.57386914051114e-06,
+      "loss": 0.2316,
+      "step": 9518
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.571535411967285e-06,
+      "loss": 0.24,
+      "step": 9519
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.569202642703944e-06,
+      "loss": 0.2495,
+      "step": 9520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5668708329011685e-06,
+      "loss": 0.2093,
+      "step": 9521
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.564539982738937e-06,
+      "loss": 0.188,
+      "step": 9522
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.562210092397152e-06,
+      "loss": 0.2335,
+      "step": 9523
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5598811620556495e-06,
+      "loss": 0.2429,
+      "step": 9524
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5575531918941843e-06,
+      "loss": 0.2366,
+      "step": 9525
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5552261820924338e-06,
+      "loss": 0.2266,
+      "step": 9526
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5529001328300133e-06,
+      "loss": 0.2226,
+      "step": 9527
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5505750442864554e-06,
+      "loss": 0.2491,
+      "step": 9528
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5482509166412137e-06,
+      "loss": 0.2379,
+      "step": 9529
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5459277500736878e-06,
+      "loss": 0.2197,
+      "step": 9530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5436055447631723e-06,
+      "loss": 0.2118,
+      "step": 9531
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5412843008889157e-06,
+      "loss": 0.2183,
+      "step": 9532
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.53896401863008e-06,
+      "loss": 0.2202,
+      "step": 9533
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.536644698165748e-06,
+      "loss": 0.2088,
+      "step": 9534
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.534326339674942e-06,
+      "loss": 0.2184,
+      "step": 9535
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5320089433366007e-06,
+      "loss": 0.2008,
+      "step": 9536
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.529692509329587e-06,
+      "loss": 0.2345,
+      "step": 9537
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5273770378327e-06,
+      "loss": 0.2281,
+      "step": 9538
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5250625290246513e-06,
+      "loss": 0.2207,
+      "step": 9539
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5227489830840806e-06,
+      "loss": 0.2225,
+      "step": 9540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.520436400189566e-06,
+      "loss": 0.2089,
+      "step": 9541
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5181247805195943e-06,
+      "loss": 0.2595,
+      "step": 9542
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.515814124252594e-06,
+      "loss": 0.2323,
+      "step": 9543
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5135044315669063e-06,
+      "loss": 0.1993,
+      "step": 9544
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5111957026408012e-06,
+      "loss": 0.2206,
+      "step": 9545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.508887937652482e-06,
+      "loss": 0.2478,
+      "step": 9546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.50658113678007e-06,
+      "loss": 0.2101,
+      "step": 9547
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.504275300201606e-06,
+      "loss": 0.2221,
+      "step": 9548
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.501970428095073e-06,
+      "loss": 0.201,
+      "step": 9549
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.499666520638366e-06,
+      "loss": 0.2228,
+      "step": 9550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.497363578009309e-06,
+      "loss": 0.2427,
+      "step": 9551
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4950616003856593e-06,
+      "loss": 0.2336,
+      "step": 9552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.492760587945084e-06,
+      "loss": 0.2762,
+      "step": 9553
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4904605408651943e-06,
+      "loss": 0.206,
+      "step": 9554
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.488161459323517e-06,
+      "loss": 0.2125,
+      "step": 9555
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.485863343497492e-06,
+      "loss": 0.1968,
+      "step": 9556
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4835661935645092e-06,
+      "loss": 0.1911,
+      "step": 9557
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4812700097018704e-06,
+      "loss": 0.241,
+      "step": 9558
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4789747920867984e-06,
+      "loss": 0.194,
+      "step": 9559
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4766805408964565e-06,
+      "loss": 0.2381,
+      "step": 9560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4743872563079167e-06,
+      "loss": 0.2301,
+      "step": 9561
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4720949384981922e-06,
+      "loss": 0.2,
+      "step": 9562
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.469803587644213e-06,
+      "loss": 0.2837,
+      "step": 9563
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.467513203922823e-06,
+      "loss": 0.2227,
+      "step": 9564
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4652237875108173e-06,
+      "loss": 0.2704,
+      "step": 9565
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4629353385848964e-06,
+      "loss": 0.2303,
+      "step": 9566
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4606478573216884e-06,
+      "loss": 0.2313,
+      "step": 9567
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4583613438977593e-06,
+      "loss": 0.2249,
+      "step": 9568
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4560757984895894e-06,
+      "loss": 0.2391,
+      "step": 9569
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4537912212735793e-06,
+      "loss": 0.2043,
+      "step": 9570
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4515076124260726e-06,
+      "loss": 0.2226,
+      "step": 9571
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4492249721233222e-06,
+      "loss": 0.2398,
+      "step": 9572
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.446943300541514e-06,
+      "loss": 0.2224,
+      "step": 9573
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.444662597856753e-06,
+      "loss": 0.236,
+      "step": 9574
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4423828642450735e-06,
+      "loss": 0.3229,
+      "step": 9575
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4401040998824405e-06,
+      "loss": 0.2266,
+      "step": 9576
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4378263049447343e-06,
+      "loss": 0.2545,
+      "step": 9577
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4355494796077625e-06,
+      "loss": 0.2368,
+      "step": 9578
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4332736240472654e-06,
+      "loss": 0.2325,
+      "step": 9579
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.430998738438905e-06,
+      "loss": 0.1996,
+      "step": 9580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.428724822958253e-06,
+      "loss": 0.255,
+      "step": 9581
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4264518777808333e-06,
+      "loss": 0.2318,
+      "step": 9582
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.424179903082071e-06,
+      "loss": 0.2143,
+      "step": 9583
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.421908899037336e-06,
+      "loss": 0.2121,
+      "step": 9584
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.41963886582191e-06,
+      "loss": 0.2172,
+      "step": 9585
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.417369803611e-06,
+      "loss": 0.2215,
+      "step": 9586
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4151017125797477e-06,
+      "loss": 0.2183,
+      "step": 9587
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.4128345929032113e-06,
+      "loss": 0.2183,
+      "step": 9588
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.410568444756375e-06,
+      "loss": 0.2435,
+      "step": 9589
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4083032683141525e-06,
+      "loss": 0.22,
+      "step": 9590
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4060390637513736e-06,
+      "loss": 0.217,
+      "step": 9591
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4037758312428065e-06,
+      "loss": 0.2307,
+      "step": 9592
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4015135709631337e-06,
+      "loss": 0.237,
+      "step": 9593
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3992522830869618e-06,
+      "loss": 0.2149,
+      "step": 9594
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.396991967788833e-06,
+      "loss": 0.233,
+      "step": 9595
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3947326252432062e-06,
+      "loss": 0.2248,
+      "step": 9596
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.392474255624465e-06,
+      "loss": 0.2517,
+      "step": 9597
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.39021685910692e-06,
+      "loss": 0.226,
+      "step": 9598
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3879604358648088e-06,
+      "loss": 0.2341,
+      "step": 9599
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3857049860722853e-06,
+      "loss": 0.1919,
+      "step": 9600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3834505099034416e-06,
+      "loss": 0.2498,
+      "step": 9601
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.38119700753228e-06,
+      "loss": 0.2446,
+      "step": 9602
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3789444791327456e-06,
+      "loss": 0.2432,
+      "step": 9603
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3766929248786906e-06,
+      "loss": 0.216,
+      "step": 9604
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.374442344943902e-06,
+      "loss": 0.227,
+      "step": 9605
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.372192739502089e-06,
+      "loss": 0.2393,
+      "step": 9606
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3699441087268825e-06,
+      "loss": 0.2688,
+      "step": 9607
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3676964527918382e-06,
+      "loss": 0.2482,
+      "step": 9608
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3654497718704505e-06,
+      "loss": 0.2317,
+      "step": 9609
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3632040661361205e-06,
+      "loss": 0.1962,
+      "step": 9610
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.360959335762178e-06,
+      "loss": 0.229,
+      "step": 9611
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3587155809218875e-06,
+      "loss": 0.2407,
+      "step": 9612
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.356472801788429e-06,
+      "loss": 0.233,
+      "step": 9613
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3542309985349075e-06,
+      "loss": 0.2081,
+      "step": 9614
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.351990171334358e-06,
+      "loss": 0.2285,
+      "step": 9615
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3497503203597316e-06,
+      "loss": 0.229,
+      "step": 9616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3475114457839136e-06,
+      "loss": 0.2269,
+      "step": 9617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.345273547779711e-06,
+      "loss": 0.23,
+      "step": 9618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3430366265198473e-06,
+      "loss": 0.2213,
+      "step": 9619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3408006821769857e-06,
+      "loss": 0.2169,
+      "step": 9620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3385657149237016e-06,
+      "loss": 0.2335,
+      "step": 9621
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.336331724932499e-06,
+      "loss": 0.2332,
+      "step": 9622
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3340987123758063e-06,
+      "loss": 0.2122,
+      "step": 9623
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.331866677425974e-06,
+      "loss": 0.2239,
+      "step": 9624
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.329635620255286e-06,
+      "loss": 0.247,
+      "step": 9625
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.327405541035943e-06,
+      "loss": 0.2168,
+      "step": 9626
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3251764399400655e-06,
+      "loss": 0.2337,
+      "step": 9627
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3229483171397126e-06,
+      "loss": 0.2445,
+      "step": 9628
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3207211728068584e-06,
+      "loss": 0.2424,
+      "step": 9629
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3184950071134005e-06,
+      "loss": 0.2413,
+      "step": 9630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3162698202311663e-06,
+      "loss": 0.2382,
+      "step": 9631
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.314045612331898e-06,
+      "loss": 0.2086,
+      "step": 9632
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3118223835872793e-06,
+      "loss": 0.2278,
+      "step": 9633
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3096001341689036e-06,
+      "loss": 0.3047,
+      "step": 9634
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3073788642482896e-06,
+      "loss": 0.2245,
+      "step": 9635
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.305158573996893e-06,
+      "loss": 0.2336,
+      "step": 9636
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.302939263586078e-06,
+      "loss": 0.2268,
+      "step": 9637
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3007209331871433e-06,
+      "loss": 0.2465,
+      "step": 9638
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.298503582971308e-06,
+      "loss": 0.2356,
+      "step": 9639
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2962872131097158e-06,
+      "loss": 0.2378,
+      "step": 9640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.294071823773431e-06,
+      "loss": 0.2062,
+      "step": 9641
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2918574151334554e-06,
+      "loss": 0.2179,
+      "step": 9642
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.289643987360699e-06,
+      "loss": 0.2124,
+      "step": 9643
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.28743154062601e-06,
+      "loss": 0.2112,
+      "step": 9644
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.28522007510015e-06,
+      "loss": 0.2236,
+      "step": 9645
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2830095909538098e-06,
+      "loss": 0.2529,
+      "step": 9646
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2808000883576035e-06,
+      "loss": 0.2344,
+      "step": 9647
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2785915674820707e-06,
+      "loss": 0.2391,
+      "step": 9648
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2763840284976696e-06,
+      "loss": 0.2421,
+      "step": 9649
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.274177471574794e-06,
+      "loss": 0.2512,
+      "step": 9650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.271971896883749e-06,
+      "loss": 0.2319,
+      "step": 9651
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.269767304594777e-06,
+      "loss": 0.191,
+      "step": 9652
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.267563694878034e-06,
+      "loss": 0.2182,
+      "step": 9653
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2653610679036015e-06,
+      "loss": 0.2418,
+      "step": 9654
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2631594238414905e-06,
+      "loss": 0.2424,
+      "step": 9655
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2609587628616317e-06,
+      "loss": 0.2191,
+      "step": 9656
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2587590851338795e-06,
+      "loss": 0.2151,
+      "step": 9657
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.256560390828018e-06,
+      "loss": 0.2585,
+      "step": 9658
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2543626801137494e-06,
+      "loss": 0.2154,
+      "step": 9659
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2521659531607002e-06,
+      "loss": 0.2632,
+      "step": 9660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.249970210138428e-06,
+      "loss": 0.2785,
+      "step": 9661
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2477754512164066e-06,
+      "loss": 0.2139,
+      "step": 9662
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2455816765640364e-06,
+      "loss": 0.2551,
+      "step": 9663
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2433888863506413e-06,
+      "loss": 0.2437,
+      "step": 9664
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.241197080745468e-06,
+      "loss": 0.2077,
+      "step": 9665
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.239006259917696e-06,
+      "loss": 0.2685,
+      "step": 9666
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.236816424036418e-06,
+      "loss": 0.21,
+      "step": 9667
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2346275732706496e-06,
+      "loss": 0.1938,
+      "step": 9668
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.232439707789345e-06,
+      "loss": 0.2546,
+      "step": 9669
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2302528277613676e-06,
+      "loss": 0.2332,
+      "step": 9670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.228066933355511e-06,
+      "loss": 0.2203,
+      "step": 9671
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.22588202474049e-06,
+      "loss": 0.2191,
+      "step": 9672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2236981020849427e-06,
+      "loss": 0.2541,
+      "step": 9673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.221515165557442e-06,
+      "loss": 0.2254,
+      "step": 9674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2193332153264685e-06,
+      "loss": 0.2074,
+      "step": 9675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2171522515604316e-06,
+      "loss": 0.2231,
+      "step": 9676
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.214972274427678e-06,
+      "loss": 0.2169,
+      "step": 9677
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.212793284096458e-06,
+      "loss": 0.2364,
+      "step": 9678
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.210615280734961e-06,
+      "loss": 0.2087,
+      "step": 9679
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2084382645112905e-06,
+      "loss": 0.2309,
+      "step": 9680
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2062622355934786e-06,
+      "loss": 0.1897,
+      "step": 9681
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.204087194149476e-06,
+      "loss": 0.1875,
+      "step": 9682
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2019131403471703e-06,
+      "loss": 0.2334,
+      "step": 9683
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1997400743543573e-06,
+      "loss": 0.2603,
+      "step": 9684
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1975679963387673e-06,
+      "loss": 0.2322,
+      "step": 9685
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1953969064680486e-06,
+      "loss": 0.2287,
+      "step": 9686
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1932268049097746e-06,
+      "loss": 0.2371,
+      "step": 9687
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.191057691831444e-06,
+      "loss": 0.1965,
+      "step": 9688
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.188889567400477e-06,
+      "loss": 0.2113,
+      "step": 9689
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.186722431784214e-06,
+      "loss": 0.2599,
+      "step": 9690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.184556285149932e-06,
+      "loss": 0.2189,
+      "step": 9691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1823911276648156e-06,
+      "loss": 0.2016,
+      "step": 9692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.180226959495987e-06,
+      "loss": 0.2076,
+      "step": 9693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.178063780810483e-06,
+      "loss": 0.215,
+      "step": 9694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.175901591775266e-06,
+      "loss": 0.2318,
+      "step": 9695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.173740392557224e-06,
+      "loss": 0.2278,
+      "step": 9696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1715801833231667e-06,
+      "loss": 0.2099,
+      "step": 9697
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1694209642398244e-06,
+      "loss": 0.2343,
+      "step": 9698
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1672627354738616e-06,
+      "loss": 0.2258,
+      "step": 9699
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1651054971918557e-06,
+      "loss": 0.2547,
+      "step": 9700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1629492495603075e-06,
+      "loss": 0.2519,
+      "step": 9701
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.160793992745653e-06,
+      "loss": 0.2278,
+      "step": 9702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.158639726914241e-06,
+      "loss": 0.2536,
+      "step": 9703
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1564864522323445e-06,
+      "loss": 0.2397,
+      "step": 9704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1543341688661646e-06,
+      "loss": 0.2237,
+      "step": 9705
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1521828769818175e-06,
+      "loss": 0.2429,
+      "step": 9706
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.150032576745358e-06,
+      "loss": 0.1731,
+      "step": 9707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.147883268322749e-06,
+      "loss": 0.2611,
+      "step": 9708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1457349518798837e-06,
+      "loss": 0.2624,
+      "step": 9709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1435876275825806e-06,
+      "loss": 0.2409,
+      "step": 9710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1414412955965788e-06,
+      "loss": 0.2507,
+      "step": 9711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1392959560875397e-06,
+      "loss": 0.2392,
+      "step": 9712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1371516092210504e-06,
+      "loss": 0.1988,
+      "step": 9713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1350082551626166e-06,
+      "loss": 0.2188,
+      "step": 9714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.132865894077678e-06,
+      "loss": 0.2341,
+      "step": 9715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.130724526131587e-06,
+      "loss": 0.2299,
+      "step": 9716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1285841514896202e-06,
+      "loss": 0.2061,
+      "step": 9717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1264447703169886e-06,
+      "loss": 0.2322,
+      "step": 9718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1243063827788135e-06,
+      "loss": 0.2089,
+      "step": 9719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.122168989040146e-06,
+      "loss": 0.2166,
+      "step": 9720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1200325892659573e-06,
+      "loss": 0.2445,
+      "step": 9721
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.117897183621141e-06,
+      "loss": 0.2609,
+      "step": 9722
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.115762772270523e-06,
+      "loss": 0.2501,
+      "step": 9723
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.113629355378844e-06,
+      "loss": 0.2499,
+      "step": 9724
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1114969331107663e-06,
+      "loss": 0.2358,
+      "step": 9725
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1093655056308843e-06,
+      "loss": 0.2264,
+      "step": 9726
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1072350731037073e-06,
+      "loss": 0.2022,
+      "step": 9727
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1051056356936736e-06,
+      "loss": 0.2019,
+      "step": 9728
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1029771935651393e-06,
+      "loss": 0.2199,
+      "step": 9729
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1008497468823874e-06,
+      "loss": 0.2101,
+      "step": 9730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.098723295809618e-06,
+      "loss": 0.2115,
+      "step": 9731
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0965978405109686e-06,
+      "loss": 0.2352,
+      "step": 9732
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.094473381150483e-06,
+      "loss": 0.2163,
+      "step": 9733
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.092349917892143e-06,
+      "loss": 0.2091,
+      "step": 9734
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0902274508998427e-06,
+      "loss": 0.2367,
+      "step": 9735
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.088105980337401e-06,
+      "loss": 0.2597,
+      "step": 9736
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.085985506368565e-06,
+      "loss": 0.2446,
+      "step": 9737
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0838660291569992e-06,
+      "loss": 0.2162,
+      "step": 9738
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.081747548866293e-06,
+      "loss": 0.2173,
+      "step": 9739
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0796300656599633e-06,
+      "loss": 0.2303,
+      "step": 9740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0775135797014426e-06,
+      "loss": 0.2167,
+      "step": 9741
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0753980911540893e-06,
+      "loss": 0.2405,
+      "step": 9742
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.073283600181192e-06,
+      "loss": 0.2212,
+      "step": 9743
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.071170106945951e-06,
+      "loss": 0.2637,
+      "step": 9744
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.069057611611495e-06,
+      "loss": 0.2171,
+      "step": 9745
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0669461143408772e-06,
+      "loss": 0.2159,
+      "step": 9746
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0648356152970644e-06,
+      "loss": 0.2102,
+      "step": 9747
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.062726114642964e-06,
+      "loss": 0.2116,
+      "step": 9748
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0606176125413906e-06,
+      "loss": 0.2273,
+      "step": 9749
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0585101091550845e-06,
+      "loss": 0.226,
+      "step": 9750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.056403604646717e-06,
+      "loss": 0.2301,
+      "step": 9751
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.054298099178877e-06,
+      "loss": 0.2287,
+      "step": 9752
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.052193592914072e-06,
+      "loss": 0.2228,
+      "step": 9753
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0500900860147387e-06,
+      "loss": 0.2231,
+      "step": 9754
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.047987578643231e-06,
+      "loss": 0.2563,
+      "step": 9755
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.045886070961836e-06,
+      "loss": 0.2455,
+      "step": 9756
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0437855631327534e-06,
+      "loss": 0.2146,
+      "step": 9757
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0416860553181055e-06,
+      "loss": 0.2307,
+      "step": 9758
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0395875476799457e-06,
+      "loss": 0.1891,
+      "step": 9759
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0374900403802465e-06,
+      "loss": 0.207,
+      "step": 9760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0353935335809e-06,
+      "loss": 0.2122,
+      "step": 9761
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0332980274437214e-06,
+      "loss": 0.2399,
+      "step": 9762
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.03120352213045e-06,
+      "loss": 0.209,
+      "step": 9763
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0291100178027543e-06,
+      "loss": 0.1893,
+      "step": 9764
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.027017514622216e-06,
+      "loss": 0.2222,
+      "step": 9765
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.02492601275034e-06,
+      "loss": 0.2423,
+      "step": 9766
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0228355123485642e-06,
+      "loss": 0.2792,
+      "step": 9767
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.020746013578238e-06,
+      "loss": 0.235,
+      "step": 9768
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.018657516600639e-06,
+      "loss": 0.2609,
+      "step": 9769
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0165700215769644e-06,
+      "loss": 0.2437,
+      "step": 9770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.014483528668336e-06,
+      "loss": 0.2199,
+      "step": 9771
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0123980380357946e-06,
+      "loss": 0.2374,
+      "step": 9772
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.010313549840315e-06,
+      "loss": 0.2278,
+      "step": 9773
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0082300642427786e-06,
+      "loss": 0.2142,
+      "step": 9774
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.006147581404005e-06,
+      "loss": 0.2775,
+      "step": 9775
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.004066101484725e-06,
+      "loss": 0.2414,
+      "step": 9776
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.001985624645596e-06,
+      "loss": 0.2481,
+      "step": 9777
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.999906151047197e-06,
+      "loss": 0.2025,
+      "step": 9778
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9978276808500325e-06,
+      "loss": 0.2291,
+      "step": 9779
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.995750214214523e-06,
+      "loss": 0.249,
+      "step": 9780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.993673751301022e-06,
+      "loss": 0.2101,
+      "step": 9781
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.991598292269794e-06,
+      "loss": 0.2254,
+      "step": 9782
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.989523837281038e-06,
+      "loss": 0.2201,
+      "step": 9783
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9874503864948636e-06,
+      "loss": 0.231,
+      "step": 9784
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9853779400713084e-06,
+      "loss": 0.2566,
+      "step": 9785
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9833064981703402e-06,
+      "loss": 0.2173,
+      "step": 9786
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9812360609518327e-06,
+      "loss": 0.2388,
+      "step": 9787
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9791666285755895e-06,
+      "loss": 0.2216,
+      "step": 9788
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.977098201201346e-06,
+      "loss": 0.2289,
+      "step": 9789
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9750307789887485e-06,
+      "loss": 0.2663,
+      "step": 9790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9729643620973653e-06,
+      "loss": 0.2358,
+      "step": 9791
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9708989506866975e-06,
+      "loss": 0.2284,
+      "step": 9792
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9688345449161543e-06,
+      "loss": 0.2399,
+      "step": 9793
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9667711449450896e-06,
+      "loss": 0.17,
+      "step": 9794
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.96470875093275e-06,
+      "loss": 0.2419,
+      "step": 9795
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9626473630383223e-06,
+      "loss": 0.2211,
+      "step": 9796
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.96058698142092e-06,
+      "loss": 0.2533,
+      "step": 9797
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9585276062395684e-06,
+      "loss": 0.2236,
+      "step": 9798
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9564692376532127e-06,
+      "loss": 0.2142,
+      "step": 9799
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9544118758207375e-06,
+      "loss": 0.2384,
+      "step": 9800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9523555209009302e-06,
+      "loss": 0.247,
+      "step": 9801
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9503001730525125e-06,
+      "loss": 0.2215,
+      "step": 9802
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9482458324341248e-06,
+      "loss": 0.2154,
+      "step": 9803
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9461924992043222e-06,
+      "loss": 0.2145,
+      "step": 9804
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.944140173521601e-06,
+      "loss": 0.2336,
+      "step": 9805
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9420888555443617e-06,
+      "loss": 0.1946,
+      "step": 9806
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9400385454309314e-06,
+      "loss": 0.281,
+      "step": 9807
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.937989243339569e-06,
+      "loss": 0.2064,
+      "step": 9808
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9359409494284443e-06,
+      "loss": 0.2235,
+      "step": 9809
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9338936638556488e-06,
+      "loss": 0.2449,
+      "step": 9810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9318473867792115e-06,
+      "loss": 0.232,
+      "step": 9811
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.929802118357059e-06,
+      "loss": 0.2212,
+      "step": 9812
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9277578587470625e-06,
+      "loss": 0.2246,
+      "step": 9813
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.925714608107005e-06,
+      "loss": 0.2362,
+      "step": 9814
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.923672366594589e-06,
+      "loss": 0.2309,
+      "step": 9815
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.921631134367448e-06,
+      "loss": 0.2297,
+      "step": 9816
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9195909115831334e-06,
+      "loss": 0.2083,
+      "step": 9817
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.917551698399112e-06,
+      "loss": 0.2298,
+      "step": 9818
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9155134949727883e-06,
+      "loss": 0.2074,
+      "step": 9819
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.913476301461471e-06,
+      "loss": 0.2241,
+      "step": 9820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9114401180223977e-06,
+      "loss": 0.2162,
+      "step": 9821
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9094049448127383e-06,
+      "loss": 0.2476,
+      "step": 9822
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9073707819895665e-06,
+      "loss": 0.2424,
+      "step": 9823
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9053376297098968e-06,
+      "loss": 0.2104,
+      "step": 9824
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9033054881306504e-06,
+      "loss": 0.2233,
+      "step": 9825
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9012743574086739e-06,
+      "loss": 0.1999,
+      "step": 9826
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8992442377007464e-06,
+      "loss": 0.265,
+      "step": 9827
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8972151291635593e-06,
+      "loss": 0.2316,
+      "step": 9828
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.895187031953719e-06,
+      "loss": 0.2854,
+      "step": 9829
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8931599462277721e-06,
+      "loss": 0.2581,
+      "step": 9830
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8911338721421735e-06,
+      "loss": 0.2231,
+      "step": 9831
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8891088098533011e-06,
+      "loss": 0.2297,
+      "step": 9832
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8870847595174651e-06,
+      "loss": 0.2476,
+      "step": 9833
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8850617212908822e-06,
+      "loss": 0.1674,
+      "step": 9834
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8830396953297058e-06,
+      "loss": 0.2354,
+      "step": 9835
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8810186817900043e-06,
+      "loss": 0.2469,
+      "step": 9836
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.878998680827758e-06,
+      "loss": 0.2366,
+      "step": 9837
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8769796925988892e-06,
+      "loss": 0.2562,
+      "step": 9838
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8749617172592276e-06,
+      "loss": 0.2019,
+      "step": 9839
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8729447549645272e-06,
+      "loss": 0.2204,
+      "step": 9840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8709288058704705e-06,
+      "loss": 0.1959,
+      "step": 9841
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8689138701326542e-06,
+      "loss": 0.2233,
+      "step": 9842
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8668999479065974e-06,
+      "loss": 0.2295,
+      "step": 9843
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8648870393477475e-06,
+      "loss": 0.2181,
+      "step": 9844
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8628751446114668e-06,
+      "loss": 0.2428,
+      "step": 9845
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.860864263853041e-06,
+      "loss": 0.2225,
+      "step": 9846
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8588543972276795e-06,
+      "loss": 0.2563,
+      "step": 9847
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.856845544890508e-06,
+      "loss": 0.2015,
+      "step": 9848
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8548377069965844e-06,
+      "loss": 0.2018,
+      "step": 9849
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8528308837008795e-06,
+      "loss": 0.199,
+      "step": 9850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8508250751582846e-06,
+      "loss": 0.2005,
+      "step": 9851
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8488202815236239e-06,
+      "loss": 0.2429,
+      "step": 9852
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8468165029516288e-06,
+      "loss": 0.2301,
+      "step": 9853
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.844813739596964e-06,
+      "loss": 0.2304,
+      "step": 9854
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.842811991614209e-06,
+      "loss": 0.2539,
+      "step": 9855
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8408112591578618e-06,
+      "loss": 0.2412,
+      "step": 9856
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8388115423823577e-06,
+      "loss": 0.2241,
+      "step": 9857
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.836812841442036e-06,
+      "loss": 0.2405,
+      "step": 9858
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8348151564911653e-06,
+      "loss": 0.2427,
+      "step": 9859
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.832818487683937e-06,
+      "loss": 0.2097,
+      "step": 9860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8308228351744683e-06,
+      "loss": 0.255,
+      "step": 9861
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8288281991167771e-06,
+      "loss": 0.2429,
+      "step": 9862
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.826834579664829e-06,
+      "loss": 0.2027,
+      "step": 9863
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8248419769724939e-06,
+      "loss": 0.2232,
+      "step": 9864
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8228503911935757e-06,
+      "loss": 0.2123,
+      "step": 9865
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8208598224817895e-06,
+      "loss": 0.2452,
+      "step": 9866
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8188702709907728e-06,
+      "loss": 0.2342,
+      "step": 9867
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8168817368740941e-06,
+      "loss": 0.2342,
+      "step": 9868
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8148942202852326e-06,
+      "loss": 0.2176,
+      "step": 9869
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8129077213775935e-06,
+      "loss": 0.2134,
+      "step": 9870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8109222403045027e-06,
+      "loss": 0.2137,
+      "step": 9871
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8089377772192078e-06,
+      "loss": 0.2256,
+      "step": 9872
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8069543322748761e-06,
+      "loss": 0.2601,
+      "step": 9873
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.804971905624605e-06,
+      "loss": 0.2328,
+      "step": 9874
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8029904974213973e-06,
+      "loss": 0.2259,
+      "step": 9875
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.801010107818194e-06,
+      "loss": 0.2274,
+      "step": 9876
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.799030736967846e-06,
+      "loss": 0.2497,
+      "step": 9877
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7970523850231313e-06,
+      "loss": 0.2322,
+      "step": 9878
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7950750521367443e-06,
+      "loss": 0.2288,
+      "step": 9879
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7930987384613063e-06,
+      "loss": 0.2221,
+      "step": 9880
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.7911234441493535e-06,
+      "loss": 0.2452,
+      "step": 9881
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.789149169353354e-06,
+      "loss": 0.2226,
+      "step": 9882
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7871759142256844e-06,
+      "loss": 0.1944,
+      "step": 9883
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7852036789186526e-06,
+      "loss": 0.2138,
+      "step": 9884
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7832324635844837e-06,
+      "loss": 0.2404,
+      "step": 9885
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7812622683753228e-06,
+      "loss": 0.2141,
+      "step": 9886
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7792930934432395e-06,
+      "loss": 0.233,
+      "step": 9887
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.777324938940219e-06,
+      "loss": 0.2026,
+      "step": 9888
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7753578050181734e-06,
+      "loss": 0.2152,
+      "step": 9889
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7733916918289362e-06,
+      "loss": 0.2135,
+      "step": 9890
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7714265995242606e-06,
+      "loss": 0.2163,
+      "step": 9891
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.769462528255814e-06,
+      "loss": 0.2248,
+      "step": 9892
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7674994781752018e-06,
+      "loss": 0.2429,
+      "step": 9893
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7655374494339344e-06,
+      "loss": 0.2497,
+      "step": 9894
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7635764421834487e-06,
+      "loss": 0.214,
+      "step": 9895
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.761616456575106e-06,
+      "loss": 0.2729,
+      "step": 9896
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.759657492760181e-06,
+      "loss": 0.2308,
+      "step": 9897
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7576995508898818e-06,
+      "loss": 0.219,
+      "step": 9898
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7557426311153275e-06,
+      "loss": 0.2173,
+      "step": 9899
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7537867335875585e-06,
+      "loss": 0.2701,
+      "step": 9900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7518318584575444e-06,
+      "loss": 0.2363,
+      "step": 9901
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7498780058761676e-06,
+      "loss": 0.242,
+      "step": 9902
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7479251759942343e-06,
+      "loss": 0.2255,
+      "step": 9903
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7459733689624724e-06,
+      "loss": 0.2395,
+      "step": 9904
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7440225849315278e-06,
+      "loss": 0.2341,
+      "step": 9905
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7420728240519751e-06,
+      "loss": 0.2593,
+      "step": 9906
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.740124086474304e-06,
+      "loss": 0.2354,
+      "step": 9907
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7381763723489208e-06,
+      "loss": 0.2123,
+      "step": 9908
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7362296818261652e-06,
+      "loss": 0.2191,
+      "step": 9909
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.734284015056289e-06,
+      "loss": 0.233,
+      "step": 9910
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7323393721894648e-06,
+      "loss": 0.2406,
+      "step": 9911
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7303957533757897e-06,
+      "loss": 0.2645,
+      "step": 9912
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7284531587652752e-06,
+      "loss": 0.2124,
+      "step": 9913
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.726511588507868e-06,
+      "loss": 0.2193,
+      "step": 9914
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7245710427534218e-06,
+      "loss": 0.248,
+      "step": 9915
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7226315216517113e-06,
+      "loss": 0.2268,
+      "step": 9916
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7206930253524471e-06,
+      "loss": 0.2009,
+      "step": 9917
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.718755554005243e-06,
+      "loss": 0.2048,
+      "step": 9918
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7168191077596424e-06,
+      "loss": 0.206,
+      "step": 9919
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7148836867651107e-06,
+      "loss": 0.2152,
+      "step": 9920
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7129492911710288e-06,
+      "loss": 0.2449,
+      "step": 9921
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7110159211266984e-06,
+      "loss": 0.2207,
+      "step": 9922
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.709083576781354e-06,
+      "loss": 0.2219,
+      "step": 9923
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.7071522582841325e-06,
+      "loss": 0.2122,
+      "step": 9924
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.70522196578411e-06,
+      "loss": 0.2288,
+      "step": 9925
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.703292699430269e-06,
+      "loss": 0.209,
+      "step": 9926
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.70136445937152e-06,
+      "loss": 0.2135,
+      "step": 9927
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6994372457566908e-06,
+      "loss": 0.2256,
+      "step": 9928
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6975110587345327e-06,
+      "loss": 0.2167,
+      "step": 9929
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6955858984537149e-06,
+      "loss": 0.2219,
+      "step": 9930
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.693661765062835e-06,
+      "loss": 0.2585,
+      "step": 9931
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6917386587104011e-06,
+      "loss": 0.2611,
+      "step": 9932
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6898165795448444e-06,
+      "loss": 0.2415,
+      "step": 9933
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6878955277145247e-06,
+      "loss": 0.2394,
+      "step": 9934
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6859755033677148e-06,
+      "loss": 0.2125,
+      "step": 9935
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.684056506652608e-06,
+      "loss": 0.2488,
+      "step": 9936
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6821385377173226e-06,
+      "loss": 0.236,
+      "step": 9937
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6802215967098932e-06,
+      "loss": 0.2575,
+      "step": 9938
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6783056837782802e-06,
+      "loss": 0.2147,
+      "step": 9939
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6763907990703598e-06,
+      "loss": 0.1986,
+      "step": 9940
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6744769427339274e-06,
+      "loss": 0.237,
+      "step": 9941
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6725641149167115e-06,
+      "loss": 0.2409,
+      "step": 9942
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6706523157663455e-06,
+      "loss": 0.2407,
+      "step": 9943
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.66874154543039e-06,
+      "loss": 0.2233,
+      "step": 9944
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6668318040563286e-06,
+      "loss": 0.2511,
+      "step": 9945
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6649230917915599e-06,
+      "loss": 0.2053,
+      "step": 9946
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6630154087834093e-06,
+      "loss": 0.2136,
+      "step": 9947
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6611087551791193e-06,
+      "loss": 0.2636,
+      "step": 9948
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6592031311258504e-06,
+      "loss": 0.2532,
+      "step": 9949
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6572985367706918e-06,
+      "loss": 0.2332,
+      "step": 9950
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6553949722606437e-06,
+      "loss": 0.1967,
+      "step": 9951
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6534924377426341e-06,
+      "loss": 0.229,
+      "step": 9952
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6515909333635055e-06,
+      "loss": 0.2731,
+      "step": 9953
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6496904592700236e-06,
+      "loss": 0.1704,
+      "step": 9954
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6477910156088776e-06,
+      "loss": 0.2357,
+      "step": 9955
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6458926025266758e-06,
+      "loss": 0.2027,
+      "step": 9956
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6439952201699388e-06,
+      "loss": 0.224,
+      "step": 9957
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6420988686851214e-06,
+      "loss": 0.2428,
+      "step": 9958
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6402035482185895e-06,
+      "loss": 0.2444,
+      "step": 9959
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6383092589166332e-06,
+      "loss": 0.2238,
+      "step": 9960
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6364160009254586e-06,
+      "loss": 0.2533,
+      "step": 9961
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6345237743911973e-06,
+      "loss": 0.2717,
+      "step": 9962
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6326325794598957e-06,
+      "loss": 0.2312,
+      "step": 9963
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6307424162775287e-06,
+      "loss": 0.1911,
+      "step": 9964
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6288532849899828e-06,
+      "loss": 0.2409,
+      "step": 9965
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.626965185743075e-06,
+      "loss": 0.2431,
+      "step": 9966
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6250781186825336e-06,
+      "loss": 0.2377,
+      "step": 9967
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6231920839540087e-06,
+      "loss": 0.2557,
+      "step": 9968
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.621307081703074e-06,
+      "loss": 0.2072,
+      "step": 9969
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6194231120752196e-06,
+      "loss": 0.2085,
+      "step": 9970
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6175401752158576e-06,
+      "loss": 0.2535,
+      "step": 9971
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6156582712703265e-06,
+      "loss": 0.2159,
+      "step": 9972
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6137774003838722e-06,
+      "loss": 0.2576,
+      "step": 9973
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6118975627016751e-06,
+      "loss": 0.25,
+      "step": 9974
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.610018758368824e-06,
+      "loss": 0.2039,
+      "step": 9975
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6081409875303365e-06,
+      "loss": 0.2322,
+      "step": 9976
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6062642503311447e-06,
+      "loss": 0.2592,
+      "step": 9977
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6043885469161013e-06,
+      "loss": 0.2388,
+      "step": 9978
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6025138774299801e-06,
+      "loss": 0.2307,
+      "step": 9979
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6006402420174826e-06,
+      "loss": 0.2603,
+      "step": 9980
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5987676408232177e-06,
+      "loss": 0.2184,
+      "step": 9981
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5968960739917199e-06,
+      "loss": 0.232,
+      "step": 9982
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5950255416674503e-06,
+      "loss": 0.234,
+      "step": 9983
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.59315604399478e-06,
+      "loss": 0.231,
+      "step": 9984
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5912875811180055e-06,
+      "loss": 0.2394,
+      "step": 9985
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.589420153181343e-06,
+      "loss": 0.198,
+      "step": 9986
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5875537603289254e-06,
+      "loss": 0.2906,
+      "step": 9987
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.585688402704814e-06,
+      "loss": 0.1826,
+      "step": 9988
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5838240804529808e-06,
+      "loss": 0.2328,
+      "step": 9989
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.581960793717322e-06,
+      "loss": 0.2404,
+      "step": 9990
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.580098542641656e-06,
+      "loss": 0.2467,
+      "step": 9991
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5782373273697193e-06,
+      "loss": 0.2552,
+      "step": 9992
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5763771480451671e-06,
+      "loss": 0.2278,
+      "step": 9993
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5745180048115749e-06,
+      "loss": 0.1931,
+      "step": 9994
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5726598978124358e-06,
+      "loss": 0.2206,
+      "step": 9995
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5708028271911752e-06,
+      "loss": 0.2191,
+      "step": 9996
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5689467930911233e-06,
+      "loss": 0.2192,
+      "step": 9997
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.567091795655536e-06,
+      "loss": 0.2096,
+      "step": 9998
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5652378350275931e-06,
+      "loss": 0.2324,
+      "step": 9999
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5633849113503889e-06,
+      "loss": 0.2072,
+      "step": 10000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5615330247669406e-06,
+      "loss": 0.2259,
+      "step": 10001
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5596821754201856e-06,
+      "loss": 0.2534,
+      "step": 10002
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5578323634529779e-06,
+      "loss": 0.2163,
+      "step": 10003
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5559835890080899e-06,
+      "loss": 0.2314,
+      "step": 10004
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5541358522282272e-06,
+      "loss": 0.2301,
+      "step": 10005
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5522891532559963e-06,
+      "loss": 0.2428,
+      "step": 10006
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5504434922339412e-06,
+      "loss": 0.2205,
+      "step": 10007
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5485988693045144e-06,
+      "loss": 0.2254,
+      "step": 10008
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5467552846100907e-06,
+      "loss": 0.2179,
+      "step": 10009
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.544912738292966e-06,
+      "loss": 0.2134,
+      "step": 10010
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5430712304953554e-06,
+      "loss": 0.2085,
+      "step": 10011
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.541230761359393e-06,
+      "loss": 0.2732,
+      "step": 10012
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5393913310271369e-06,
+      "loss": 0.2194,
+      "step": 10013
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5375529396405585e-06,
+      "loss": 0.2259,
+      "step": 10014
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5357155873415562e-06,
+      "loss": 0.197,
+      "step": 10015
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5338792742719431e-06,
+      "loss": 0.2347,
+      "step": 10016
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5320440005734543e-06,
+      "loss": 0.2222,
+      "step": 10017
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5302097663877412e-06,
+      "loss": 0.2269,
+      "step": 10018
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5283765718563809e-06,
+      "loss": 0.1953,
+      "step": 10019
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5265444171208597e-06,
+      "loss": 0.2012,
+      "step": 10020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5247133023226e-06,
+      "loss": 0.2287,
+      "step": 10021
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.522883227602932e-06,
+      "loss": 0.2404,
+      "step": 10022
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5210541931031041e-06,
+      "loss": 0.2293,
+      "step": 10023
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.519226198964297e-06,
+      "loss": 0.2181,
+      "step": 10024
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.517399245327596e-06,
+      "loss": 0.2009,
+      "step": 10025
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.515573332334015e-06,
+      "loss": 0.2157,
+      "step": 10026
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5137484601244878e-06,
+      "loss": 0.2673,
+      "step": 10027
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5119246288398586e-06,
+      "loss": 0.2176,
+      "step": 10028
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5101018386209065e-06,
+      "loss": 0.2074,
+      "step": 10029
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5082800896083188e-06,
+      "loss": 0.232,
+      "step": 10030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.506459381942703e-06,
+      "loss": 0.2663,
+      "step": 10031
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5046397157645952e-06,
+      "loss": 0.2129,
+      "step": 10032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5028210912144397e-06,
+      "loss": 0.2184,
+      "step": 10033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5010035084326074e-06,
+      "loss": 0.2288,
+      "step": 10034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.499186967559386e-06,
+      "loss": 0.2325,
+      "step": 10035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.497371468734982e-06,
+      "loss": 0.2035,
+      "step": 10036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4955570120995266e-06,
+      "loss": 0.2022,
+      "step": 10037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4937435977930675e-06,
+      "loss": 0.2527,
+      "step": 10038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4919312259555645e-06,
+      "loss": 0.2375,
+      "step": 10039
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4901198967269143e-06,
+      "loss": 0.2233,
+      "step": 10040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4883096102469162e-06,
+      "loss": 0.2132,
+      "step": 10041
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4865003666552972e-06,
+      "loss": 0.2317,
+      "step": 10042
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4846921660917022e-06,
+      "loss": 0.2289,
+      "step": 10043
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.482885008695693e-06,
+      "loss": 0.2559,
+      "step": 10044
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4810788946067593e-06,
+      "loss": 0.2028,
+      "step": 10045
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4792738239643e-06,
+      "loss": 0.2588,
+      "step": 10046
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4774697969076383e-06,
+      "loss": 0.2356,
+      "step": 10047
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.475666813576018e-06,
+      "loss": 0.2026,
+      "step": 10048
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4738648741086013e-06,
+      "loss": 0.2202,
+      "step": 10049
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4720639786444685e-06,
+      "loss": 0.2093,
+      "step": 10050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4702641273226214e-06,
+      "loss": 0.25,
+      "step": 10051
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.468465320281976e-06,
+      "loss": 0.2172,
+      "step": 10052
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4666675576613725e-06,
+      "loss": 0.233,
+      "step": 10053
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4648708395995752e-06,
+      "loss": 0.2785,
+      "step": 10054
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4630751662352549e-06,
+      "loss": 0.2217,
+      "step": 10055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4612805377070153e-06,
+      "loss": 0.2364,
+      "step": 10056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4594869541533713e-06,
+      "loss": 0.1951,
+      "step": 10057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4576944157127598e-06,
+      "loss": 0.2153,
+      "step": 10058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4559029225235338e-06,
+      "loss": 0.2452,
+      "step": 10059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4541124747239693e-06,
+      "loss": 0.1882,
+      "step": 10060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.452323072452259e-06,
+      "loss": 0.2314,
+      "step": 10061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4505347158465194e-06,
+      "loss": 0.249,
+      "step": 10062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.448747405044783e-06,
+      "loss": 0.2229,
+      "step": 10063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.446961140184998e-06,
+      "loss": 0.2624,
+      "step": 10064
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.445175921405041e-06,
+      "loss": 0.255,
+      "step": 10065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4433917488426978e-06,
+      "loss": 0.2306,
+      "step": 10066
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4416086226356872e-06,
+      "loss": 0.2194,
+      "step": 10067
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.439826542921629e-06,
+      "loss": 0.2348,
+      "step": 10068
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4380455098380695e-06,
+      "loss": 0.215,
+      "step": 10069
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4362655235224858e-06,
+      "loss": 0.2316,
+      "step": 10070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.434486584112258e-06,
+      "loss": 0.2512,
+      "step": 10071
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4327086917446925e-06,
+      "loss": 0.2226,
+      "step": 10072
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4309318465570182e-06,
+      "loss": 0.2479,
+      "step": 10073
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4291560486863752e-06,
+      "loss": 0.2251,
+      "step": 10074
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4273812982698325e-06,
+      "loss": 0.2248,
+      "step": 10075
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4256075954443671e-06,
+      "loss": 0.2313,
+      "step": 10076
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4238349403468809e-06,
+      "loss": 0.2419,
+      "step": 10077
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.422063333114198e-06,
+      "loss": 0.2278,
+      "step": 10078
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4202927738830574e-06,
+      "loss": 0.2415,
+      "step": 10079
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4185232627901146e-06,
+      "loss": 0.232,
+      "step": 10080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4167547999719538e-06,
+      "loss": 0.267,
+      "step": 10081
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4149873855650691e-06,
+      "loss": 0.2273,
+      "step": 10082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4132210197058764e-06,
+      "loss": 0.1861,
+      "step": 10083
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4114557025307134e-06,
+      "loss": 0.2299,
+      "step": 10084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.409691434175831e-06,
+      "loss": 0.2553,
+      "step": 10085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.407928214777407e-06,
+      "loss": 0.2302,
+      "step": 10086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4061660444715308e-06,
+      "loss": 0.2316,
+      "step": 10087
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4044049233942136e-06,
+      "loss": 0.2215,
+      "step": 10088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.40264485168139e-06,
+      "loss": 0.1842,
+      "step": 10089
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.400885829468908e-06,
+      "loss": 0.2004,
+      "step": 10090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3991278568925308e-06,
+      "loss": 0.2167,
+      "step": 10091
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3973709340879582e-06,
+      "loss": 0.2507,
+      "step": 10092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3956150611907865e-06,
+      "loss": 0.2661,
+      "step": 10093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3938602383365407e-06,
+      "loss": 0.2136,
+      "step": 10094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3921064656606707e-06,
+      "loss": 0.2073,
+      "step": 10095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3903537432985354e-06,
+      "loss": 0.2417,
+      "step": 10096
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3886020713854243e-06,
+      "loss": 0.2186,
+      "step": 10097
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3868514500565328e-06,
+      "loss": 0.2115,
+      "step": 10098
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3851018794469793e-06,
+      "loss": 0.2377,
+      "step": 10099
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3833533596918113e-06,
+      "loss": 0.2253,
+      "step": 10100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3816058909259805e-06,
+      "loss": 0.2296,
+      "step": 10101
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3798594732843606e-06,
+      "loss": 0.2166,
+      "step": 10102
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3781141069017556e-06,
+      "loss": 0.2065,
+      "step": 10103
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3763697919128727e-06,
+      "loss": 0.2311,
+      "step": 10104
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3746265284523508e-06,
+      "loss": 0.2501,
+      "step": 10105
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3728843166547422e-06,
+      "loss": 0.2556,
+      "step": 10106
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3711431566545129e-06,
+      "loss": 0.2099,
+      "step": 10107
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3694030485860587e-06,
+      "loss": 0.2442,
+      "step": 10108
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.367663992583687e-06,
+      "loss": 0.2098,
+      "step": 10109
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3659259887816206e-06,
+      "loss": 0.2396,
+      "step": 10110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.364189037314012e-06,
+      "loss": 0.2279,
+      "step": 10111
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3624531383149225e-06,
+      "loss": 0.2138,
+      "step": 10112
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3607182919183348e-06,
+      "loss": 0.2522,
+      "step": 10113
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3589844982581567e-06,
+      "loss": 0.2437,
+      "step": 10114
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3572517574682048e-06,
+      "loss": 0.1939,
+      "step": 10115
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3555200696822235e-06,
+      "loss": 0.2351,
+      "step": 10116
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3537894350338726e-06,
+      "loss": 0.2013,
+      "step": 10117
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3520598536567202e-06,
+      "loss": 0.2283,
+      "step": 10118
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.350331325684273e-06,
+      "loss": 0.2492,
+      "step": 10119
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.348603851249941e-06,
+      "loss": 0.2205,
+      "step": 10120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3468774304870556e-06,
+      "loss": 0.2384,
+      "step": 10121
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3451520635288756e-06,
+      "loss": 0.2511,
+      "step": 10122
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3434277505085673e-06,
+      "loss": 0.2317,
+      "step": 10123
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3417044915592197e-06,
+      "loss": 0.2243,
+      "step": 10124
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3399822868138462e-06,
+      "loss": 0.2296,
+      "step": 10125
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3382611364053705e-06,
+      "loss": 0.256,
+      "step": 10126
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.336541040466638e-06,
+      "loss": 0.2306,
+      "step": 10127
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3348219991304124e-06,
+      "loss": 0.2246,
+      "step": 10128
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3331040125293758e-06,
+      "loss": 0.2331,
+      "step": 10129
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3313870807961326e-06,
+      "loss": 0.2484,
+      "step": 10130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3296712040631998e-06,
+      "loss": 0.2386,
+      "step": 10131
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3279563824630164e-06,
+      "loss": 0.2566,
+      "step": 10132
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3262426161279417e-06,
+      "loss": 0.2228,
+      "step": 10133
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3245299051902499e-06,
+      "loss": 0.2463,
+      "step": 10134
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.322818249782135e-06,
+      "loss": 0.1973,
+      "step": 10135
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.32110765003571e-06,
+      "loss": 0.2551,
+      "step": 10136
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3193981060830024e-06,
+      "loss": 0.2353,
+      "step": 10137
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3176896180559666e-06,
+      "loss": 0.2292,
+      "step": 10138
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3159821860864708e-06,
+      "loss": 0.2443,
+      "step": 10139
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3142758103062975e-06,
+      "loss": 0.2319,
+      "step": 10140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3125704908471581e-06,
+      "loss": 0.2163,
+      "step": 10141
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.310866227840674e-06,
+      "loss": 0.2059,
+      "step": 10142
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3091630214183819e-06,
+      "loss": 0.2299,
+      "step": 10143
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3074608717117497e-06,
+      "loss": 0.2405,
+      "step": 10144
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.305759778852149e-06,
+      "loss": 0.2478,
+      "step": 10145
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.304059742970885e-06,
+      "loss": 0.2253,
+      "step": 10146
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3023607641991692e-06,
+      "loss": 0.215,
+      "step": 10147
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.300662842668135e-06,
+      "loss": 0.2059,
+      "step": 10148
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2989659785088392e-06,
+      "loss": 0.2086,
+      "step": 10149
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.297270171852249e-06,
+      "loss": 0.1927,
+      "step": 10150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.295575422829256e-06,
+      "loss": 0.2122,
+      "step": 10151
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2938817315706674e-06,
+      "loss": 0.2405,
+      "step": 10152
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2921890982072088e-06,
+      "loss": 0.1992,
+      "step": 10153
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.290497522869522e-06,
+      "loss": 0.222,
+      "step": 10154
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2888070056881746e-06,
+      "loss": 0.2664,
+      "step": 10155
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2871175467936437e-06,
+      "loss": 0.2433,
+      "step": 10156
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2854291463163332e-06,
+      "loss": 0.2247,
+      "step": 10157
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.283741804386559e-06,
+      "loss": 0.2067,
+      "step": 10158
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.282055521134557e-06,
+      "loss": 0.2077,
+      "step": 10159
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2803702966904795e-06,
+      "loss": 0.2049,
+      "step": 10160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2786861311844028e-06,
+      "loss": 0.2052,
+      "step": 10161
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.277003024746311e-06,
+      "loss": 0.2148,
+      "step": 10162
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.275320977506122e-06,
+      "loss": 0.2418,
+      "step": 10163
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.27363998959366e-06,
+      "loss": 0.2198,
+      "step": 10164
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2719600611386666e-06,
+      "loss": 0.2448,
+      "step": 10165
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2702811922708114e-06,
+      "loss": 0.2446,
+      "step": 10166
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2686033831196736e-06,
+      "loss": 0.2358,
+      "step": 10167
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2669266338147534e-06,
+      "loss": 0.2074,
+      "step": 10168
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2652509444854705e-06,
+      "loss": 0.2629,
+      "step": 10169
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2635763152611578e-06,
+      "loss": 0.2515,
+      "step": 10170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2619027462710776e-06,
+      "loss": 0.2154,
+      "step": 10171
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.260230237644396e-06,
+      "loss": 0.2348,
+      "step": 10172
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2585587895102046e-06,
+      "loss": 0.2207,
+      "step": 10173
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.256888401997519e-06,
+      "loss": 0.2271,
+      "step": 10174
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.255219075235261e-06,
+      "loss": 0.1992,
+      "step": 10175
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2535508093522773e-06,
+      "loss": 0.2377,
+      "step": 10176
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.251883604477332e-06,
+      "loss": 0.2231,
+      "step": 10177
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2502174607391048e-06,
+      "loss": 0.2159,
+      "step": 10178
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2485523782661984e-06,
+      "loss": 0.2151,
+      "step": 10179
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2468883571871315e-06,
+      "loss": 0.2058,
+      "step": 10180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2452253976303364e-06,
+      "loss": 0.23,
+      "step": 10181
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.24356349972417e-06,
+      "loss": 0.2186,
+      "step": 10182
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2419026635969055e-06,
+      "loss": 0.231,
+      "step": 10183
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2402428893767315e-06,
+      "loss": 0.2201,
+      "step": 10184
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2385841771917555e-06,
+      "loss": 0.2001,
+      "step": 10185
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.236926527170002e-06,
+      "loss": 0.2282,
+      "step": 10186
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2352699394394201e-06,
+      "loss": 0.2299,
+      "step": 10187
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.233614414127871e-06,
+      "loss": 0.2334,
+      "step": 10188
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2319599513631308e-06,
+      "loss": 0.2595,
+      "step": 10189
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2303065512729022e-06,
+      "loss": 0.2015,
+      "step": 10190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2286542139848018e-06,
+      "loss": 0.245,
+      "step": 10191
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.227002939626362e-06,
+      "loss": 0.204,
+      "step": 10192
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.225352728325035e-06,
+      "loss": 0.2373,
+      "step": 10193
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2237035802081915e-06,
+      "loss": 0.2489,
+      "step": 10194
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2220554954031165e-06,
+      "loss": 0.2053,
+      "step": 10195
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.220408474037022e-06,
+      "loss": 0.2281,
+      "step": 10196
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.218762516237026e-06,
+      "loss": 0.222,
+      "step": 10197
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.217117622130175e-06,
+      "loss": 0.2886,
+      "step": 10198
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2154737918434277e-06,
+      "loss": 0.2178,
+      "step": 10199
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2138310255036612e-06,
+      "loss": 0.2502,
+      "step": 10200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2121893232376702e-06,
+      "loss": 0.2747,
+      "step": 10201
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2105486851721688e-06,
+      "loss": 0.2109,
+      "step": 10202
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2089091114337852e-06,
+      "loss": 0.2491,
+      "step": 10203
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2072706021490737e-06,
+      "loss": 0.2809,
+      "step": 10204
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.205633157444498e-06,
+      "loss": 0.2183,
+      "step": 10205
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.203996777446445e-06,
+      "loss": 0.2185,
+      "step": 10206
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2023614622812174e-06,
+      "loss": 0.2376,
+      "step": 10207
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.200727212075034e-06,
+      "loss": 0.1836,
+      "step": 10208
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1990940269540323e-06,
+      "loss": 0.2529,
+      "step": 10209
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1974619070442716e-06,
+      "loss": 0.2371,
+      "step": 10210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1958308524717194e-06,
+      "loss": 0.2083,
+      "step": 10211
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1942008633622747e-06,
+      "loss": 0.2446,
+      "step": 10212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1925719398417424e-06,
+      "loss": 0.2324,
+      "step": 10213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1909440820358497e-06,
+      "loss": 0.2338,
+      "step": 10214
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1893172900702449e-06,
+      "loss": 0.2166,
+      "step": 10215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.187691564070487e-06,
+      "loss": 0.2645,
+      "step": 10216
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1860669041620581e-06,
+      "loss": 0.2069,
+      "step": 10217
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1844433104703555e-06,
+      "loss": 0.2657,
+      "step": 10218
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.182820783120691e-06,
+      "loss": 0.2261,
+      "step": 10219
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.181199322238306e-06,
+      "loss": 0.2354,
+      "step": 10220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1795789279483472e-06,
+      "loss": 0.2449,
+      "step": 10221
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1779596003758808e-06,
+      "loss": 0.2081,
+      "step": 10222
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1763413396458988e-06,
+      "loss": 0.2154,
+      "step": 10223
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.174724145883303e-06,
+      "loss": 0.2238,
+      "step": 10224
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1731080192129134e-06,
+      "loss": 0.2502,
+      "step": 10225
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.171492959759472e-06,
+      "loss": 0.2286,
+      "step": 10226
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1698789676476323e-06,
+      "loss": 0.228,
+      "step": 10227
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1682660430019715e-06,
+      "loss": 0.2191,
+      "step": 10228
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.1666541859469832e-06,
+      "loss": 0.2105,
+      "step": 10229
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.165043396607073e-06,
+      "loss": 0.2219,
+      "step": 10230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.163433675106571e-06,
+      "loss": 0.2136,
+      "step": 10231
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1618250215697235e-06,
+      "loss": 0.2342,
+      "step": 10232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1602174361206925e-06,
+      "loss": 0.2406,
+      "step": 10233
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1586109188835553e-06,
+      "loss": 0.2152,
+      "step": 10234
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1570054699823095e-06,
+      "loss": 0.2266,
+      "step": 10235
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1554010895408746e-06,
+      "loss": 0.2202,
+      "step": 10236
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.153797777683081e-06,
+      "loss": 0.2039,
+      "step": 10237
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1521955345326768e-06,
+      "loss": 0.2102,
+      "step": 10238
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1505943602133345e-06,
+      "loss": 0.224,
+      "step": 10239
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1489942548486354e-06,
+      "loss": 0.2054,
+      "step": 10240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.147395218562084e-06,
+      "loss": 0.2236,
+      "step": 10241
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1457972514771015e-06,
+      "loss": 0.2222,
+      "step": 10242
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1442003537170226e-06,
+      "loss": 0.2377,
+      "step": 10243
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1426045254051038e-06,
+      "loss": 0.2175,
+      "step": 10244
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1410097666645181e-06,
+      "loss": 0.243,
+      "step": 10245
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1394160776183555e-06,
+      "loss": 0.2449,
+      "step": 10246
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1378234583896257e-06,
+      "loss": 0.206,
+      "step": 10247
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1362319091012508e-06,
+      "loss": 0.2219,
+      "step": 10248
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1346414298760738e-06,
+      "loss": 0.2416,
+      "step": 10249
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1330520208368567e-06,
+      "loss": 0.2154,
+      "step": 10250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1314636821062728e-06,
+      "loss": 0.2277,
+      "step": 10251
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.129876413806918e-06,
+      "loss": 0.2135,
+      "step": 10252
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1282902160613052e-06,
+      "loss": 0.2137,
+      "step": 10253
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1267050889918652e-06,
+      "loss": 0.2214,
+      "step": 10254
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.12512103272094e-06,
+      "loss": 0.2389,
+      "step": 10255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1235380473707984e-06,
+      "loss": 0.2189,
+      "step": 10256
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1219561330636196e-06,
+      "loss": 0.2004,
+      "step": 10257
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1203752899215042e-06,
+      "loss": 0.2413,
+      "step": 10258
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.118795518066466e-06,
+      "loss": 0.245,
+      "step": 10259
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1172168176204362e-06,
+      "loss": 0.2497,
+      "step": 10260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1156391887052718e-06,
+      "loss": 0.2336,
+      "step": 10261
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.114062631442736e-06,
+      "loss": 0.2232,
+      "step": 10262
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1124871459545127e-06,
+      "loss": 0.2391,
+      "step": 10263
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1109127323622115e-06,
+      "loss": 0.2288,
+      "step": 10264
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.109339390787345e-06,
+      "loss": 0.2272,
+      "step": 10265
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1077671213513546e-06,
+      "loss": 0.2106,
+      "step": 10266
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.106195924175593e-06,
+      "loss": 0.2338,
+      "step": 10267
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1046257993813286e-06,
+      "loss": 0.2307,
+      "step": 10268
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1030567470897556e-06,
+      "loss": 0.2202,
+      "step": 10269
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.1014887674219776e-06,
+      "loss": 0.2021,
+      "step": 10270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0999218604990158e-06,
+      "loss": 0.2416,
+      "step": 10271
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0983560264418152e-06,
+      "loss": 0.2008,
+      "step": 10272
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0967912653712291e-06,
+      "loss": 0.2293,
+      "step": 10273
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0952275774080362e-06,
+      "loss": 0.202,
+      "step": 10274
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0936649626729244e-06,
+      "loss": 0.2018,
+      "step": 10275
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0921034212865027e-06,
+      "loss": 0.2111,
+      "step": 10276
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0905429533693011e-06,
+      "loss": 0.2263,
+      "step": 10277
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0889835590417603e-06,
+      "loss": 0.2274,
+      "step": 10278
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.08742523842424e-06,
+      "loss": 0.2154,
+      "step": 10279
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0858679916370229e-06,
+      "loss": 0.2544,
+      "step": 10280
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0843118188002992e-06,
+      "loss": 0.227,
+      "step": 10281
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0827567200341815e-06,
+      "loss": 0.2152,
+      "step": 10282
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0812026954587002e-06,
+      "loss": 0.2287,
+      "step": 10283
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0796497451937992e-06,
+      "loss": 0.2245,
+      "step": 10284
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0780978693593413e-06,
+      "loss": 0.2349,
+      "step": 10285
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0765470680751105e-06,
+      "loss": 0.2351,
+      "step": 10286
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0749973414607979e-06,
+      "loss": 0.1834,
+      "step": 10287
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0734486896360257e-06,
+      "loss": 0.2091,
+      "step": 10288
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.0719011127203205e-06,
+      "loss": 0.2607,
+      "step": 10289
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.070354610833133e-06,
+      "loss": 0.237,
+      "step": 10290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0688091840938264e-06,
+      "loss": 0.2298,
+      "step": 10291
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0672648326216833e-06,
+      "loss": 0.2167,
+      "step": 10292
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.065721556535902e-06,
+      "loss": 0.2278,
+      "step": 10293
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0641793559556017e-06,
+      "loss": 0.216,
+      "step": 10294
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0626382309998146e-06,
+      "loss": 0.2308,
+      "step": 10295
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0610981817874916e-06,
+      "loss": 0.215,
+      "step": 10296
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.059559208437501e-06,
+      "loss": 0.213,
+      "step": 10297
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0580213110686266e-06,
+      "loss": 0.258,
+      "step": 10298
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0564844897995679e-06,
+      "loss": 0.2399,
+      "step": 10299
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0549487447489453e-06,
+      "loss": 0.2162,
+      "step": 10300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0534140760352905e-06,
+      "loss": 0.2082,
+      "step": 10301
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0518804837770607e-06,
+      "loss": 0.2016,
+      "step": 10302
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0503479680926225e-06,
+      "loss": 0.2171,
+      "step": 10303
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0488165291002584e-06,
+      "loss": 0.2079,
+      "step": 10304
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0472861669181783e-06,
+      "loss": 0.2252,
+      "step": 10305
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.045756881664498e-06,
+      "loss": 0.2144,
+      "step": 10306
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0442286734572532e-06,
+      "loss": 0.2313,
+      "step": 10307
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0427015424143993e-06,
+      "loss": 0.2194,
+      "step": 10308
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0411754886538022e-06,
+      "loss": 0.2359,
+      "step": 10309
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.039650512293256e-06,
+      "loss": 0.2576,
+      "step": 10310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0381266134504614e-06,
+      "loss": 0.1849,
+      "step": 10311
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.036603792243036e-06,
+      "loss": 0.2863,
+      "step": 10312
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0350820487885226e-06,
+      "loss": 0.2084,
+      "step": 10313
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0335613832043756e-06,
+      "loss": 0.2085,
+      "step": 10314
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0320417956079625e-06,
+      "loss": 0.2211,
+      "step": 10315
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.030523286116573e-06,
+      "loss": 0.1963,
+      "step": 10316
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0290058548474119e-06,
+      "loss": 0.2587,
+      "step": 10317
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0274895019176016e-06,
+      "loss": 0.2265,
+      "step": 10318
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0259742274441807e-06,
+      "loss": 0.243,
+      "step": 10319
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0244600315441023e-06,
+      "loss": 0.2051,
+      "step": 10320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0229469143342408e-06,
+      "loss": 0.2173,
+      "step": 10321
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0214348759313846e-06,
+      "loss": 0.2487,
+      "step": 10322
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0199239164522388e-06,
+      "loss": 0.2142,
+      "step": 10323
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0184140360134237e-06,
+      "loss": 0.2004,
+      "step": 10324
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.016905234731479e-06,
+      "loss": 0.227,
+      "step": 10325
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0153975127228614e-06,
+      "loss": 0.2437,
+      "step": 10326
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0138908701039435e-06,
+      "loss": 0.2166,
+      "step": 10327
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.01238530699101e-06,
+      "loss": 0.2173,
+      "step": 10328
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.010880823500273e-06,
+      "loss": 0.2394,
+      "step": 10329
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0093774197478517e-06,
+      "loss": 0.2367,
+      "step": 10330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0078750958497846e-06,
+      "loss": 0.2329,
+      "step": 10331
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0063738519220273e-06,
+      "loss": 0.2647,
+      "step": 10332
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0048736880804522e-06,
+      "loss": 0.2239,
+      "step": 10333
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0033746044408465e-06,
+      "loss": 0.2089,
+      "step": 10334
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0018766011189196e-06,
+      "loss": 0.2211,
+      "step": 10335
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.0003796782302887e-06,
+      "loss": 0.1981,
+      "step": 10336
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.988838358904984e-07,
+      "loss": 0.219,
+      "step": 10337
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.97389074215001e-07,
+      "loss": 0.241,
+      "step": 10338
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.958953933191677e-07,
+      "loss": 0.1979,
+      "step": 10339
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.944027933182881e-07,
+      "loss": 0.2411,
+      "step": 10340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.929112743275664e-07,
+      "loss": 0.2222,
+      "step": 10341
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.914208364621242e-07,
+      "loss": 0.2376,
+      "step": 10342
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.899314798370007e-07,
+      "loss": 0.2021,
+      "step": 10343
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.88443204567151e-07,
+      "loss": 0.2472,
+      "step": 10344
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.869560107674414e-07,
+      "loss": 0.2628,
+      "step": 10345
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.854698985526682e-07,
+      "loss": 0.2411,
+      "step": 10346
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.839848680375286e-07,
+      "loss": 0.2085,
+      "step": 10347
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.825009193366502e-07,
+      "loss": 0.2388,
+      "step": 10348
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.810180525645634e-07,
+      "loss": 0.2444,
+      "step": 10349
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.795362678357234e-07,
+      "loss": 0.2744,
+      "step": 10350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.780555652645035e-07,
+      "loss": 0.1954,
+      "step": 10351
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.76575944965189e-07,
+      "loss": 0.2296,
+      "step": 10352
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.75097407051982e-07,
+      "loss": 0.1933,
+      "step": 10353
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.736199516390043e-07,
+      "loss": 0.2097,
+      "step": 10354
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.721435788402915e-07,
+      "loss": 0.2407,
+      "step": 10355
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.70668288769796e-07,
+      "loss": 0.2217,
+      "step": 10356
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.691940815413864e-07,
+      "loss": 0.241,
+      "step": 10357
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.67720957268845e-07,
+      "loss": 0.2125,
+      "step": 10358
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.662489160658794e-07,
+      "loss": 0.2225,
+      "step": 10359
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.647779580461052e-07,
+      "loss": 0.2304,
+      "step": 10360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.633080833230551e-07,
+      "loss": 0.2238,
+      "step": 10361
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.618392920101848e-07,
+      "loss": 0.2175,
+      "step": 10362
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.603715842208604e-07,
+      "loss": 0.2286,
+      "step": 10363
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.58904960068363e-07,
+      "loss": 0.2436,
+      "step": 10364
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.574394196658936e-07,
+      "loss": 0.205,
+      "step": 10365
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.559749631265668e-07,
+      "loss": 0.2332,
+      "step": 10366
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.545115905634222e-07,
+      "loss": 0.2028,
+      "step": 10367
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.530493020894027e-07,
+      "loss": 0.2222,
+      "step": 10368
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.515880978173747e-07,
+      "loss": 0.2268,
+      "step": 10369
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.50127977860123e-07,
+      "loss": 0.2631,
+      "step": 10370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.486689423303424e-07,
+      "loss": 0.2108,
+      "step": 10371
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.472109913406479e-07,
+      "loss": 0.2363,
+      "step": 10372
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.457541250035762e-07,
+      "loss": 0.2324,
+      "step": 10373
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.442983434315655e-07,
+      "loss": 0.2655,
+      "step": 10374
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.428436467369811e-07,
+      "loss": 0.2141,
+      "step": 10375
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.413900350321064e-07,
+      "loss": 0.204,
+      "step": 10376
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.399375084291318e-07,
+      "loss": 0.2024,
+      "step": 10377
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.384860670401757e-07,
+      "loss": 0.232,
+      "step": 10378
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.370357109772621e-07,
+      "loss": 0.2494,
+      "step": 10379
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.355864403523345e-07,
+      "loss": 0.268,
+      "step": 10380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.341382552772621e-07,
+      "loss": 0.2455,
+      "step": 10381
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.326911558638101e-07,
+      "loss": 0.2346,
+      "step": 10382
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.312451422236761e-07,
+      "loss": 0.2555,
+      "step": 10383
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.298002144684725e-07,
+      "loss": 0.2788,
+      "step": 10384
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.283563727097233e-07,
+      "loss": 0.2129,
+      "step": 10385
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.269136170588661e-07,
+      "loss": 0.2395,
+      "step": 10386
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.254719476272649e-07,
+      "loss": 0.1955,
+      "step": 10387
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.240313645261894e-07,
+      "loss": 0.2243,
+      "step": 10388
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.225918678668338e-07,
+      "loss": 0.2218,
+      "step": 10389
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.211534577603026e-07,
+      "loss": 0.2202,
+      "step": 10390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.197161343176136e-07,
+      "loss": 0.2167,
+      "step": 10391
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.182798976497131e-07,
+      "loss": 0.2265,
+      "step": 10392
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.168447478674507e-07,
+      "loss": 0.2517,
+      "step": 10393
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.154106850815979e-07,
+      "loss": 0.2277,
+      "step": 10394
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.139777094028428e-07,
+      "loss": 0.2154,
+      "step": 10395
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.12545820941787e-07,
+      "loss": 0.2114,
+      "step": 10396
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.111150198089535e-07,
+      "loss": 0.2728,
+      "step": 10397
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.096853061147758e-07,
+      "loss": 0.2324,
+      "step": 10398
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.082566799695991e-07,
+      "loss": 0.25,
+      "step": 10399
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.068291414836983e-07,
+      "loss": 0.2259,
+      "step": 10400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.054026907672535e-07,
+      "loss": 0.2289,
+      "step": 10401
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.039773279303637e-07,
+      "loss": 0.2369,
+      "step": 10402
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.025530530830472e-07,
+      "loss": 0.23,
+      "step": 10403
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.011298663352329e-07,
+      "loss": 0.2078,
+      "step": 10404
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.997077677967664e-07,
+      "loss": 0.2109,
+      "step": 10405
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.982867575774162e-07,
+      "loss": 0.1951,
+      "step": 10406
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.968668357868582e-07,
+      "loss": 0.1905,
+      "step": 10407
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.954480025346911e-07,
+      "loss": 0.2355,
+      "step": 10408
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.940302579304227e-07,
+      "loss": 0.2055,
+      "step": 10409
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.926136020834785e-07,
+      "loss": 0.2312,
+      "step": 10410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.911980351032095e-07,
+      "loss": 0.2237,
+      "step": 10411
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.897835570988682e-07,
+      "loss": 0.2341,
+      "step": 10412
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.883701681796324e-07,
+      "loss": 0.2169,
+      "step": 10413
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.869578684545948e-07,
+      "loss": 0.2979,
+      "step": 10414
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.855466580327631e-07,
+      "loss": 0.233,
+      "step": 10415
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.841365370230536e-07,
+      "loss": 0.2738,
+      "step": 10416
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.827275055343143e-07,
+      "loss": 0.2356,
+      "step": 10417
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.813195636752914e-07,
+      "loss": 0.2611,
+      "step": 10418
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.799127115546629e-07,
+      "loss": 0.2013,
+      "step": 10419
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.785069492810121e-07,
+      "loss": 0.1971,
+      "step": 10420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.771022769628423e-07,
+      "loss": 0.2237,
+      "step": 10421
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.756986947085716e-07,
+      "loss": 0.2056,
+      "step": 10422
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.742962026265383e-07,
+      "loss": 0.2303,
+      "step": 10423
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.728948008249843e-07,
+      "loss": 0.2251,
+      "step": 10424
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.714944894120813e-07,
+      "loss": 0.2131,
+      "step": 10425
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.70095268495908e-07,
+      "loss": 0.2627,
+      "step": 10426
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.686971381844677e-07,
+      "loss": 0.2177,
+      "step": 10427
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.673000985856694e-07,
+      "loss": 0.1939,
+      "step": 10428
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.6590414980734e-07,
+      "loss": 0.2362,
+      "step": 10429
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.645092919572317e-07,
+      "loss": 0.243,
+      "step": 10430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.631155251430017e-07,
+      "loss": 0.1972,
+      "step": 10431
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.617228494722257e-07,
+      "loss": 0.2205,
+      "step": 10432
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.603312650523975e-07,
+      "loss": 0.2297,
+      "step": 10433
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.589407719909248e-07,
+      "loss": 0.2459,
+      "step": 10434
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.575513703951299e-07,
+      "loss": 0.2169,
+      "step": 10435
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.56163060372257e-07,
+      "loss": 0.2335,
+      "step": 10436
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.547758420294571e-07,
+      "loss": 0.2054,
+      "step": 10437
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.533897154738046e-07,
+      "loss": 0.205,
+      "step": 10438
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.520046808122855e-07,
+      "loss": 0.2249,
+      "step": 10439
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.506207381518044e-07,
+      "loss": 0.2346,
+      "step": 10440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.492378875991757e-07,
+      "loss": 0.2277,
+      "step": 10441
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.478561292611376e-07,
+      "loss": 0.2352,
+      "step": 10442
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.464754632443345e-07,
+      "loss": 0.2619,
+      "step": 10443
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.450958896553379e-07,
+      "loss": 0.2055,
+      "step": 10444
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.437174086006261e-07,
+      "loss": 0.2087,
+      "step": 10445
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.423400201865938e-07,
+      "loss": 0.1936,
+      "step": 10446
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.409637245195595e-07,
+      "loss": 0.2218,
+      "step": 10447
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.395885217057481e-07,
+      "loss": 0.2103,
+      "step": 10448
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.38214411851303e-07,
+      "loss": 0.1962,
+      "step": 10449
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.368413950622844e-07,
+      "loss": 0.2158,
+      "step": 10450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.354694714446659e-07,
+      "loss": 0.271,
+      "step": 10451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.34098641104341e-07,
+      "loss": 0.2344,
+      "step": 10452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.327289041471136e-07,
+      "loss": 0.2478,
+      "step": 10453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.313602606787057e-07,
+      "loss": 0.2228,
+      "step": 10454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.299927108047578e-07,
+      "loss": 0.242,
+      "step": 10455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.286262546308221e-07,
+      "loss": 0.2059,
+      "step": 10456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.272608922623659e-07,
+      "loss": 0.2639,
+      "step": 10457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.258966238047749e-07,
+      "loss": 0.202,
+      "step": 10458
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.24533449363345e-07,
+      "loss": 0.2373,
+      "step": 10459
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.231713690432969e-07,
+      "loss": 0.2306,
+      "step": 10460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.218103829497597e-07,
+      "loss": 0.2036,
+      "step": 10461
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.20450491187778e-07,
+      "loss": 0.2117,
+      "step": 10462
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.190916938623177e-07,
+      "loss": 0.2133,
+      "step": 10463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.17733991078255e-07,
+      "loss": 0.229,
+      "step": 10464
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.163773829403826e-07,
+      "loss": 0.207,
+      "step": 10465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.150218695534084e-07,
+      "loss": 0.2167,
+      "step": 10466
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.136674510219555e-07,
+      "loss": 0.2205,
+      "step": 10467
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.123141274505685e-07,
+      "loss": 0.2345,
+      "step": 10468
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.109618989436973e-07,
+      "loss": 0.2705,
+      "step": 10469
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.096107656057133e-07,
+      "loss": 0.1916,
+      "step": 10470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.082607275409066e-07,
+      "loss": 0.2369,
+      "step": 10471
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.069117848534752e-07,
+      "loss": 0.1971,
+      "step": 10472
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.055639376475376e-07,
+      "loss": 0.2764,
+      "step": 10473
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.042171860271242e-07,
+      "loss": 0.1992,
+      "step": 10474
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.028715300961848e-07,
+      "loss": 0.2432,
+      "step": 10475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.015269699585797e-07,
+      "loss": 0.21,
+      "step": 10476
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.001835057180928e-07,
+      "loss": 0.2496,
+      "step": 10477
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.988411374784127e-07,
+      "loss": 0.2031,
+      "step": 10478
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.974998653431548e-07,
+      "loss": 0.2624,
+      "step": 10479
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.961596894158396e-07,
+      "loss": 0.2555,
+      "step": 10480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.948206097999094e-07,
+      "loss": 0.2635,
+      "step": 10481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.934826265987199e-07,
+      "loss": 0.222,
+      "step": 10482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.921457399155418e-07,
+      "loss": 0.1963,
+      "step": 10483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.908099498535576e-07,
+      "loss": 0.2362,
+      "step": 10484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.894752565158764e-07,
+      "loss": 0.2438,
+      "step": 10485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.881416600055109e-07,
+      "loss": 0.1963,
+      "step": 10486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.868091604253952e-07,
+      "loss": 0.2525,
+      "step": 10487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.854777578783756e-07,
+      "loss": 0.2189,
+      "step": 10488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.841474524672182e-07,
+      "loss": 0.2842,
+      "step": 10489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.828182442945992e-07,
+      "loss": 0.2211,
+      "step": 10490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.814901334631131e-07,
+      "loss": 0.2232,
+      "step": 10491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.801631200752646e-07,
+      "loss": 0.2115,
+      "step": 10492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.788372042334868e-07,
+      "loss": 0.2704,
+      "step": 10493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.775123860401128e-07,
+      "loss": 0.2078,
+      "step": 10494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.761886655973977e-07,
+      "loss": 0.2189,
+      "step": 10495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.748660430075178e-07,
+      "loss": 0.2325,
+      "step": 10496
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.735445183725515e-07,
+      "loss": 0.2114,
+      "step": 10497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.722240917945039e-07,
+      "loss": 0.2441,
+      "step": 10498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.709047633752902e-07,
+      "loss": 0.2379,
+      "step": 10499
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.695865332167374e-07,
+      "loss": 0.2186,
+      "step": 10500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.682694014205987e-07,
+      "loss": 0.2294,
+      "step": 10501
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.669533680885332e-07,
+      "loss": 0.2323,
+      "step": 10502
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.656384333221145e-07,
+      "loss": 0.3199,
+      "step": 10503
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.643245972228397e-07,
+      "loss": 0.2066,
+      "step": 10504
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.630118598921126e-07,
+      "loss": 0.2026,
+      "step": 10505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.617002214312591e-07,
+      "loss": 0.2462,
+      "step": 10506
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.603896819415146e-07,
+      "loss": 0.1993,
+      "step": 10507
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.590802415240283e-07,
+      "loss": 0.2013,
+      "step": 10508
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.577719002798761e-07,
+      "loss": 0.2182,
+      "step": 10509
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.564646583100371e-07,
+      "loss": 0.2528,
+      "step": 10510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.551585157154073e-07,
+      "loss": 0.2503,
+      "step": 10511
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.538534725968061e-07,
+      "loss": 0.2437,
+      "step": 10512
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.525495290549578e-07,
+      "loss": 0.2578,
+      "step": 10513
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.51246685190507e-07,
+      "loss": 0.2102,
+      "step": 10514
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.499449411040132e-07,
+      "loss": 0.2604,
+      "step": 10515
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.486442968959495e-07,
+      "loss": 0.2711,
+      "step": 10516
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.473447526667038e-07,
+      "loss": 0.1944,
+      "step": 10517
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.460463085165842e-07,
+      "loss": 0.2215,
+      "step": 10518
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.447489645458056e-07,
+      "loss": 0.2374,
+      "step": 10519
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.434527208545078e-07,
+      "loss": 0.2305,
+      "step": 10520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.421575775427359e-07,
+      "loss": 0.2671,
+      "step": 10521
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.408635347104564e-07,
+      "loss": 0.2239,
+      "step": 10522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.39570592457548e-07,
+      "loss": 0.2064,
+      "step": 10523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.382787508838057e-07,
+      "loss": 0.2425,
+      "step": 10524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.369880100889365e-07,
+      "loss": 0.2527,
+      "step": 10525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.356983701725706e-07,
+      "loss": 0.2268,
+      "step": 10526
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.344098312342435e-07,
+      "loss": 0.2054,
+      "step": 10527
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.331223933734122e-07,
+      "loss": 0.2083,
+      "step": 10528
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.318360566894455e-07,
+      "loss": 0.231,
+      "step": 10529
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.305508212816275e-07,
+      "loss": 0.2008,
+      "step": 10530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.292666872491605e-07,
+      "loss": 0.2489,
+      "step": 10531
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.279836546911567e-07,
+      "loss": 0.2001,
+      "step": 10532
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.267017237066454e-07,
+      "loss": 0.256,
+      "step": 10533
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.25420894394574e-07,
+      "loss": 0.2184,
+      "step": 10534
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.241411668538017e-07,
+      "loss": 0.2906,
+      "step": 10535
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228625411830997e-07,
+      "loss": 0.2281,
+      "step": 10536
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215850174811639e-07,
+      "loss": 0.225,
+      "step": 10537
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.203085958465938e-07,
+      "loss": 0.2385,
+      "step": 10538
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.190332763779122e-07,
+      "loss": 0.2176,
+      "step": 10539
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.177590591735506e-07,
+      "loss": 0.2248,
+      "step": 10540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.164859443318583e-07,
+      "loss": 0.2288,
+      "step": 10541
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.152139319511019e-07,
+      "loss": 0.2635,
+      "step": 10542
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.139430221294613e-07,
+      "loss": 0.2603,
+      "step": 10543
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.126732149650261e-07,
+      "loss": 0.2255,
+      "step": 10544
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.114045105558098e-07,
+      "loss": 0.2724,
+      "step": 10545
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.101369089997356e-07,
+      "loss": 0.202,
+      "step": 10546
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.088704103946403e-07,
+      "loss": 0.2172,
+      "step": 10547
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.07605014838279e-07,
+      "loss": 0.2232,
+      "step": 10548
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.06340722428317e-07,
+      "loss": 0.2064,
+      "step": 10549
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.050775332623427e-07,
+      "loss": 0.2273,
+      "step": 10550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.038154474378533e-07,
+      "loss": 0.2396,
+      "step": 10551
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.025544650522558e-07,
+      "loss": 0.2115,
+      "step": 10552
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.012945862028875e-07,
+      "loss": 0.2054,
+      "step": 10553
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.000358109869837e-07,
+      "loss": 0.2082,
+      "step": 10554
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.987781395017068e-07,
+      "loss": 0.1918,
+      "step": 10555
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.975215718441258e-07,
+      "loss": 0.2272,
+      "step": 10556
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.962661081112265e-07,
+      "loss": 0.2044,
+      "step": 10557
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.950117483999147e-07,
+      "loss": 0.2382,
+      "step": 10558
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.937584928070062e-07,
+      "loss": 0.2357,
+      "step": 10559
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.925063414292304e-07,
+      "loss": 0.2444,
+      "step": 10560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.91255294363235e-07,
+      "loss": 0.2216,
+      "step": 10561
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.900053517055827e-07,
+      "loss": 0.2211,
+      "step": 10562
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.887565135527463e-07,
+      "loss": 0.2033,
+      "step": 10563
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.875087800011187e-07,
+      "loss": 0.2082,
+      "step": 10564
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.86262151147003e-07,
+      "loss": 0.2046,
+      "step": 10565
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.850166270866187e-07,
+      "loss": 0.2036,
+      "step": 10566
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.83772207916104e-07,
+      "loss": 0.2369,
+      "step": 10567
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.825288937315038e-07,
+      "loss": 0.2178,
+      "step": 10568
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.812866846287862e-07,
+      "loss": 0.2275,
+      "step": 10569
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.800455807038298e-07,
+      "loss": 0.1949,
+      "step": 10570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.788055820524258e-07,
+      "loss": 0.2201,
+      "step": 10571
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.775666887702847e-07,
+      "loss": 0.2291,
+      "step": 10572
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.763289009530282e-07,
+      "loss": 0.2228,
+      "step": 10573
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.750922186961917e-07,
+      "loss": 0.236,
+      "step": 10574
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.738566420952319e-07,
+      "loss": 0.2128,
+      "step": 10575
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.726221712455144e-07,
+      "loss": 0.2391,
+      "step": 10576
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.713888062423162e-07,
+      "loss": 0.2082,
+      "step": 10577
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.701565471808413e-07,
+      "loss": 0.2307,
+      "step": 10578
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.689253941561952e-07,
+      "loss": 0.2387,
+      "step": 10579
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.67695347263405e-07,
+      "loss": 0.234,
+      "step": 10580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.664664065974119e-07,
+      "loss": 0.218,
+      "step": 10581
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.652385722530663e-07,
+      "loss": 0.2075,
+      "step": 10582
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.640118443251442e-07,
+      "loss": 0.2082,
+      "step": 10583
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.627862229083248e-07,
+      "loss": 0.2352,
+      "step": 10584
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.615617080972059e-07,
+      "loss": 0.1945,
+      "step": 10585
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.603382999863033e-07,
+      "loss": 0.21,
+      "step": 10586
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.591159986700468e-07,
+      "loss": 0.1883,
+      "step": 10587
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.57894804242774e-07,
+      "loss": 0.2178,
+      "step": 10588
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.566747167987447e-07,
+      "loss": 0.2335,
+      "step": 10589
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.554557364321267e-07,
+      "loss": 0.2495,
+      "step": 10590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.542378632370099e-07,
+      "loss": 0.2119,
+      "step": 10591
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.530210973073941e-07,
+      "loss": 0.2305,
+      "step": 10592
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.518054387371924e-07,
+      "loss": 0.2078,
+      "step": 10593
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.505908876202382e-07,
+      "loss": 0.1863,
+      "step": 10594
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.493774440502714e-07,
+      "loss": 0.2194,
+      "step": 10595
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.481651081209522e-07,
+      "loss": 0.2204,
+      "step": 10596
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.469538799258556e-07,
+      "loss": 0.2261,
+      "step": 10597
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.457437595584653e-07,
+      "loss": 0.2111,
+      "step": 10598
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.445347471121865e-07,
+      "loss": 0.2359,
+      "step": 10599
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.433268426803345e-07,
+      "loss": 0.2647,
+      "step": 10600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.421200463561399e-07,
+      "loss": 0.2614,
+      "step": 10601
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.409143582327498e-07,
+      "loss": 0.253,
+      "step": 10602
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.397097784032246e-07,
+      "loss": 0.2421,
+      "step": 10603
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.385063069605384e-07,
+      "loss": 0.2033,
+      "step": 10604
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.373039439975786e-07,
+      "loss": 0.2029,
+      "step": 10605
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.361026896071492e-07,
+      "loss": 0.2411,
+      "step": 10606
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.349025438819661e-07,
+      "loss": 0.2614,
+      "step": 10607
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.337035069146668e-07,
+      "loss": 0.2284,
+      "step": 10608
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.325055787977907e-07,
+      "loss": 0.2104,
+      "step": 10609
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.313087596238071e-07,
+      "loss": 0.2215,
+      "step": 10610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.301130494850854e-07,
+      "loss": 0.2119,
+      "step": 10611
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.289184484739186e-07,
+      "loss": 0.2646,
+      "step": 10612
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.277249566825094e-07,
+      "loss": 0.2109,
+      "step": 10613
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.265325742029776e-07,
+      "loss": 0.2073,
+      "step": 10614
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.253413011273529e-07,
+      "loss": 0.2443,
+      "step": 10615
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.241511375475867e-07,
+      "loss": 0.2243,
+      "step": 10616
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.229620835555371e-07,
+      "loss": 0.1901,
+      "step": 10617
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.21774139242986e-07,
+      "loss": 0.2875,
+      "step": 10618
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.205873047016198e-07,
+      "loss": 0.2484,
+      "step": 10619
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.194015800230435e-07,
+      "loss": 0.2144,
+      "step": 10620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.182169652987757e-07,
+      "loss": 0.2349,
+      "step": 10621
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.170334606202515e-07,
+      "loss": 0.2079,
+      "step": 10622
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.158510660788164e-07,
+      "loss": 0.2315,
+      "step": 10623
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.14669781765737e-07,
+      "loss": 0.2389,
+      "step": 10624
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.134896077721858e-07,
+      "loss": 0.2359,
+      "step": 10625
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.123105441892513e-07,
+      "loss": 0.2205,
+      "step": 10626
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.111325911079457e-07,
+      "loss": 0.199,
+      "step": 10627
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.099557486191815e-07,
+      "loss": 0.2334,
+      "step": 10628
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.087800168137991e-07,
+      "loss": 0.2138,
+      "step": 10629
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.076053957825411e-07,
+      "loss": 0.2443,
+      "step": 10630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.064318856160683e-07,
+      "loss": 0.1968,
+      "step": 10631
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.052594864049615e-07,
+      "loss": 0.2305,
+      "step": 10632
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.0408819823971e-07,
+      "loss": 0.2243,
+      "step": 10633
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.029180212107183e-07,
+      "loss": 0.2422,
+      "step": 10634
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.017489554083056e-07,
+      "loss": 0.1953,
+      "step": 10635
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.005810009227048e-07,
+      "loss": 0.2121,
+      "step": 10636
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.994141578440637e-07,
+      "loss": 0.1786,
+      "step": 10637
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.982484262624455e-07,
+      "loss": 0.2385,
+      "step": 10638
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.97083806267823e-07,
+      "loss": 0.2468,
+      "step": 10639
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.959202979500894e-07,
+      "loss": 0.2801,
+      "step": 10640
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.947579013990479e-07,
+      "loss": 0.2583,
+      "step": 10641
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.935966167044166e-07,
+      "loss": 0.2026,
+      "step": 10642
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.924364439558288e-07,
+      "loss": 0.2306,
+      "step": 10643
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.912773832428331e-07,
+      "loss": 0.2116,
+      "step": 10644
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.901194346548894e-07,
+      "loss": 0.2575,
+      "step": 10645
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.889625982813713e-07,
+      "loss": 0.2268,
+      "step": 10646
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.878068742115689e-07,
+      "loss": 0.2528,
+      "step": 10647
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.866522625346876e-07,
+      "loss": 0.2388,
+      "step": 10648
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.854987633398446e-07,
+      "loss": 0.2388,
+      "step": 10649
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.843463767160701e-07,
+      "loss": 0.2164,
+      "step": 10650
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.831951027523114e-07,
+      "loss": 0.2254,
+      "step": 10651
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.82044941537429e-07,
+      "loss": 0.2049,
+      "step": 10652
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.808958931601955e-07,
+      "loss": 0.2187,
+      "step": 10653
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.797479577093029e-07,
+      "loss": 0.2609,
+      "step": 10654
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.786011352733506e-07,
+      "loss": 0.2102,
+      "step": 10655
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.77455425940851e-07,
+      "loss": 0.2009,
+      "step": 10656
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.763108298002434e-07,
+      "loss": 0.2196,
+      "step": 10657
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.751673469398655e-07,
+      "loss": 0.2255,
+      "step": 10658
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.7402497744798e-07,
+      "loss": 0.2312,
+      "step": 10659
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.728837214127597e-07,
+      "loss": 0.2336,
+      "step": 10660
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.717435789222874e-07,
+      "loss": 0.2353,
+      "step": 10661
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.706045500645712e-07,
+      "loss": 0.2125,
+      "step": 10662
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.694666349275207e-07,
+      "loss": 0.215,
+      "step": 10663
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.683298335989623e-07,
+      "loss": 0.2203,
+      "step": 10664
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.671941461666458e-07,
+      "loss": 0.1957,
+      "step": 10665
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.660595727182261e-07,
+      "loss": 0.2263,
+      "step": 10666
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.649261133412714e-07,
+      "loss": 0.2208,
+      "step": 10667
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.637937681232702e-07,
+      "loss": 0.2312,
+      "step": 10668
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.626625371516175e-07,
+      "loss": 0.281,
+      "step": 10669
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.615324205136319e-07,
+      "loss": 0.186,
+      "step": 10670
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.604034182965401e-07,
+      "loss": 0.2253,
+      "step": 10671
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.592755305874758e-07,
+      "loss": 0.2332,
+      "step": 10672
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.581487574735011e-07,
+      "loss": 0.2239,
+      "step": 10673
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.570230990415847e-07,
+      "loss": 0.2097,
+      "step": 10674
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.558985553786039e-07,
+      "loss": 0.2046,
+      "step": 10675
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.547751265713641e-07,
+      "loss": 0.2068,
+      "step": 10676
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.536528127065693e-07,
+      "loss": 0.2221,
+      "step": 10677
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.52531613870847e-07,
+      "loss": 0.207,
+      "step": 10678
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.514115301507378e-07,
+      "loss": 0.1835,
+      "step": 10679
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.502925616326893e-07,
+      "loss": 0.2164,
+      "step": 10680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.491747084030724e-07,
+      "loss": 0.2794,
+      "step": 10681
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.480579705481681e-07,
+      "loss": 0.2049,
+      "step": 10682
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.469423481541658e-07,
+      "loss": 0.2315,
+      "step": 10683
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.458278413071799e-07,
+      "loss": 0.207,
+      "step": 10684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.447144500932316e-07,
+      "loss": 0.2248,
+      "step": 10685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.436021745982522e-07,
+      "loss": 0.2446,
+      "step": 10686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.424910149080964e-07,
+      "loss": 0.2187,
+      "step": 10687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.413809711085272e-07,
+      "loss": 0.223,
+      "step": 10688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.402720432852226e-07,
+      "loss": 0.2297,
+      "step": 10689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.391642315237744e-07,
+      "loss": 0.2648,
+      "step": 10690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.380575359096839e-07,
+      "loss": 0.2102,
+      "step": 10691
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.369519565283765e-07,
+      "loss": 0.2164,
+      "step": 10692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.35847493465182e-07,
+      "loss": 0.2572,
+      "step": 10693
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.347441468053472e-07,
+      "loss": 0.2074,
+      "step": 10694
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.336419166340373e-07,
+      "loss": 0.2025,
+      "step": 10695
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.325408030363227e-07,
+      "loss": 0.2384,
+      "step": 10696
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.314408060971921e-07,
+      "loss": 0.2412,
+      "step": 10697
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.30341925901549e-07,
+      "loss": 0.1859,
+      "step": 10698
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.292441625342076e-07,
+      "loss": 0.2163,
+      "step": 10699
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.281475160799015e-07,
+      "loss": 0.1986,
+      "step": 10700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.270519866232731e-07,
+      "loss": 0.1868,
+      "step": 10701
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.259575742488765e-07,
+      "loss": 0.2726,
+      "step": 10702
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.248642790411873e-07,
+      "loss": 0.2464,
+      "step": 10703
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.237721010845914e-07,
+      "loss": 0.225,
+      "step": 10704
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.226810404633814e-07,
+      "loss": 0.2426,
+      "step": 10705
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.215910972617766e-07,
+      "loss": 0.2299,
+      "step": 10706
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.205022715638996e-07,
+      "loss": 0.2032,
+      "step": 10707
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.1941456345379e-07,
+      "loss": 0.248,
+      "step": 10708
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.183279730154039e-07,
+      "loss": 0.258,
+      "step": 10709
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.172425003326059e-07,
+      "loss": 0.2064,
+      "step": 10710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.161581454891823e-07,
+      "loss": 0.2365,
+      "step": 10711
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.150749085688244e-07,
+      "loss": 0.2008,
+      "step": 10712
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.139927896551422e-07,
+      "loss": 0.2494,
+      "step": 10713
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.129117888316571e-07,
+      "loss": 0.2278,
+      "step": 10714
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.118319061818072e-07,
+      "loss": 0.2389,
+      "step": 10715
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.107531417889377e-07,
+      "loss": 0.2369,
+      "step": 10716
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.09675495736317e-07,
+      "loss": 0.2235,
+      "step": 10717
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.085989681071185e-07,
+      "loss": 0.1933,
+      "step": 10718
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.075235589844374e-07,
+      "loss": 0.2488,
+      "step": 10719
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.064492684512773e-07,
+      "loss": 0.2271,
+      "step": 10720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.053760965905536e-07,
+      "loss": 0.2141,
+      "step": 10721
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.043040434850998e-07,
+      "loss": 0.218,
+      "step": 10722
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.032331092176618e-07,
+      "loss": 0.2337,
+      "step": 10723
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.021632938708948e-07,
+      "loss": 0.2524,
+      "step": 10724
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.010945975273779e-07,
+      "loss": 0.212,
+      "step": 10725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.00027020269595e-07,
+      "loss": 0.2008,
+      "step": 10726
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.989605621799437e-07,
+      "loss": 0.2334,
+      "step": 10727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.97895223340743e-07,
+      "loss": 0.2331,
+      "step": 10728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.968310038342155e-07,
+      "loss": 0.2294,
+      "step": 10729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.957679037425022e-07,
+      "loss": 0.2204,
+      "step": 10730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.947059231476608e-07,
+      "loss": 0.2013,
+      "step": 10731
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.936450621316557e-07,
+      "loss": 0.2208,
+      "step": 10732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.925853207763714e-07,
+      "loss": 0.2614,
+      "step": 10733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.915266991636025e-07,
+      "loss": 0.2573,
+      "step": 10734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.904691973750553e-07,
+      "loss": 0.1912,
+      "step": 10735
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.89412815492356e-07,
+      "loss": 0.2364,
+      "step": 10736
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.883575535970381e-07,
+      "loss": 0.2191,
+      "step": 10737
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.873034117705527e-07,
+      "loss": 0.2188,
+      "step": 10738
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.862503900942617e-07,
+      "loss": 0.206,
+      "step": 10739
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.851984886494398e-07,
+      "loss": 0.2116,
+      "step": 10740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.841477075172823e-07,
+      "loss": 0.2034,
+      "step": 10741
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.830980467788892e-07,
+      "loss": 0.2183,
+      "step": 10742
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.820495065152758e-07,
+      "loss": 0.2516,
+      "step": 10743
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.810020868073773e-07,
+      "loss": 0.2153,
+      "step": 10744
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.799557877360355e-07,
+      "loss": 0.258,
+      "step": 10745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.789106093820095e-07,
+      "loss": 0.2445,
+      "step": 10746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.778665518259695e-07,
+      "loss": 0.2139,
+      "step": 10747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7682361514849783e-07,
+      "loss": 0.2364,
+      "step": 10748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.757817994300967e-07,
+      "loss": 0.2692,
+      "step": 10749
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7474110475117683e-07,
+      "loss": 0.1794,
+      "step": 10750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7370153119205896e-07,
+      "loss": 0.2406,
+      "step": 10751
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7266307883298887e-07,
+      "loss": 0.2241,
+      "step": 10752
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7162574775411415e-07,
+      "loss": 0.2095,
+      "step": 10753
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.7058953803550243e-07,
+      "loss": 0.2479,
+      "step": 10754
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.6955444975713137e-07,
+      "loss": 0.2029,
+      "step": 10755
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6852048299889205e-07,
+      "loss": 0.2281,
+      "step": 10756
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6748763784059234e-07,
+      "loss": 0.2294,
+      "step": 10757
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.664559143619501e-07,
+      "loss": 0.2604,
+      "step": 10758
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.654253126426e-07,
+      "loss": 0.2246,
+      "step": 10759
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6439583276208674e-07,
+      "loss": 0.2256,
+      "step": 10760
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.633674747998701e-07,
+      "loss": 0.2262,
+      "step": 10761
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6234023883532485e-07,
+      "loss": 0.2053,
+      "step": 10762
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.613141249477343e-07,
+      "loss": 0.2245,
+      "step": 10763
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6028913321630006e-07,
+      "loss": 0.2458,
+      "step": 10764
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.592652637201322e-07,
+      "loss": 0.2514,
+      "step": 10765
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5824251653826257e-07,
+      "loss": 0.2176,
+      "step": 10766
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5722089174962624e-07,
+      "loss": 0.2145,
+      "step": 10767
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.562003894330785e-07,
+      "loss": 0.2491,
+      "step": 10768
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.551810096673864e-07,
+      "loss": 0.2134,
+      "step": 10769
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.541627525312303e-07,
+      "loss": 0.2409,
+      "step": 10770
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.531456181032023e-07,
+      "loss": 0.2534,
+      "step": 10771
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.521296064618097e-07,
+      "loss": 0.1867,
+      "step": 10772
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5111471768546965e-07,
+      "loss": 0.2368,
+      "step": 10773
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.5010095185251955e-07,
+      "loss": 0.215,
+      "step": 10774
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4908830904120345e-07,
+      "loss": 0.2206,
+      "step": 10775
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.480767893296822e-07,
+      "loss": 0.2254,
+      "step": 10776
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.470663927960317e-07,
+      "loss": 0.2578,
+      "step": 10777
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.460571195182345e-07,
+      "loss": 0.2049,
+      "step": 10778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.450489695741916e-07,
+      "loss": 0.2599,
+      "step": 10779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4404194304171754e-07,
+      "loss": 0.2325,
+      "step": 10780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.430360399985367e-07,
+      "loss": 0.2351,
+      "step": 10781
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.42031260522292e-07,
+      "loss": 0.2361,
+      "step": 10782
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.4102760469053314e-07,
+      "loss": 0.2327,
+      "step": 10783
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.400250725807281e-07,
+      "loss": 0.2442,
+      "step": 10784
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3902366427025663e-07,
+      "loss": 0.258,
+      "step": 10785
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.380233798364103e-07,
+      "loss": 0.2466,
+      "step": 10786
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3702421935639725e-07,
+      "loss": 0.1814,
+      "step": 10787
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3602618290733586e-07,
+      "loss": 0.215,
+      "step": 10788
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3502927056625783e-07,
+      "loss": 0.2326,
+      "step": 10789
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3403348241010996e-07,
+      "loss": 0.2275,
+      "step": 10790
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3303881851575076e-07,
+      "loss": 0.2244,
+      "step": 10791
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3204527895995216e-07,
+      "loss": 0.2287,
+      "step": 10792
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.310528638194028e-07,
+      "loss": 0.257,
+      "step": 10793
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.3006157317069807e-07,
+      "loss": 0.2278,
+      "step": 10794
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.290714070903501e-07,
+      "loss": 0.2413,
+      "step": 10795
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2808236565478607e-07,
+      "loss": 0.2062,
+      "step": 10796
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2709444894034166e-07,
+      "loss": 0.2474,
+      "step": 10797
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.261076570232691e-07,
+      "loss": 0.2404,
+      "step": 10798
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2512198997973426e-07,
+      "loss": 0.2394,
+      "step": 10799
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2413744788581286e-07,
+      "loss": 0.2233,
+      "step": 10800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.231540308175008e-07,
+      "loss": 0.208,
+      "step": 10801
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2217173885069735e-07,
+      "loss": 0.2295,
+      "step": 10802
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2119057206122194e-07,
+      "loss": 0.2573,
+      "step": 10803
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.202105305248055e-07,
+      "loss": 0.2056,
+      "step": 10804
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1923161431708933e-07,
+      "loss": 0.2119,
+      "step": 10805
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.182538235136313e-07,
+      "loss": 0.2312,
+      "step": 10806
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.172771581899043e-07,
+      "loss": 0.2561,
+      "step": 10807
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1630161842128643e-07,
+      "loss": 0.251,
+      "step": 10808
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.153272042830775e-07,
+      "loss": 0.2313,
+      "step": 10809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.143539158504872e-07,
+      "loss": 0.2004,
+      "step": 10810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.133817531986356e-07,
+      "loss": 0.2117,
+      "step": 10811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.124107164025592e-07,
+      "loss": 0.2398,
+      "step": 10812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1144080553720655e-07,
+      "loss": 0.2037,
+      "step": 10813
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.104720206774376e-07,
+      "loss": 0.2485,
+      "step": 10814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0950436189803264e-07,
+      "loss": 0.2186,
+      "step": 10815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.085378292736752e-07,
+      "loss": 0.226,
+      "step": 10816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07572422878964e-07,
+      "loss": 0.2275,
+      "step": 10817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.066081427884194e-07,
+      "loss": 0.2101,
+      "step": 10818
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.056449890764652e-07,
+      "loss": 0.2473,
+      "step": 10819
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.046829618174419e-07,
+      "loss": 0.193,
+      "step": 10820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.037220610856035e-07,
+      "loss": 0.2321,
+      "step": 10821
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.027622869551123e-07,
+      "loss": 0.1933,
+      "step": 10822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0180363950005403e-07,
+      "loss": 0.1959,
+      "step": 10823
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0084611879441626e-07,
+      "loss": 0.2652,
+      "step": 10824
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9988972491210474e-07,
+      "loss": 0.2412,
+      "step": 10825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9893445792694215e-07,
+      "loss": 0.2713,
+      "step": 10826
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.979803179126562e-07,
+      "loss": 0.2235,
+      "step": 10827
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.970273049428913e-07,
+      "loss": 0.2152,
+      "step": 10828
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.960754190912069e-07,
+      "loss": 0.2315,
+      "step": 10829
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9512466043107097e-07,
+      "loss": 0.2087,
+      "step": 10830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9417502903586975e-07,
+      "loss": 0.2118,
+      "step": 10831
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9322652497889964e-07,
+      "loss": 0.2431,
+      "step": 10832
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.9227914833336707e-07,
+      "loss": 0.238,
+      "step": 10833
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.913328991723969e-07,
+      "loss": 0.2202,
+      "step": 10834
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.90387777569024e-07,
+      "loss": 0.2633,
+      "step": 10835
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8944378359619836e-07,
+      "loss": 0.2236,
+      "step": 10836
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.885009173267784e-07,
+      "loss": 0.2282,
+      "step": 10837
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8755917883354087e-07,
+      "loss": 0.2382,
+      "step": 10838
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.866185681891693e-07,
+      "loss": 0.221,
+      "step": 10839
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8567908546627063e-07,
+      "loss": 0.2158,
+      "step": 10840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8474073073735017e-07,
+      "loss": 0.2357,
+      "step": 10841
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8380350407484165e-07,
+      "loss": 0.2642,
+      "step": 10842
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.828674055510789e-07,
+      "loss": 0.2339,
+      "step": 10843
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8193243523831576e-07,
+      "loss": 0.2429,
+      "step": 10844
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.809985932087179e-07,
+      "loss": 0.218,
+      "step": 10845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.800658795343609e-07,
+      "loss": 0.2172,
+      "step": 10846
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7913429428723554e-07,
+      "loss": 0.1705,
+      "step": 10847
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.782038375392477e-07,
+      "loss": 0.232,
+      "step": 10848
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7727450936221155e-07,
+      "loss": 0.2094,
+      "step": 10849
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.76346309827858e-07,
+      "loss": 0.2219,
+      "step": 10850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7541923900783146e-07,
+      "loss": 0.2387,
+      "step": 10851
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7449329697368307e-07,
+      "loss": 0.2228,
+      "step": 10852
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.735684837968822e-07,
+      "loss": 0.2112,
+      "step": 10853
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7264479954881025e-07,
+      "loss": 0.2083,
+      "step": 10854
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7172224430076005e-07,
+      "loss": 0.239,
+      "step": 10855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7080081812393973e-07,
+      "loss": 0.2221,
+      "step": 10856
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6988052108946903e-07,
+      "loss": 0.2144,
+      "step": 10857
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6896135326837776e-07,
+      "loss": 0.2192,
+      "step": 10858
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6804331473161254e-07,
+      "loss": 0.2054,
+      "step": 10859
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.671264055500334e-07,
+      "loss": 0.2134,
+      "step": 10860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.662106257944087e-07,
+      "loss": 0.1994,
+      "step": 10861
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6529597553542184e-07,
+      "loss": 0.2147,
+      "step": 10862
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6438245484366974e-07,
+      "loss": 0.2237,
+      "step": 10863
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.634700637896643e-07,
+      "loss": 0.2098,
+      "step": 10864
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6255880244382423e-07,
+      "loss": 0.2051,
+      "step": 10865
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.616486708764849e-07,
+      "loss": 0.2087,
+      "step": 10866
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.607396691578968e-07,
+      "loss": 0.2543,
+      "step": 10867
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.5983179735821714e-07,
+      "loss": 0.2506,
+      "step": 10868
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.589250555475232e-07,
+      "loss": 0.2501,
+      "step": 10869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.580194437957957e-07,
+      "loss": 0.2312,
+      "step": 10870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.571149621729369e-07,
+      "loss": 0.2388,
+      "step": 10871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.562116107487595e-07,
+      "loss": 0.2472,
+      "step": 10872
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.55309389592986e-07,
+      "loss": 0.223,
+      "step": 10873
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.544082987752523e-07,
+      "loss": 0.2266,
+      "step": 10874
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.535083383651111e-07,
+      "loss": 0.2009,
+      "step": 10875
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5260950843202354e-07,
+      "loss": 0.2194,
+      "step": 10876
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5171180904536747e-07,
+      "loss": 0.2281,
+      "step": 10877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5081524027442746e-07,
+      "loss": 0.2317,
+      "step": 10878
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.499198021884048e-07,
+      "loss": 0.2148,
+      "step": 10879
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4902549485641586e-07,
+      "loss": 0.218,
+      "step": 10880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.481323183474855e-07,
+      "loss": 0.2495,
+      "step": 10881
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.472402727305518e-07,
+      "loss": 0.2727,
+      "step": 10882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.463493580744681e-07,
+      "loss": 0.261,
+      "step": 10883
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.454595744479977e-07,
+      "loss": 0.2632,
+      "step": 10884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4457092191982063e-07,
+      "loss": 0.2013,
+      "step": 10885
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.436834005585221e-07,
+      "loss": 0.219,
+      "step": 10886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4279701043260893e-07,
+      "loss": 0.2567,
+      "step": 10887
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.419117516104914e-07,
+      "loss": 0.226,
+      "step": 10888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4102762416050315e-07,
+      "loss": 0.2211,
+      "step": 10889
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.401446281508813e-07,
+      "loss": 0.2405,
+      "step": 10890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3926276364978137e-07,
+      "loss": 0.2286,
+      "step": 10891
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.383820307252672e-07,
+      "loss": 0.2009,
+      "step": 10892
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.375024294453194e-07,
+      "loss": 0.1905,
+      "step": 10893
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.366239598778287e-07,
+      "loss": 0.1979,
+      "step": 10894
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.357466220905975e-07,
+      "loss": 0.2318,
+      "step": 10895
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.348704161513416e-07,
+      "loss": 0.2564,
+      "step": 10896
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3399534212769365e-07,
+      "loss": 0.2446,
+      "step": 10897
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.331214000871946e-07,
+      "loss": 0.2226,
+      "step": 10898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.322485900972955e-07,
+      "loss": 0.1931,
+      "step": 10899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.31376912225369e-07,
+      "loss": 0.2317,
+      "step": 10900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.3050636653869147e-07,
+      "loss": 0.2215,
+      "step": 10901
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.296369531044557e-07,
+      "loss": 0.2195,
+      "step": 10902
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2876867198976646e-07,
+      "loss": 0.2445,
+      "step": 10903
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2790152326164004e-07,
+      "loss": 0.2305,
+      "step": 10904
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.270355069870079e-07,
+      "loss": 0.2531,
+      "step": 10905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2617062323271494e-07,
+      "loss": 0.2363,
+      "step": 10906
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.253068720655128e-07,
+      "loss": 0.2442,
+      "step": 10907
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.244442535520714e-07,
+      "loss": 0.2285,
+      "step": 10908
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2358276775897087e-07,
+      "loss": 0.2211,
+      "step": 10909
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2272241475270634e-07,
+      "loss": 0.2314,
+      "step": 10910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2186319459967973e-07,
+      "loss": 0.221,
+      "step": 10911
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.210051073662096e-07,
+      "loss": 0.241,
+      "step": 10912
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.2014815311852796e-07,
+      "loss": 0.221,
+      "step": 10913
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.192923319227803e-07,
+      "loss": 0.2112,
+      "step": 10914
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1843764384501707e-07,
+      "loss": 0.2387,
+      "step": 10915
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1758408895121215e-07,
+      "loss": 0.231,
+      "step": 10916
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.167316673072429e-07,
+      "loss": 0.2136,
+      "step": 10917
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.15880378978905e-07,
+      "loss": 0.2184,
+      "step": 10918
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.150302240319042e-07,
+      "loss": 0.2275,
+      "step": 10919
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.141812025318547e-07,
+      "loss": 0.2436,
+      "step": 10920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.13333314544294e-07,
+      "loss": 0.2455,
+      "step": 10921
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.124865601346616e-07,
+      "loss": 0.2436,
+      "step": 10922
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.116409393683134e-07,
+      "loss": 0.2439,
+      "step": 10923
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.107964523105189e-07,
+      "loss": 0.248,
+      "step": 10924
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0995309902646095e-07,
+      "loss": 0.2403,
+      "step": 10925
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.091108795812292e-07,
+      "loss": 0.2277,
+      "step": 10926
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.082697940398349e-07,
+      "loss": 0.26,
+      "step": 10927
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.074298424671912e-07,
+      "loss": 0.2488,
+      "step": 10928
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0659102492813117e-07,
+      "loss": 0.2777,
+      "step": 10929
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.0575334148739974e-07,
+      "loss": 0.2111,
+      "step": 10930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.049167922096485e-07,
+      "loss": 0.234,
+      "step": 10931
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0408137715945085e-07,
+      "loss": 0.2356,
+      "step": 10932
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.032470964012851e-07,
+      "loss": 0.2305,
+      "step": 10933
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.024139499995449e-07,
+      "loss": 0.2492,
+      "step": 10934
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.0158193801853697e-07,
+      "loss": 0.1955,
+      "step": 10935
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.007510605224767e-07,
+      "loss": 0.2012,
+      "step": 10936
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.999213175754961e-07,
+      "loss": 0.2393,
+      "step": 10937
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.99092709241639e-07,
+      "loss": 0.2432,
+      "step": 10938
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9826523558485916e-07,
+      "loss": 0.2289,
+      "step": 10939
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9743889666902724e-07,
+      "loss": 0.2278,
+      "step": 10940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.966136925579205e-07,
+      "loss": 0.2137,
+      "step": 10941
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.95789623315233e-07,
+      "loss": 0.2192,
+      "step": 10942
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.949666890045705e-07,
+      "loss": 0.2288,
+      "step": 10943
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.941448896894505e-07,
+      "loss": 0.2462,
+      "step": 10944
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.933242254332991e-07,
+      "loss": 0.2337,
+      "step": 10945
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9250469629946373e-07,
+      "loss": 0.1937,
+      "step": 10946
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.916863023511973e-07,
+      "loss": 0.2161,
+      "step": 10947
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9086904365166425e-07,
+      "loss": 0.1972,
+      "step": 10948
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9005292026394736e-07,
+      "loss": 0.2306,
+      "step": 10949
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8923793225103635e-07,
+      "loss": 0.2141,
+      "step": 10950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.884240796758375e-07,
+      "loss": 0.2016,
+      "step": 10951
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8761136260116727e-07,
+      "loss": 0.2239,
+      "step": 10952
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8679978108975045e-07,
+      "loss": 0.2391,
+      "step": 10953
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.859893352042336e-07,
+      "loss": 0.2089,
+      "step": 10954
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.851800250071668e-07,
+      "loss": 0.243,
+      "step": 10955
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.843718505610149e-07,
+      "loss": 0.2602,
+      "step": 10956
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.835648119281614e-07,
+      "loss": 0.2104,
+      "step": 10957
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.827589091708932e-07,
+      "loss": 0.1967,
+      "step": 10958
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.8195414235141216e-07,
+      "loss": 0.205,
+      "step": 10959
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.811505115318386e-07,
+      "loss": 0.2306,
+      "step": 10960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.803480167741929e-07,
+      "loss": 0.2426,
+      "step": 10961
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.795466581404221e-07,
+      "loss": 0.1987,
+      "step": 10962
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.787464356923736e-07,
+      "loss": 0.1908,
+      "step": 10963
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.779473494918111e-07,
+      "loss": 0.2073,
+      "step": 10964
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.771493996004171e-07,
+      "loss": 0.2258,
+      "step": 10965
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7635258607977554e-07,
+      "loss": 0.2176,
+      "step": 10966
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.755569089913873e-07,
+      "loss": 0.2303,
+      "step": 10967
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.747623683966716e-07,
+      "loss": 0.4324,
+      "step": 10968
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.739689643569493e-07,
+      "loss": 0.2211,
+      "step": 10969
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.731766969334615e-07,
+      "loss": 0.2264,
+      "step": 10970
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.723855661873575e-07,
+      "loss": 0.272,
+      "step": 10971
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.71595572179697e-07,
+      "loss": 0.1971,
+      "step": 10972
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7080671497146105e-07,
+      "loss": 0.2006,
+      "step": 10973
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.7001899462353265e-07,
+      "loss": 0.228,
+      "step": 10974
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.692324111967132e-07,
+      "loss": 0.2094,
+      "step": 10975
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6844696475171405e-07,
+      "loss": 0.2417,
+      "step": 10976
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.676626553491601e-07,
+      "loss": 0.2425,
+      "step": 10977
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6687948304958467e-07,
+      "loss": 0.2697,
+      "step": 10978
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.660974479134376e-07,
+      "loss": 0.2343,
+      "step": 10979
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.653165500010807e-07,
+      "loss": 0.2584,
+      "step": 10980
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6453678937278736e-07,
+      "loss": 0.2363,
+      "step": 10981
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6375816608874115e-07,
+      "loss": 0.2227,
+      "step": 10982
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6298068020903896e-07,
+      "loss": 0.2417,
+      "step": 10983
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6220433179369284e-07,
+      "loss": 0.2438,
+      "step": 10984
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6142912090262485e-07,
+      "loss": 0.2483,
+      "step": 10985
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6065504759566373e-07,
+      "loss": 0.2382,
+      "step": 10986
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5988211193256173e-07,
+      "loss": 0.2144,
+      "step": 10987
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.591103139729745e-07,
+      "loss": 0.2521,
+      "step": 10988
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5833965377647095e-07,
+      "loss": 0.181,
+      "step": 10989
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5757013140253694e-07,
+      "loss": 0.2571,
+      "step": 10990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.568017469105649e-07,
+      "loss": 0.2653,
+      "step": 10991
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.560345003598658e-07,
+      "loss": 0.238,
+      "step": 10992
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5526839180965547e-07,
+      "loss": 0.2175,
+      "step": 10993
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.545034213190667e-07,
+      "loss": 0.2221,
+      "step": 10994
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.537395889471422e-07,
+      "loss": 0.2505,
+      "step": 10995
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5297689475283827e-07,
+      "loss": 0.238,
+      "step": 10996
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.522153387950227e-07,
+      "loss": 0.2419,
+      "step": 10997
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5145492113247516e-07,
+      "loss": 0.2198,
+      "step": 10998
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5069564182388873e-07,
+      "loss": 0.2182,
+      "step": 10999
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.499375009278665e-07,
+      "loss": 0.1852,
+      "step": 11000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.491804985029267e-07,
+      "loss": 0.2181,
+      "step": 11001
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4842463460749587e-07,
+      "loss": 0.1951,
+      "step": 11002
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.476699092999174e-07,
+      "loss": 0.2249,
+      "step": 11003
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4691632263844133e-07,
+      "loss": 0.2382,
+      "step": 11004
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.461638746812328e-07,
+      "loss": 0.2578,
+      "step": 11005
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4541256548637205e-07,
+      "loss": 0.2315,
+      "step": 11006
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4466239511184427e-07,
+      "loss": 0.2739,
+      "step": 11007
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.439133636155533e-07,
+      "loss": 0.237,
+      "step": 11008
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4316547105531105e-07,
+      "loss": 0.2266,
+      "step": 11009
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.424187174888448e-07,
+      "loss": 0.2313,
+      "step": 11010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.416731029737901e-07,
+      "loss": 0.2239,
+      "step": 11011
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.409286275676975e-07,
+      "loss": 0.2556,
+      "step": 11012
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.401852913280278e-07,
+      "loss": 0.2214,
+      "step": 11013
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.394430943121567e-07,
+      "loss": 0.201,
+      "step": 11014
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3870203657736843e-07,
+      "loss": 0.2369,
+      "step": 11015
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3796211818086057e-07,
+      "loss": 0.2063,
+      "step": 11016
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.372233391797457e-07,
+      "loss": 0.2479,
+      "step": 11017
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3648569963104326e-07,
+      "loss": 0.1991,
+      "step": 11018
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.357491995916894e-07,
+      "loss": 0.2782,
+      "step": 11019
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3501383911852692e-07,
+      "loss": 0.247,
+      "step": 11020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3427961826831544e-07,
+      "loss": 0.2665,
+      "step": 11021
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.33546537097728e-07,
+      "loss": 0.2445,
+      "step": 11022
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.328145956633443e-07,
+      "loss": 0.2275,
+      "step": 11023
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3208379402165746e-07,
+      "loss": 0.2043,
+      "step": 11024
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.313541322290774e-07,
+      "loss": 0.221,
+      "step": 11025
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.306256103419191e-07,
+      "loss": 0.2529,
+      "step": 11026
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2989822841641417e-07,
+      "loss": 0.2303,
+      "step": 11027
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2917198650870608e-07,
+      "loss": 0.2398,
+      "step": 11028
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2844688467484664e-07,
+      "loss": 0.2236,
+      "step": 11029
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2772292297080277e-07,
+      "loss": 0.2243,
+      "step": 11030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2700010145245642e-07,
+      "loss": 0.2313,
+      "step": 11031
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2627842017559297e-07,
+      "loss": 0.2262,
+      "step": 11032
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2555787919591787e-07,
+      "loss": 0.2283,
+      "step": 11033
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2483847856904494e-07,
+      "loss": 0.2155,
+      "step": 11034
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.241202183504998e-07,
+      "loss": 0.231,
+      "step": 11035
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2340309859572139e-07,
+      "loss": 0.1989,
+      "step": 11036
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2268711936006047e-07,
+      "loss": 0.2411,
+      "step": 11037
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2197228069877617e-07,
+      "loss": 0.1964,
+      "step": 11038
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2125858266704769e-07,
+      "loss": 0.2138,
+      "step": 11039
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.205460253199576e-07,
+      "loss": 0.2215,
+      "step": 11040
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1983460871250526e-07,
+      "loss": 0.2309,
+      "step": 11041
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.191243328996001e-07,
+      "loss": 0.2043,
+      "step": 11042
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1841519793606658e-07,
+      "loss": 0.2221,
+      "step": 11043
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1770720387663423e-07,
+      "loss": 0.2281,
+      "step": 11044
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.170003507759527e-07,
+      "loss": 0.2125,
+      "step": 11045
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.1629463868857834e-07,
+      "loss": 0.2392,
+      "step": 11046
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1559006766898092e-07,
+      "loss": 0.2184,
+      "step": 11047
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1488663777154361e-07,
+      "loss": 0.2291,
+      "step": 11048
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1418434905055794e-07,
+      "loss": 0.2152,
+      "step": 11049
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.134832015602306e-07,
+      "loss": 0.2468,
+      "step": 11050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1278319535467993e-07,
+      "loss": 0.2129,
+      "step": 11051
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1208433048793273e-07,
+      "loss": 0.2103,
+      "step": 11052
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1138660701393254e-07,
+      "loss": 0.2082,
+      "step": 11053
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1069002498653122e-07,
+      "loss": 0.2168,
+      "step": 11054
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.099945844594958e-07,
+      "loss": 0.1804,
+      "step": 11055
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0930028548650171e-07,
+      "loss": 0.2079,
+      "step": 11056
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0860712812113603e-07,
+      "loss": 0.2102,
+      "step": 11057
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.079151124169043e-07,
+      "loss": 0.2197,
+      "step": 11058
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0722423842721549e-07,
+      "loss": 0.1965,
+      "step": 11059
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0653450620539527e-07,
+      "loss": 0.2062,
+      "step": 11060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0584591580467938e-07,
+      "loss": 0.2247,
+      "step": 11061
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0515846727821696e-07,
+      "loss": 0.2472,
+      "step": 11062
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0447216067906727e-07,
+      "loss": 0.2107,
+      "step": 11063
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0378699606020457e-07,
+      "loss": 0.1995,
+      "step": 11064
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0310297347450822e-07,
+      "loss": 0.2351,
+      "step": 11065
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0242009297477772e-07,
+      "loss": 0.2169,
+      "step": 11066
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.017383546137208e-07,
+      "loss": 0.2501,
+      "step": 11067
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0105775844395379e-07,
+      "loss": 0.2232,
+      "step": 11068
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0037830451801132e-07,
+      "loss": 0.2068,
+      "step": 11069
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9969999288833307e-07,
+      "loss": 0.2033,
+      "step": 11070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.990228236072772e-07,
+      "loss": 0.23,
+      "step": 11071
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9834679672710854e-07,
+      "loss": 0.2271,
+      "step": 11072
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9767191230000536e-07,
+      "loss": 0.2108,
+      "step": 11073
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9699817037805935e-07,
+      "loss": 0.2251,
+      "step": 11074
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.963255710132722e-07,
+      "loss": 0.2175,
+      "step": 11075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9565411425755908e-07,
+      "loss": 0.2297,
+      "step": 11076
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9498380016274352e-07,
+      "loss": 0.1989,
+      "step": 11077
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9431462878056416e-07,
+      "loss": 0.213,
+      "step": 11078
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9364660016266965e-07,
+      "loss": 0.3032,
+      "step": 11079
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.929797143606238e-07,
+      "loss": 0.2451,
+      "step": 11080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9231397142589535e-07,
+      "loss": 0.2611,
+      "step": 11081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9164937140987492e-07,
+      "loss": 0.2311,
+      "step": 11082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9098591436385316e-07,
+      "loss": 0.2362,
+      "step": 11083
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9032360033904406e-07,
+      "loss": 0.2342,
+      "step": 11084
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8966242938656177e-07,
+      "loss": 0.2487,
+      "step": 11085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8900240155744375e-07,
+      "loss": 0.235,
+      "step": 11086
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.883435169026293e-07,
+      "loss": 0.1995,
+      "step": 11087
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8768577547297606e-07,
+      "loss": 0.2241,
+      "step": 11088
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8702917731925173e-07,
+      "loss": 0.1837,
+      "step": 11089
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8637372249213246e-07,
+      "loss": 0.2224,
+      "step": 11090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.857194110422128e-07,
+      "loss": 0.2336,
+      "step": 11091
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8506624301999398e-07,
+      "loss": 0.2245,
+      "step": 11092
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8441421847588902e-07,
+      "loss": 0.2419,
+      "step": 11093
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8376333746022432e-07,
+      "loss": 0.2456,
+      "step": 11094
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8311360002323807e-07,
+      "loss": 0.2425,
+      "step": 11095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8246500621507846e-07,
+      "loss": 0.2634,
+      "step": 11096
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8181755608580875e-07,
+      "loss": 0.2518,
+      "step": 11097
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8117124968540067e-07,
+      "loss": 0.2094,
+      "step": 11098
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.805260870637393e-07,
+      "loss": 0.2454,
+      "step": 11099
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.798820682706198e-07,
+      "loss": 0.2222,
+      "step": 11100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.792391933557508e-07,
+      "loss": 0.2257,
+      "step": 11101
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7859746236875251e-07,
+      "loss": 0.2689,
+      "step": 11102
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.779568753591554e-07,
+      "loss": 0.1814,
+      "step": 11103
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7731743237640485e-07,
+      "loss": 0.2425,
+      "step": 11104
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.766791334698531e-07,
+      "loss": 0.2117,
+      "step": 11105
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7604197868876736e-07,
+      "loss": 0.2131,
+      "step": 11106
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7540596808232834e-07,
+      "loss": 0.2211,
+      "step": 11107
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7477110169962173e-07,
+      "loss": 0.2457,
+      "step": 11108
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7413737958965337e-07,
+      "loss": 0.1934,
+      "step": 11109
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.735048018013341e-07,
+      "loss": 0.1969,
+      "step": 11110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.728733683834882e-07,
+      "loss": 0.2494,
+      "step": 11111
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.722430793848534e-07,
+      "loss": 0.2127,
+      "step": 11112
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7161393485407906e-07,
+      "loss": 0.2626,
+      "step": 11113
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7098593483972469e-07,
+      "loss": 0.2444,
+      "step": 11114
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7035907939026153e-07,
+      "loss": 0.2305,
+      "step": 11115
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.697333685540725e-07,
+      "loss": 0.1915,
+      "step": 11116
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6910880237945237e-07,
+      "loss": 0.2304,
+      "step": 11117
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.684853809146092e-07,
+      "loss": 0.2234,
+      "step": 11118
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.678631042076595e-07,
+      "loss": 0.2318,
+      "step": 11119
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.672419723066332e-07,
+      "loss": 0.2675,
+      "step": 11120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6662198525947525e-07,
+      "loss": 0.2408,
+      "step": 11121
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6600314311403408e-07,
+      "loss": 0.2408,
+      "step": 11122
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6538544591807648e-07,
+      "loss": 0.2264,
+      "step": 11123
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6476889371928094e-07,
+      "loss": 0.224,
+      "step": 11124
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6415348656523278e-07,
+      "loss": 0.2394,
+      "step": 11125
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6353922450343227e-07,
+      "loss": 0.2382,
+      "step": 11126
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6292610758129156e-07,
+      "loss": 0.2393,
+      "step": 11127
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6231413584613276e-07,
+      "loss": 0.2413,
+      "step": 11128
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6170330934518973e-07,
+      "loss": 0.244,
+      "step": 11129
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.610936281256098e-07,
+      "loss": 0.2668,
+      "step": 11130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.604850922344503e-07,
+      "loss": 0.1995,
+      "step": 11131
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.59877701718682e-07,
+      "loss": 0.1996,
+      "step": 11132
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5927145662518404e-07,
+      "loss": 0.2297,
+      "step": 11133
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5866635700074904e-07,
+      "loss": 0.2559,
+      "step": 11134
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.580624028920813e-07,
+      "loss": 0.2504,
+      "step": 11135
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5745959434579516e-07,
+      "loss": 0.2584,
+      "step": 11136
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5685793140842174e-07,
+      "loss": 0.2443,
+      "step": 11137
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.562574141263956e-07,
+      "loss": 0.1977,
+      "step": 11138
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5565804254606964e-07,
+      "loss": 0.2221,
+      "step": 11139
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5505981671370518e-07,
+      "loss": 0.2076,
+      "step": 11140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5446273667547695e-07,
+      "loss": 0.2179,
+      "step": 11141
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5386680247746809e-07,
+      "loss": 0.1826,
+      "step": 11142
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5327201416567683e-07,
+      "loss": 0.2402,
+      "step": 11143
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5267837178600974e-07,
+      "loss": 0.2763,
+      "step": 11144
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5208587538428686e-07,
+      "loss": 0.231,
+      "step": 11145
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.5149452500624162e-07,
+      "loss": 0.2405,
+      "step": 11146
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.509043206975158e-07,
+      "loss": 0.1994,
+      "step": 11147
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.50315262503663e-07,
+      "loss": 0.2162,
+      "step": 11148
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.497273504701502e-07,
+      "loss": 0.209,
+      "step": 11149
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.491405846423527e-07,
+      "loss": 0.2027,
+      "step": 11150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4855496506556265e-07,
+      "loss": 0.2449,
+      "step": 11151
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.479704917849789e-07,
+      "loss": 0.2105,
+      "step": 11152
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4738716484571367e-07,
+      "loss": 0.2202,
+      "step": 11153
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4680498429279099e-07,
+      "loss": 0.202,
+      "step": 11154
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4622395017114486e-07,
+      "loss": 0.1899,
+      "step": 11155
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4564406252562446e-07,
+      "loss": 0.2295,
+      "step": 11156
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4506532140098562e-07,
+      "loss": 0.2058,
+      "step": 11157
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.444877268418976e-07,
+      "loss": 0.2179,
+      "step": 11158
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4391127889294309e-07,
+      "loss": 0.224,
+      "step": 11159
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4333597759861484e-07,
+      "loss": 0.2005,
+      "step": 11160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4276182300331564e-07,
+      "loss": 0.1784,
+      "step": 11161
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4218881515136174e-07,
+      "loss": 0.2208,
+      "step": 11162
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416169540869794e-07,
+      "loss": 0.212,
+      "step": 11163
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4104623985431e-07,
+      "loss": 0.2252,
+      "step": 11164
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4047667249739992e-07,
+      "loss": 0.1986,
+      "step": 11165
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3990825206021406e-07,
+      "loss": 0.2523,
+      "step": 11166
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3934097858662397e-07,
+      "loss": 0.1864,
+      "step": 11167
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3877485212041464e-07,
+      "loss": 0.2458,
+      "step": 11168
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.382098727052794e-07,
+      "loss": 0.2149,
+      "step": 11169
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.376460403848301e-07,
+      "loss": 0.1828,
+      "step": 11170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3708335520258185e-07,
+      "loss": 0.2394,
+      "step": 11171
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3652181720196665e-07,
+      "loss": 0.235,
+      "step": 11172
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3596142642632814e-07,
+      "loss": 0.2522,
+      "step": 11173
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.354021829189167e-07,
+      "loss": 0.2217,
+      "step": 11174
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.348440867228995e-07,
+      "loss": 0.2256,
+      "step": 11175
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.342871378813504e-07,
+      "loss": 0.2515,
+      "step": 11176
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3373133643725667e-07,
+      "loss": 0.2552,
+      "step": 11177
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3317668243352066e-07,
+      "loss": 0.2105,
+      "step": 11178
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.326231759129498e-07,
+      "loss": 0.253,
+      "step": 11179
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3207081691826827e-07,
+      "loss": 0.2618,
+      "step": 11180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3151960549210695e-07,
+      "loss": 0.251,
+      "step": 11181
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3096954167701347e-07,
+      "loss": 0.2312,
+      "step": 11182
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.304206255154422e-07,
+      "loss": 0.2341,
+      "step": 11183
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2987285704976093e-07,
+      "loss": 0.2104,
+      "step": 11184
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2932623632224917e-07,
+      "loss": 0.2209,
+      "step": 11185
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2878076337509648e-07,
+      "loss": 0.2391,
+      "step": 11186
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2823643825040587e-07,
+      "loss": 0.233,
+      "step": 11187
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.276932609901904e-07,
+      "loss": 0.2194,
+      "step": 11188
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2715123163637486e-07,
+      "loss": 0.2145,
+      "step": 11189
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2661035023079416e-07,
+      "loss": 0.2299,
+      "step": 11190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.260706168151965e-07,
+      "loss": 0.2095,
+      "step": 11191
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.255320314312419e-07,
+      "loss": 0.2287,
+      "step": 11192
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2499459412049718e-07,
+      "loss": 0.2158,
+      "step": 11193
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2445830492444742e-07,
+      "loss": 0.2299,
+      "step": 11194
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.239231638844862e-07,
+      "loss": 0.2361,
+      "step": 11195
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2338917104191382e-07,
+      "loss": 0.2695,
+      "step": 11196
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2285632643795065e-07,
+      "loss": 0.2418,
+      "step": 11197
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.223246301137204e-07,
+      "loss": 0.2331,
+      "step": 11198
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2179408211026367e-07,
+      "loss": 0.2189,
+      "step": 11199
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.212646824685293e-07,
+      "loss": 0.2248,
+      "step": 11200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2073643122937793e-07,
+      "loss": 0.197,
+      "step": 11201
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2020932843358369e-07,
+      "loss": 0.2658,
+      "step": 11202
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.196833741218306e-07,
+      "loss": 0.2204,
+      "step": 11203
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1915856833471295e-07,
+      "loss": 0.1983,
+      "step": 11204
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1863491111273827e-07,
+      "loss": 0.2185,
+      "step": 11205
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1811240249632427e-07,
+      "loss": 0.2098,
+      "step": 11206
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1759104252579867e-07,
+      "loss": 0.1922,
+      "step": 11207
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1707083124140594e-07,
+      "loss": 0.2282,
+      "step": 11208
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1655176868329564e-07,
+      "loss": 0.2465,
+      "step": 11209
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1603385489153073e-07,
+      "loss": 0.2785,
+      "step": 11210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1551708990608756e-07,
+      "loss": 0.2211,
+      "step": 11211
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1500147376685088e-07,
+      "loss": 0.2131,
+      "step": 11212
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1448700651361721e-07,
+      "loss": 0.1759,
+      "step": 11213
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1397368818609976e-07,
+      "loss": 0.2451,
+      "step": 11214
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1346151882391187e-07,
+      "loss": 0.234,
+      "step": 11215
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1295049846659189e-07,
+      "loss": 0.2088,
+      "step": 11216
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1244062715357827e-07,
+      "loss": 0.2306,
+      "step": 11217
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1193190492422456e-07,
+      "loss": 0.2412,
+      "step": 11218
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1142433181779765e-07,
+      "loss": 0.226,
+      "step": 11219
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1091790787347456e-07,
+      "loss": 0.2294,
+      "step": 11220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1041263313034067e-07,
+      "loss": 0.2471,
+      "step": 11221
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0990850762739813e-07,
+      "loss": 0.2156,
+      "step": 11222
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.094055314035558e-07,
+      "loss": 0.2652,
+      "step": 11223
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0890370449763765e-07,
+      "loss": 0.24,
+      "step": 11224
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0840302694837433e-07,
+      "loss": 0.2471,
+      "step": 11225
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0790349879440997e-07,
+      "loss": 0.2496,
+      "step": 11226
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0740512007430203e-07,
+      "loss": 0.2092,
+      "step": 11227
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0690789082651642e-07,
+      "loss": 0.2029,
+      "step": 11228
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0641181108943076e-07,
+      "loss": 0.205,
+      "step": 11229
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0591688090133611e-07,
+      "loss": 0.2085,
+      "step": 11230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.054231003004319e-07,
+      "loss": 0.234,
+      "step": 11231
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0493046932483263e-07,
+      "loss": 0.2367,
+      "step": 11232
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0443898801255791e-07,
+      "loss": 0.2363,
+      "step": 11233
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0394865640154405e-07,
+      "loss": 0.2326,
+      "step": 11234
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0345947452963745e-07,
+      "loss": 0.2281,
+      "step": 11235
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0297144243459623e-07,
+      "loss": 0.2315,
+      "step": 11236
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0248456015408358e-07,
+      "loss": 0.2109,
+      "step": 11237
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0199882772568614e-07,
+      "loss": 0.258,
+      "step": 11238
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0151424518688889e-07,
+      "loss": 0.2435,
+      "step": 11239
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0103081257509694e-07,
+      "loss": 0.2331,
+      "step": 11240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0054852992762543e-07,
+      "loss": 0.2676,
+      "step": 11241
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0006739728169289e-07,
+      "loss": 0.1944,
+      "step": 11242
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.958741467444133e-08,
+      "loss": 0.2394,
+      "step": 11243
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.91085821429144e-08,
+      "loss": 0.2545,
+      "step": 11244
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.863089972407092e-08,
+      "loss": 0.2152,
+      "step": 11245
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.815436745478135e-08,
+      "loss": 0.2505,
+      "step": 11246
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.767898537182463e-08,
+      "loss": 0.2342,
+      "step": 11247
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.720475351189473e-08,
+      "loss": 0.2448,
+      "step": 11248
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.673167191159404e-08,
+      "loss": 0.2274,
+      "step": 11249
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.625974060743836e-08,
+      "loss": 0.2138,
+      "step": 11250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.578895963585021e-08,
+      "loss": 0.2108,
+      "step": 11251
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.53193290331672e-08,
+      "loss": 0.2189,
+      "step": 11252
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.485084883564033e-08,
+      "loss": 0.1969,
+      "step": 11253
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.438351907942567e-08,
+      "loss": 0.2242,
+      "step": 11254
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.391733980059436e-08,
+      "loss": 0.2386,
+      "step": 11255
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.34523110351293e-08,
+      "loss": 0.237,
+      "step": 11256
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.298843281892178e-08,
+      "loss": 0.2202,
+      "step": 11257
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.252570518777814e-08,
+      "loss": 0.2389,
+      "step": 11258
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.206412817740983e-08,
+      "loss": 0.2347,
+      "step": 11259
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.160370182344835e-08,
+      "loss": 0.2561,
+      "step": 11260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.114442616142527e-08,
+      "loss": 0.2118,
+      "step": 11261
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.06863012267939e-08,
+      "loss": 0.2156,
+      "step": 11262
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.022932705491427e-08,
+      "loss": 0.2106,
+      "step": 11263
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.977350368105486e-08,
+      "loss": 0.244,
+      "step": 11264
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.931883114039918e-08,
+      "loss": 0.206,
+      "step": 11265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.88653094680425e-08,
+      "loss": 0.2283,
+      "step": 11266
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.841293869898514e-08,
+      "loss": 0.208,
+      "step": 11267
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.796171886814752e-08,
+      "loss": 0.216,
+      "step": 11268
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.751165001035511e-08,
+      "loss": 0.2639,
+      "step": 11269
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.706273216034511e-08,
+      "loss": 0.2615,
+      "step": 11270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.661496535276814e-08,
+      "loss": 0.2371,
+      "step": 11271
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.616834962218322e-08,
+      "loss": 0.2096,
+      "step": 11272
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.572288500306447e-08,
+      "loss": 0.2273,
+      "step": 11273
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.527857152979268e-08,
+      "loss": 0.2129,
+      "step": 11274
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.483540923666211e-08,
+      "loss": 0.1968,
+      "step": 11275
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.439339815787872e-08,
+      "loss": 0.2517,
+      "step": 11276
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.39525383275569e-08,
+      "loss": 0.232,
+      "step": 11277
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.351282977972607e-08,
+      "loss": 0.2286,
+      "step": 11278
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.307427254832412e-08,
+      "loss": 0.2143,
+      "step": 11279
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 8.263686666719894e-08,
+      "loss": 0.2319,
+      "step": 11280
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.220061217011355e-08,
+      "loss": 0.2357,
+      "step": 11281
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.176550909074098e-08,
+      "loss": 0.2332,
+      "step": 11282
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.133155746266108e-08,
+      "loss": 0.2076,
+      "step": 11283
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.089875731937035e-08,
+      "loss": 0.2166,
+      "step": 11284
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.046710869427375e-08,
+      "loss": 0.2352,
+      "step": 11285
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.003661162068466e-08,
+      "loss": 0.2452,
+      "step": 11286
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.960726613183645e-08,
+      "loss": 0.2201,
+      "step": 11287
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.917907226086262e-08,
+      "loss": 0.2202,
+      "step": 11288
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.875203004081511e-08,
+      "loss": 0.2184,
+      "step": 11289
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.832613950465583e-08,
+      "loss": 0.2307,
+      "step": 11290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.790140068525353e-08,
+      "loss": 0.2362,
+      "step": 11291
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.74778136153953e-08,
+      "loss": 0.2235,
+      "step": 11292
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.705537832777331e-08,
+      "loss": 0.2127,
+      "step": 11293
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.66340948549915e-08,
+      "loss": 0.2127,
+      "step": 11294
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.621396322956886e-08,
+      "loss": 0.2381,
+      "step": 11295
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.57949834839311e-08,
+      "loss": 0.2243,
+      "step": 11296
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.537715565041736e-08,
+      "loss": 0.2095,
+      "step": 11297
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.496047976127685e-08,
+      "loss": 0.2222,
+      "step": 11298
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.454495584867216e-08,
+      "loss": 0.2002,
+      "step": 11299
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.413058394467431e-08,
+      "loss": 0.2097,
+      "step": 11300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.371736408126606e-08,
+      "loss": 0.2505,
+      "step": 11301
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.330529629034022e-08,
+      "loss": 0.2402,
+      "step": 11302
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.289438060370302e-08,
+      "loss": 0.2123,
+      "step": 11303
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.248461705307075e-08,
+      "loss": 0.2302,
+      "step": 11304
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.207600567006977e-08,
+      "loss": 0.2305,
+      "step": 11305
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.166854648623988e-08,
+      "loss": 0.2478,
+      "step": 11306
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.126223953302924e-08,
+      "loss": 0.2123,
+      "step": 11307
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.085708484179942e-08,
+      "loss": 0.2369,
+      "step": 11308
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.045308244382043e-08,
+      "loss": 0.2244,
+      "step": 11309
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.00502323702773e-08,
+      "loss": 0.2152,
+      "step": 11310
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.964853465226018e-08,
+      "loss": 0.2066,
+      "step": 11311
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.924798932077592e-08,
+      "loss": 0.2619,
+      "step": 11312
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.88485964067398e-08,
+      "loss": 0.2367,
+      "step": 11313
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.845035594098048e-08,
+      "loss": 0.1889,
+      "step": 11314
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.80532679542334e-08,
+      "loss": 0.2745,
+      "step": 11315
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.765733247714734e-08,
+      "loss": 0.2232,
+      "step": 11316
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.726254954028455e-08,
+      "loss": 0.2629,
+      "step": 11317
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.686891917411397e-08,
+      "loss": 0.2356,
+      "step": 11318
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.647644140901632e-08,
+      "loss": 0.2384,
+      "step": 11319
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.608511627528901e-08,
+      "loss": 0.2582,
+      "step": 11320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.56949438031329e-08,
+      "loss": 0.2226,
+      "step": 11321
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.530592402266389e-08,
+      "loss": 0.2543,
+      "step": 11322
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.491805696390963e-08,
+      "loss": 0.2482,
+      "step": 11323
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.45313426568045e-08,
+      "loss": 0.2277,
+      "step": 11324
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.414578113119795e-08,
+      "loss": 0.2538,
+      "step": 11325
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.376137241684954e-08,
+      "loss": 0.2204,
+      "step": 11326
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.337811654342885e-08,
+      "loss": 0.209,
+      "step": 11327
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.299601354051887e-08,
+      "loss": 0.2176,
+      "step": 11328
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.261506343760937e-08,
+      "loss": 0.2254,
+      "step": 11329
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.223526626410514e-08,
+      "loss": 0.2254,
+      "step": 11330
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.185662204932108e-08,
+      "loss": 0.2096,
+      "step": 11331
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.147913082248213e-08,
+      "loss": 0.2243,
+      "step": 11332
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.1102792612725e-08,
+      "loss": 0.2223,
+      "step": 11333
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.072760744909478e-08,
+      "loss": 0.2167,
+      "step": 11334
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.035357536055164e-08,
+      "loss": 0.1947,
+      "step": 11335
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.998069637596415e-08,
+      "loss": 0.2203,
+      "step": 11336
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.960897052411429e-08,
+      "loss": 0.2356,
+      "step": 11337
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9238397833690784e-08,
+      "loss": 0.1991,
+      "step": 11338
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8868978333299076e-08,
+      "loss": 0.2393,
+      "step": 11339
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.850071205144969e-08,
+      "loss": 0.1951,
+      "step": 11340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8133599016568227e-08,
+      "loss": 0.2406,
+      "step": 11341
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.776763925699036e-08,
+      "loss": 0.2172,
+      "step": 11342
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7402832800961813e-08,
+      "loss": 0.2094,
+      "step": 11343
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7039179676640074e-08,
+      "loss": 0.199,
+      "step": 11344
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.6676679912094354e-08,
+      "loss": 0.2152,
+      "step": 11345
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.6315333535300605e-08,
+      "loss": 0.2276,
+      "step": 11346
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.595514057415318e-08,
+      "loss": 0.2331,
+      "step": 11347
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.5596101056449855e-08,
+      "loss": 0.2322,
+      "step": 11348
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.523821500990511e-08,
+      "loss": 0.2598,
+      "step": 11349
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4881482462141866e-08,
+      "loss": 0.2498,
+      "step": 11350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.4525903440693084e-08,
+      "loss": 0.2489,
+      "step": 11351
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.417147797300348e-08,
+      "loss": 0.2181,
+      "step": 11352
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.3818206086431174e-08,
+      "loss": 0.2364,
+      "step": 11353
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.3466087808241024e-08,
+      "loss": 0.2643,
+      "step": 11354
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.311512316561295e-08,
+      "loss": 0.1936,
+      "step": 11355
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2765312185635294e-08,
+      "loss": 0.2288,
+      "step": 11356
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.241665489530645e-08,
+      "loss": 0.2219,
+      "step": 11357
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.206915132153822e-08,
+      "loss": 0.217,
+      "step": 11358
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1722801491152497e-08,
+      "loss": 0.2367,
+      "step": 11359
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1377605430881215e-08,
+      "loss": 0.2201,
+      "step": 11360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.1033563167369735e-08,
+      "loss": 0.2217,
+      "step": 11361
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.069067472717015e-08,
+      "loss": 0.2437,
+      "step": 11362
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.034894013675129e-08,
+      "loss": 0.2011,
+      "step": 11363
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.000835942248705e-08,
+      "loss": 0.2182,
+      "step": 11364
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.966893261066641e-08,
+      "loss": 0.2202,
+      "step": 11365
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.9330659727486735e-08,
+      "loss": 0.2206,
+      "step": 11366
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.8993540799057154e-08,
+      "loss": 0.233,
+      "step": 11367
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.865757585139852e-08,
+      "loss": 0.2481,
+      "step": 11368
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.832276491044174e-08,
+      "loss": 0.2021,
+      "step": 11369
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.798910800202949e-08,
+      "loss": 0.2622,
+      "step": 11370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.76566051519145e-08,
+      "loss": 0.229,
+      "step": 11371
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7325256385759574e-08,
+      "loss": 0.2194,
+      "step": 11372
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.699506172914092e-08,
+      "loss": 0.2337,
+      "step": 11373
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.6666021207544816e-08,
+      "loss": 0.2239,
+      "step": 11374
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.633813484636762e-08,
+      "loss": 0.2016,
+      "step": 11375
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.601140267091408e-08,
+      "loss": 0.2228,
+      "step": 11376
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5685824706407366e-08,
+      "loss": 0.2278,
+      "step": 11377
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.536140097797403e-08,
+      "loss": 0.2354,
+      "step": 11378
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5038131510654056e-08,
+      "loss": 0.202,
+      "step": 11379
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.471601632940081e-08,
+      "loss": 0.2462,
+      "step": 11380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.439505545907441e-08,
+      "loss": 0.2402,
+      "step": 11381
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.407524892445003e-08,
+      "loss": 0.2141,
+      "step": 11382
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.375659675021126e-08,
+      "loss": 0.2368,
+      "step": 11383
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.343909896095177e-08,
+      "loss": 0.223,
+      "step": 11384
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.312275558117695e-08,
+      "loss": 0.2145,
+      "step": 11385
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.28075666353056e-08,
+      "loss": 0.2298,
+      "step": 11386
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.249353214766327e-08,
+      "loss": 0.2443,
+      "step": 11387
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.218065214249056e-08,
+      "loss": 0.2004,
+      "step": 11388
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.186892664393649e-08,
+      "loss": 0.2571,
+      "step": 11389
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.155835567605848e-08,
+      "loss": 0.2256,
+      "step": 11390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.124893926283069e-08,
+      "loss": 0.2019,
+      "step": 11391
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.094067742813401e-08,
+      "loss": 0.2109,
+      "step": 11392
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.063357019576108e-08,
+      "loss": 0.2159,
+      "step": 11393
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0327617589417944e-08,
+      "loss": 0.2454,
+      "step": 11394
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0022819632715704e-08,
+      "loss": 0.2265,
+      "step": 11395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.971917634918221e-08,
+      "loss": 0.237,
+      "step": 11396
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.941668776225371e-08,
+      "loss": 0.2212,
+      "step": 11397
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.9115353895278203e-08,
+      "loss": 0.2121,
+      "step": 11398
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.881517477151209e-08,
+      "loss": 0.2382,
+      "step": 11399
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.851615041412515e-08,
+      "loss": 0.2226,
+      "step": 11400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.821828084619727e-08,
+      "loss": 0.2036,
+      "step": 11401
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.792156609072006e-08,
+      "loss": 0.2458,
+      "step": 11402
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.762600617059353e-08,
+      "loss": 0.2233,
+      "step": 11403
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7331601108632765e-08,
+      "loss": 0.2232,
+      "step": 11404
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7038350927559585e-08,
+      "loss": 0.2332,
+      "step": 11405
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.6746255650007554e-08,
+      "loss": 0.2255,
+      "step": 11406
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.645531529852364e-08,
+      "loss": 0.2379,
+      "step": 11407
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.616552989556321e-08,
+      "loss": 0.2119,
+      "step": 11408
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.5876899463490045e-08,
+      "loss": 0.2147,
+      "step": 11409
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.558942402458798e-08,
+      "loss": 0.2417,
+      "step": 11410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.530310360103928e-08,
+      "loss": 0.2323,
+      "step": 11411
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.501793821494792e-08,
+      "loss": 0.2469,
+      "step": 11412
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.47339278883213e-08,
+      "loss": 0.2038,
+      "step": 11413
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.445107264308356e-08,
+      "loss": 0.2085,
+      "step": 11414
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.416937250106223e-08,
+      "loss": 0.2222,
+      "step": 11415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.3888827484004904e-08,
+      "loss": 0.2157,
+      "step": 11416
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.3609437613560945e-08,
+      "loss": 0.2189,
+      "step": 11417
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.33312029112981e-08,
+      "loss": 0.2489,
+      "step": 11418
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.3054123398689186e-08,
+      "loss": 0.204,
+      "step": 11419
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.27781990971221e-08,
+      "loss": 0.2087,
+      "step": 11420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2503430027893134e-08,
+      "loss": 0.236,
+      "step": 11421
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2229816212210326e-08,
+      "loss": 0.2252,
+      "step": 11422
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.195735767119179e-08,
+      "loss": 0.2139,
+      "step": 11423
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.168605442586736e-08,
+      "loss": 0.2486,
+      "step": 11424
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.1415906497178626e-08,
+      "loss": 0.2332,
+      "step": 11425
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.11469139059739e-08,
+      "loss": 0.2327,
+      "step": 11426
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.087907667301826e-08,
+      "loss": 0.24,
+      "step": 11427
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.061239481898181e-08,
+      "loss": 0.2491,
+      "step": 11428
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0346868364449774e-08,
+      "loss": 0.2236,
+      "step": 11429
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.008249732991575e-08,
+      "loss": 0.2349,
+      "step": 11430
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.981928173578674e-08,
+      "loss": 0.2436,
+      "step": 11431
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9557221602376506e-08,
+      "loss": 0.1967,
+      "step": 11432
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.9296316949912194e-08,
+      "loss": 0.2294,
+      "step": 11433
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.90365677985327e-08,
+      "loss": 0.2151,
+      "step": 11434
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.877797416828698e-08,
+      "loss": 0.2258,
+      "step": 11435
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8520536079132407e-08,
+      "loss": 0.256,
+      "step": 11436
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8264253550941422e-08,
+      "loss": 0.2162,
+      "step": 11437
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8009126603493197e-08,
+      "loss": 0.1959,
+      "step": 11438
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.7755155256478647e-08,
+      "loss": 0.2178,
+      "step": 11439
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.750233952950376e-08,
+      "loss": 0.2484,
+      "step": 11440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.725067944207793e-08,
+      "loss": 0.2109,
+      "step": 11441
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.700017501362895e-08,
+      "loss": 0.2179,
+      "step": 11442
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6750826263488015e-08,
+      "loss": 0.2215,
+      "step": 11443
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6502633210904737e-08,
+      "loss": 0.2211,
+      "step": 11444
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.625559587503379e-08,
+      "loss": 0.2055,
+      "step": 11445
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6009714274941588e-08,
+      "loss": 0.2122,
+      "step": 11446
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.576498842960795e-08,
+      "loss": 0.2361,
+      "step": 11447
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.55214183579211e-08,
+      "loss": 0.2021,
+      "step": 11448
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5279004078681002e-08,
+      "loss": 0.2152,
+      "step": 11449
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.503774561059935e-08,
+      "loss": 0.2085,
+      "step": 11450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4797642972294588e-08,
+      "loss": 0.1876,
+      "step": 11451
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4558696182300222e-08,
+      "loss": 0.2133,
+      "step": 11452
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.43209052590615e-08,
+      "loss": 0.212,
+      "step": 11453
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4084270220928737e-08,
+      "loss": 0.2289,
+      "step": 11454
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3848791086167332e-08,
+      "loss": 0.21,
+      "step": 11455
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3614467872954404e-08,
+      "loss": 0.2084,
+      "step": 11456
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3381300599373824e-08,
+      "loss": 0.2272,
+      "step": 11457
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3149289283421192e-08,
+      "loss": 0.2204,
+      "step": 11458
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2918433943008854e-08,
+      "loss": 0.1903,
+      "step": 11459
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.268873459595089e-08,
+      "loss": 0.2183,
+      "step": 11460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2460191259978114e-08,
+      "loss": 0.2249,
+      "step": 11461
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2232803952729753e-08,
+      "loss": 0.2209,
+      "step": 11462
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.200657269175843e-08,
+      "loss": 0.2108,
+      "step": 11463
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1781497494523518e-08,
+      "loss": 0.2026,
+      "step": 11464
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.155757837839778e-08,
+      "loss": 0.2421,
+      "step": 11465
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.133481536066406e-08,
+      "loss": 0.2129,
+      "step": 11466
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.1113208458516942e-08,
+      "loss": 0.2246,
+      "step": 11467
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0892757689059406e-08,
+      "loss": 0.2195,
+      "step": 11468
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0673463069309508e-08,
+      "loss": 0.2434,
+      "step": 11469
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0455324616190374e-08,
+      "loss": 0.2316,
+      "step": 11470
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0238342346540206e-08,
+      "loss": 0.2484,
+      "step": 11471
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.0022516277107273e-08,
+      "loss": 0.1727,
+      "step": 11472
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9807846424548247e-08,
+      "loss": 0.2244,
+      "step": 11473
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9594332805433214e-08,
+      "loss": 0.2152,
+      "step": 11474
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9381975436242316e-08,
+      "loss": 0.2039,
+      "step": 11475
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9170774333365782e-08,
+      "loss": 0.2272,
+      "step": 11476
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8960729513105567e-08,
+      "loss": 0.247,
+      "step": 11477
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8751840991672042e-08,
+      "loss": 0.2234,
+      "step": 11478
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.854410878519064e-08,
+      "loss": 0.2519,
+      "step": 11479
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8337532909691867e-08,
+      "loss": 0.1978,
+      "step": 11480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8132113381122974e-08,
+      "loss": 0.2295,
+      "step": 11481
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.792785021533794e-08,
+      "loss": 0.2304,
+      "step": 11482
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7724743428102484e-08,
+      "loss": 0.2354,
+      "step": 11483
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7522793035094075e-08,
+      "loss": 0.2543,
+      "step": 11484
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7321999051898573e-08,
+      "loss": 0.1913,
+      "step": 11485
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7122361494016914e-08,
+      "loss": 0.2324,
+      "step": 11486
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.692388037685344e-08,
+      "loss": 0.2135,
+      "step": 11487
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6726555715732562e-08,
+      "loss": 0.2088,
+      "step": 11488
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.653038752588043e-08,
+      "loss": 0.2488,
+      "step": 11489
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.63353758224416e-08,
+      "loss": 0.2335,
+      "step": 11490
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6141520620465698e-08,
+      "loss": 0.2515,
+      "step": 11491
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5948821934915757e-08,
+      "loss": 0.2211,
+      "step": 11492
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5757279780666546e-08,
+      "loss": 0.2227,
+      "step": 11493
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.556689417249957e-08,
+      "loss": 0.2162,
+      "step": 11494
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5377665125111408e-08,
+      "loss": 0.2163,
+      "step": 11495
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5189592653105377e-08,
+      "loss": 0.2298,
+      "step": 11496
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5002676771001534e-08,
+      "loss": 0.2215,
+      "step": 11497
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4816917493223336e-08,
+      "loss": 0.2294,
+      "step": 11498
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4632314834109316e-08,
+      "loss": 0.2465,
+      "step": 11499
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4448868807908077e-08,
+      "loss": 0.222,
+      "step": 11500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4266579428779957e-08,
+      "loss": 0.2348,
+      "step": 11501
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4085446710793703e-08,
+      "loss": 0.1896,
+      "step": 11502
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3905470667929798e-08,
+      "loss": 0.2254,
+      "step": 11503
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3726651314078797e-08,
+      "loss": 0.2569,
+      "step": 11504
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.354898866304466e-08,
+      "loss": 0.2232,
+      "step": 11505
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3372482728538082e-08,
+      "loss": 0.2449,
+      "step": 11506
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.319713352418317e-08,
+      "loss": 0.216,
+      "step": 11507
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.302294106351576e-08,
+      "loss": 0.2655,
+      "step": 11508
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2849905359978432e-08,
+      "loss": 0.2052,
+      "step": 11509
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2678026426928835e-08,
+      "loss": 0.2565,
+      "step": 11510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2507304277631359e-08,
+      "loss": 0.2719,
+      "step": 11511
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2337738925265463e-08,
+      "loss": 0.2433,
+      "step": 11512
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2169330382915677e-08,
+      "loss": 0.2218,
+      "step": 11513
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2002078663583272e-08,
+      "loss": 0.2145,
+      "step": 11514
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1835983780174586e-08,
+      "loss": 0.2071,
+      "step": 11515
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1671045745512699e-08,
+      "loss": 0.2041,
+      "step": 11516
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1507264572327425e-08,
+      "loss": 0.2015,
+      "step": 11517
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.134464027325699e-08,
+      "loss": 0.2733,
+      "step": 11518
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1183172860858016e-08,
+      "loss": 0.2331,
+      "step": 11519
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1022862347590534e-08,
+      "loss": 0.1931,
+      "step": 11520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0863708745827983e-08,
+      "loss": 0.1825,
+      "step": 11521
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0705712067855532e-08,
+      "loss": 0.2297,
+      "step": 11522
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.054887232586843e-08,
+      "loss": 0.2216,
+      "step": 11523
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0393189531970326e-08,
+      "loss": 0.2348,
+      "step": 11524
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0238663698178274e-08,
+      "loss": 0.1958,
+      "step": 11525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0085294836419401e-08,
+      "loss": 0.2509,
+      "step": 11526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.933082958530903e-09,
+      "loss": 0.2383,
+      "step": 11527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.782028076261718e-09,
+      "loss": 0.1882,
+      "step": 11528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.63213020127085e-09,
+      "loss": 0.2154,
+      "step": 11529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.483389345129045e-09,
+      "loss": 0.232,
+      "step": 11530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.335805519315454e-09,
+      "loss": 0.2144,
+      "step": 11531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.189378735219301e-09,
+      "loss": 0.2726,
+      "step": 11532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.044109004146539e-09,
+      "loss": 0.2352,
+      "step": 11533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.899996337306538e-09,
+      "loss": 0.2625,
+      "step": 11534
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.75704074582373e-09,
+      "loss": 0.2036,
+      "step": 11535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.615242240732624e-09,
+      "loss": 0.228,
+      "step": 11536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.47460083297613e-09,
+      "loss": 0.2767,
+      "step": 11537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.335116533412234e-09,
+      "loss": 0.2285,
+      "step": 11538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.19678935280399e-09,
+      "loss": 0.2135,
+      "step": 11539
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.059619301829523e-09,
+      "loss": 0.2186,
+      "step": 11540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.923606391075367e-09,
+      "loss": 0.2244,
+      "step": 11541
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.788750631039786e-09,
+      "loss": 0.231,
+      "step": 11542
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.655052032132792e-09,
+      "loss": 0.2197,
+      "step": 11543
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.522510604672795e-09,
+      "loss": 0.2309,
+      "step": 11544
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.3911263588882786e-09,
+      "loss": 0.2219,
+      "step": 11545
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.260899304924462e-09,
+      "loss": 0.2554,
+      "step": 11546
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.131829452828309e-09,
+      "loss": 0.2241,
+      "step": 11547
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0039168125635154e-09,
+      "loss": 0.2294,
+      "step": 11548
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.87716139400385e-09,
+      "loss": 0.2349,
+      "step": 11549
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.751563206931488e-09,
+      "loss": 0.225,
+      "step": 11550
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.627122261042007e-09,
+      "loss": 0.2287,
+      "step": 11551
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.503838565937725e-09,
+      "loss": 0.2248,
+      "step": 11552
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.381712131137696e-09,
+      "loss": 0.2229,
+      "step": 11553
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.260742966066047e-09,
+      "loss": 0.2354,
+      "step": 11554
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.140931080060308e-09,
+      "loss": 0.2327,
+      "step": 11555
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.022276482366418e-09,
+      "loss": 0.2607,
+      "step": 11556
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.9047791821453814e-09,
+      "loss": 0.2096,
+      "step": 11557
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.788439188463279e-09,
+      "loss": 0.2398,
+      "step": 11558
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.673256510301261e-09,
+      "loss": 0.2186,
+      "step": 11559
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.559231156550548e-09,
+      "loss": 0.232,
+      "step": 11560
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.446363136010768e-09,
+      "loss": 0.2182,
+      "step": 11561
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.334652457393285e-09,
+      "loss": 0.21,
+      "step": 11562
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.224099129322868e-09,
+      "loss": 0.2243,
+      "step": 11563
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.1147031603276935e-09,
+      "loss": 0.211,
+      "step": 11564
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.006464558856005e-09,
+      "loss": 0.2414,
+      "step": 11565
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.899383333259455e-09,
+      "loss": 0.257,
+      "step": 11566
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.793459491804764e-09,
+      "loss": 0.2239,
+      "step": 11567
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.688693042667058e-09,
+      "loss": 0.2197,
+      "step": 11568
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.585083993931538e-09,
+      "loss": 0.2251,
+      "step": 11569
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.482632353595139e-09,
+      "loss": 0.2206,
+      "step": 11570
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.381338129566537e-09,
+      "loss": 0.2128,
+      "step": 11571
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2812013296644744e-09,
+      "loss": 0.2124,
+      "step": 11572
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.182221961617772e-09,
+      "loss": 0.2253,
+      "step": 11573
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.084400033065316e-09,
+      "loss": 0.2277,
+      "step": 11574
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.987735551556071e-09,
+      "loss": 0.2467,
+      "step": 11575
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.892228524554064e-09,
+      "loss": 0.2463,
+      "step": 11576
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.797878959430068e-09,
+      "loss": 0.2514,
+      "step": 11577
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7046868634649235e-09,
+      "loss": 0.21,
+      "step": 11578
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.612652243851211e-09,
+      "loss": 0.2368,
+      "step": 11579
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.5217751076949133e-09,
+      "loss": 0.217,
+      "step": 11580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4320554620087542e-09,
+      "loss": 0.1989,
+      "step": 11581
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3434933137188593e-09,
+      "loss": 0.233,
+      "step": 11582
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.2560886696597626e-09,
+      "loss": 0.2022,
+      "step": 11583
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.1698415365777335e-09,
+      "loss": 0.2303,
+      "step": 11584
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0847519211307795e-09,
+      "loss": 0.2257,
+      "step": 11585
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.00081982988365e-09,
+      "loss": 0.1908,
+      "step": 11586
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.9180452693178263e-09,
+      "loss": 0.2382,
+      "step": 11587
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8364282458215316e-09,
+      "loss": 0.2503,
+      "step": 11588
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.755968765691397e-09,
+      "loss": 0.2053,
+      "step": 11589
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6766668351424493e-09,
+      "loss": 0.284,
+      "step": 11590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5985224602914637e-09,
+      "loss": 0.2092,
+      "step": 11591
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5215356471719462e-09,
+      "loss": 0.2777,
+      "step": 11592
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4457064017258113e-09,
+      "loss": 0.2236,
+      "step": 11593
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.371034729805044e-09,
+      "loss": 0.2334,
+      "step": 11594
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2975206371750323e-09,
+      "loss": 0.218,
+      "step": 11595
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2251641295062408e-09,
+      "loss": 0.2394,
+      "step": 11596
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.153965212387532e-09,
+      "loss": 0.1802,
+      "step": 11597
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0839238913128445e-09,
+      "loss": 0.226,
+      "step": 11598
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.015040171687854e-09,
+      "loss": 0.1966,
+      "step": 11599
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.947314058828309e-09,
+      "loss": 0.1993,
+      "step": 11600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.880745557963359e-09,
+      "loss": 0.2288,
+      "step": 11601
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8153346742305622e-09,
+      "loss": 0.226,
+      "step": 11602
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7510814126775464e-09,
+      "loss": 0.2519,
+      "step": 11603
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.687985778265344e-09,
+      "loss": 0.2211,
+      "step": 11604
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.626047775863393e-09,
+      "loss": 0.2286,
+      "step": 11605
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5652674102528687e-09,
+      "loss": 0.2222,
+      "step": 11606
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5056446861216878e-09,
+      "loss": 0.2345,
+      "step": 11607
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4471796080761657e-09,
+      "loss": 0.2128,
+      "step": 11608
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3898721806276937e-09,
+      "loss": 0.2298,
+      "step": 11609
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.333722408197735e-09,
+      "loss": 0.2527,
+      "step": 11610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2787302951211554e-09,
+      "loss": 0.209,
+      "step": 11611
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2248958456428926e-09,
+      "loss": 0.2761,
+      "step": 11612
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1722190639179563e-09,
+      "loss": 0.206,
+      "step": 11613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1206999540114282e-09,
+      "loss": 0.2244,
+      "step": 11614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0703385199017922e-09,
+      "loss": 0.2016,
+      "step": 11615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0211347654726088e-09,
+      "loss": 0.2414,
+      "step": 11616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.730886945258366e-10,
+      "loss": 0.2447,
+      "step": 11617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.262003107668448e-10,
+      "loss": 0.2277,
+      "step": 11618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.804696178160709e-10,
+      "loss": 0.2489,
+      "step": 11619
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.358966192023587e-10,
+      "loss": 0.2442,
+      "step": 11620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.924813183662894e-10,
+      "loss": 0.2058,
+      "step": 11621
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.50223718658516e-10,
+      "loss": 0.2054,
+      "step": 11622
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.091238233414287e-10,
+      "loss": 0.2136,
+      "step": 11623
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.691816355874902e-10,
+      "loss": 0.2233,
+      "step": 11624
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.303971584792345e-10,
+      "loss": 0.2218,
+      "step": 11625
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.927703950092677e-10,
+      "loss": 0.2122,
+      "step": 11626
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.563013480835988e-10,
+      "loss": 0.2488,
+      "step": 11627
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.209900205149776e-10,
+      "loss": 0.1929,
+      "step": 11628
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.868364150295568e-10,
+      "loss": 0.2209,
+      "step": 11629
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.5384053426522633e-10,
+      "loss": 0.198,
+      "step": 11630
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.2200238076661734e-10,
+      "loss": 0.2289,
+      "step": 11631
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9132195699176367e-10,
+      "loss": 0.2245,
+      "step": 11632
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.617992653087709e-10,
+      "loss": 0.218,
+      "step": 11633
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.334343079974822e-10,
+      "loss": 0.2494,
+      "step": 11634
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0622708724448167e-10,
+      "loss": 0.2314,
+      "step": 11635
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.801776051530869e-10,
+      "loss": 0.2377,
+      "step": 11636
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.552858637300259e-10,
+      "loss": 0.2616,
+      "step": 11637
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3155186490042558e-10,
+      "loss": 0.237,
+      "step": 11638
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0897561049448844e-10,
+      "loss": 0.2631,
+      "step": 11639
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8755710225415447e-10,
+      "loss": 0.2016,
+      "step": 11640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6729634183310083e-10,
+      "loss": 0.2013,
+      "step": 11641
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4819333079507668e-10,
+      "loss": 0.1955,
+      "step": 11642
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.302480706139031e-10,
+      "loss": 0.2237,
+      "step": 11643
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1346056267680372e-10,
+      "loss": 0.2084,
+      "step": 11644
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.78308082777435e-11,
+      "loss": 0.2278,
+      "step": 11645
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.335880862242462e-11,
+      "loss": 0.2292,
+      "step": 11646
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.004456482995192e-11,
+      "loss": 0.2346,
+      "step": 11647
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.788807792617146e-11,
+      "loss": 0.2379,
+      "step": 11648
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6889348850331916e-11,
+      "loss": 0.2016,
+      "step": 11649
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7048378451753904e-11,
+      "loss": 0.1929,
+      "step": 11650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.8365167488164645e-11,
+      "loss": 0.193,
+      "step": 11651
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0839716632359283e-11,
+      "loss": 0.2284,
+      "step": 11652
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.447202646220891e-11,
+      "loss": 0.2315,
+      "step": 11653
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.262097472317877e-12,
+      "loss": 0.1927,
+      "step": 11654
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.209930062366474e-12,
+      "loss": 0.2246,
+      "step": 11655
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.315524547102932e-12,
+      "loss": 0.2146,
+      "step": 11656
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.788811480167411e-13,
+      "loss": 0.2392,
+      "step": 11657
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.2258,
+      "step": 11658
+    },
+    {
+      "epoch": 2.0,
+      "step": 11658,
+      "total_flos": 6261095013023744.0,
+      "train_loss": 0.3254167705679721,
+      "train_runtime": 122351.6027,
+      "train_samples_per_second": 12.197,
+      "train_steps_per_second": 0.095
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11658,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 50000,
+  "total_flos": 6261095013023744.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}