diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,93256 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 15540,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428575e-08,
+      "loss": 1.0405,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-08,
+      "loss": 1.0125,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428573e-08,
+      "loss": 1.0108,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.142857142857143e-07,
+      "loss": 1.0151,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4285714285714287e-07,
+      "loss": 0.9754,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7142857142857146e-07,
+      "loss": 1.0091,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.0125,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.285714285714286e-07,
+      "loss": 1.0404,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5714285714285716e-07,
+      "loss": 1.0334,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428575e-07,
+      "loss": 1.022,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1428571428571433e-07,
+      "loss": 1.0428,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.428571428571429e-07,
+      "loss": 0.9879,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7142857142857145e-07,
+      "loss": 0.9653,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.0869,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.285714285714286e-07,
+      "loss": 1.0539,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.571428571428572e-07,
+      "loss": 1.0299,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.857142857142857e-07,
+      "loss": 0.9906,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.142857142857143e-07,
+      "loss": 1.0332,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.428571428571429e-07,
+      "loss": 1.0768,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 1.0153,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.0794,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.285714285714287e-07,
+      "loss": 0.9556,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.571428571428571e-07,
+      "loss": 0.935,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.857142857142858e-07,
+      "loss": 1.0366,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.0009,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.428571428571429e-07,
+      "loss": 0.9534,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.714285714285715e-07,
+      "loss": 0.9825,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.9897,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.285714285714285e-07,
+      "loss": 0.9654,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.571428571428572e-07,
+      "loss": 0.9999,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.857142857142857e-07,
+      "loss": 0.9536,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.142857142857144e-07,
+      "loss": 0.9763,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.42857142857143e-07,
+      "loss": 0.9416,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.714285714285715e-07,
+      "loss": 0.8997,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.8878,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0285714285714286e-06,
+      "loss": 0.9336,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0571428571428573e-06,
+      "loss": 0.8653,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0857142857142858e-06,
+      "loss": 0.9067,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1142857142857145e-06,
+      "loss": 0.9404,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 0.8989,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1714285714285715e-06,
+      "loss": 0.8679,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.9055,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2285714285714286e-06,
+      "loss": 0.8636,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2571428571428573e-06,
+      "loss": 0.8272,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2857142857142856e-06,
+      "loss": 0.9012,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3142857142857143e-06,
+      "loss": 0.8983,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.342857142857143e-06,
+      "loss": 0.9004,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3714285714285717e-06,
+      "loss": 0.9362,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.9066,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.8758,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4571428571428573e-06,
+      "loss": 0.9263,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4857142857142858e-06,
+      "loss": 0.8375,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5142857142857145e-06,
+      "loss": 0.827,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.542857142857143e-06,
+      "loss": 0.8135,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5714285714285714e-06,
+      "loss": 0.8733,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.8195,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6285714285714288e-06,
+      "loss": 0.858,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.657142857142857e-06,
+      "loss": 0.8579,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6857142857142858e-06,
+      "loss": 0.7985,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 0.79,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7428571428571432e-06,
+      "loss": 0.8097,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7714285714285714e-06,
+      "loss": 0.8172,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.8506,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8285714285714288e-06,
+      "loss": 0.8328,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8571428571428573e-06,
+      "loss": 0.7951,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.885714285714286e-06,
+      "loss": 0.8287,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9142857142857145e-06,
+      "loss": 0.8194,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.942857142857143e-06,
+      "loss": 0.8516,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9714285714285714e-06,
+      "loss": 0.7728,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.8552,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.028571428571429e-06,
+      "loss": 0.7961,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0571428571428573e-06,
+      "loss": 0.7996,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0857142857142858e-06,
+      "loss": 0.7685,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1142857142857147e-06,
+      "loss": 0.8025,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.786,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1714285714285716e-06,
+      "loss": 0.82,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 0.761,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.228571428571429e-06,
+      "loss": 0.8316,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.257142857142857e-06,
+      "loss": 0.7888,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 0.8038,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3142857142857145e-06,
+      "loss": 0.8439,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.342857142857143e-06,
+      "loss": 0.774,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.371428571428572e-06,
+      "loss": 0.7688,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.8152,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.428571428571429e-06,
+      "loss": 0.7878,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4571428571428573e-06,
+      "loss": 0.8003,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.485714285714286e-06,
+      "loss": 0.7731,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5142857142857147e-06,
+      "loss": 0.7897,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.542857142857143e-06,
+      "loss": 0.8036,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": 0.7828,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-06,
+      "loss": 0.7528,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6285714285714286e-06,
+      "loss": 0.7635,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6571428571428575e-06,
+      "loss": 0.7347,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.685714285714286e-06,
+      "loss": 0.8061,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7142857142857144e-06,
+      "loss": 0.8143,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7428571428571433e-06,
+      "loss": 0.7667,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7714285714285714e-06,
+      "loss": 0.7544,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.7236,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8285714285714288e-06,
+      "loss": 0.7459,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.8146,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.885714285714286e-06,
+      "loss": 0.7525,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9142857142857146e-06,
+      "loss": 7.246,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9428571428571427e-06,
+      "loss": 0.7448,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9714285714285716e-06,
+      "loss": 0.7562,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 0.7381,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.028571428571429e-06,
+      "loss": 0.7634,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0571428571428575e-06,
+      "loss": 0.7569,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.085714285714286e-06,
+      "loss": 0.7627,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.114285714285715e-06,
+      "loss": 0.752,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.142857142857143e-06,
+      "loss": 0.7671,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1714285714285714e-06,
+      "loss": 0.7822,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.7219,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2285714285714288e-06,
+      "loss": 0.7468,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2571428571428577e-06,
+      "loss": 0.7535,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.285714285714286e-06,
+      "loss": 0.7548,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.314285714285714e-06,
+      "loss": 0.8188,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.342857142857143e-06,
+      "loss": 0.8135,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3714285714285716e-06,
+      "loss": 0.7663,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.7124,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 0.732,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4571428571428574e-06,
+      "loss": 0.7691,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4857142857142863e-06,
+      "loss": 0.7477,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5142857142857144e-06,
+      "loss": 0.7457,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.542857142857143e-06,
+      "loss": 0.7392,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.7647,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.7597,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.628571428571429e-06,
+      "loss": 0.7084,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6571428571428576e-06,
+      "loss": 0.7179,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6857142857142857e-06,
+      "loss": 0.7796,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7142857142857146e-06,
+      "loss": 0.7288,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.742857142857143e-06,
+      "loss": 0.739,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.771428571428572e-06,
+      "loss": 0.7357,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.7886,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.828571428571429e-06,
+      "loss": 0.7777,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.857142857142858e-06,
+      "loss": 0.7532,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885714285714286e-06,
+      "loss": 0.719,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.914285714285714e-06,
+      "loss": 0.7642,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.942857142857143e-06,
+      "loss": 5.532,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.971428571428572e-06,
+      "loss": 0.6878,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.765,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.028571428571429e-06,
+      "loss": 0.7516,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.057142857142858e-06,
+      "loss": 0.7507,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0857142857142865e-06,
+      "loss": 0.7205,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.114285714285715e-06,
+      "loss": 0.765,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1428571428571435e-06,
+      "loss": 0.7489,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1714285714285715e-06,
+      "loss": 0.7791,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.7277,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.228571428571429e-06,
+      "loss": 0.7328,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.257142857142857e-06,
+      "loss": 0.7298,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.7511,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.314285714285714e-06,
+      "loss": 0.6983,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.342857142857143e-06,
+      "loss": 0.7227,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.371428571428572e-06,
+      "loss": 0.7267,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4e-06,
+      "loss": 0.7464,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.428571428571429e-06,
+      "loss": 0.7249,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.457142857142858e-06,
+      "loss": 0.7616,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.485714285714286e-06,
+      "loss": 0.7807,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.514285714285714e-06,
+      "loss": 0.7614,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.542857142857143e-06,
+      "loss": 0.7045,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 0.7189,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.7439,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.628571428571429e-06,
+      "loss": 0.7188,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657142857142857e-06,
+      "loss": 0.7227,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.685714285714286e-06,
+      "loss": 0.7228,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.714285714285715e-06,
+      "loss": 0.7316,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.742857142857144e-06,
+      "loss": 0.7095,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.771428571428572e-06,
+      "loss": 0.7217,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.7106,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8285714285714295e-06,
+      "loss": 0.6927,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.857142857142858e-06,
+      "loss": 0.723,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.885714285714286e-06,
+      "loss": 0.7099,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9142857142857145e-06,
+      "loss": 0.7451,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9428571428571435e-06,
+      "loss": 0.7803,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971428571428572e-06,
+      "loss": 0.7222,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.7466,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.028571428571429e-06,
+      "loss": 0.727,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.057142857142857e-06,
+      "loss": 0.7466,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.085714285714286e-06,
+      "loss": 0.7082,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.114285714285715e-06,
+      "loss": 0.7039,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 0.6806,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.171428571428571e-06,
+      "loss": 0.6984,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-06,
+      "loss": 0.6969,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.22857142857143e-06,
+      "loss": 0.7645,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.257142857142857e-06,
+      "loss": 0.7124,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.285714285714286e-06,
+      "loss": 0.7457,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.314285714285715e-06,
+      "loss": 0.7018,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.342857142857143e-06,
+      "loss": 0.7108,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.371428571428572e-06,
+      "loss": 0.7243,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 0.6914,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.428571428571429e-06,
+      "loss": 0.7424,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.457142857142858e-06,
+      "loss": 0.7415,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.485714285714287e-06,
+      "loss": 0.7539,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.514285714285714e-06,
+      "loss": 0.7564,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.542857142857143e-06,
+      "loss": 0.751,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.571428571428572e-06,
+      "loss": 0.6958,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.7406,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.628571428571429e-06,
+      "loss": 0.701,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6571428571428576e-06,
+      "loss": 0.7622,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6857142857142865e-06,
+      "loss": 0.7071,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.7379,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.742857142857143e-06,
+      "loss": 0.6939,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.771428571428572e-06,
+      "loss": 0.7417,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 0.7519,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.828571428571429e-06,
+      "loss": 0.7395,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.857142857142858e-06,
+      "loss": 0.7872,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.885714285714285e-06,
+      "loss": 0.6577,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.914285714285714e-06,
+      "loss": 0.7157,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.942857142857143e-06,
+      "loss": 0.7311,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.971428571428572e-06,
+      "loss": 0.6683,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-06,
+      "loss": 0.6792,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.028571428571429e-06,
+      "loss": 0.713,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.057142857142858e-06,
+      "loss": 0.7138,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.085714285714286e-06,
+      "loss": 0.7348,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.114285714285715e-06,
+      "loss": 0.7065,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.142857142857144e-06,
+      "loss": 0.6845,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.171428571428572e-06,
+      "loss": 0.7065,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.6936,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.22857142857143e-06,
+      "loss": 0.7569,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.257142857142857e-06,
+      "loss": 0.6863,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 0.7118,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.314285714285715e-06,
+      "loss": 0.7013,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.342857142857143e-06,
+      "loss": 0.7267,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.371428571428572e-06,
+      "loss": 0.7191,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.7282,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.7326,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4571428571428575e-06,
+      "loss": 4.3688,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.485714285714286e-06,
+      "loss": 0.6912,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.514285714285715e-06,
+      "loss": 0.6769,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.542857142857143e-06,
+      "loss": 0.7102,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.571428571428572e-06,
+      "loss": 0.6939,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.6933,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.628571428571428e-06,
+      "loss": 0.7026,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.657142857142857e-06,
+      "loss": 0.7142,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.685714285714286e-06,
+      "loss": 0.7391,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.714285714285714e-06,
+      "loss": 0.6919,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.742857142857143e-06,
+      "loss": 0.6774,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.771428571428572e-06,
+      "loss": 0.7303,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 0.7103,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.828571428571429e-06,
+      "loss": 0.7282,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 0.7032,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.885714285714287e-06,
+      "loss": 0.6858,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.914285714285715e-06,
+      "loss": 0.7282,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.942857142857144e-06,
+      "loss": 0.7198,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.971428571428573e-06,
+      "loss": 0.6752,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-06,
+      "loss": 0.7313,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.028571428571429e-06,
+      "loss": 0.7385,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.057142857142858e-06,
+      "loss": 0.7083,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.085714285714286e-06,
+      "loss": 0.6826,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.114285714285715e-06,
+      "loss": 0.7002,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.7016,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1714285714285725e-06,
+      "loss": 0.7159,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.6696,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2285714285714294e-06,
+      "loss": 0.7301,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.257142857142858e-06,
+      "loss": 0.6685,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.285714285714286e-06,
+      "loss": 0.7164,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.314285714285715e-06,
+      "loss": 0.6881,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.342857142857144e-06,
+      "loss": 0.6981,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.371428571428571e-06,
+      "loss": 0.7175,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4e-06,
+      "loss": 0.6909,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 0.6915,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.457142857142857e-06,
+      "loss": 0.7204,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.485714285714286e-06,
+      "loss": 0.6987,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.514285714285715e-06,
+      "loss": 0.6589,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.542857142857144e-06,
+      "loss": 0.7135,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.571428571428572e-06,
+      "loss": 0.6994,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.6806,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.62857142857143e-06,
+      "loss": 0.6737,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.657142857142858e-06,
+      "loss": 0.7524,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.685714285714286e-06,
+      "loss": 0.739,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.714285714285716e-06,
+      "loss": 0.7597,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.742857142857144e-06,
+      "loss": 0.688,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.771428571428572e-06,
+      "loss": 0.6677,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 0.6895,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.828571428571428e-06,
+      "loss": 0.7457,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.7173,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.885714285714286e-06,
+      "loss": 0.69,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.914285714285715e-06,
+      "loss": 0.7008,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.942857142857144e-06,
+      "loss": 0.6885,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.971428571428572e-06,
+      "loss": 0.6639,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7149,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.02857142857143e-06,
+      "loss": 0.7197,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.057142857142857e-06,
+      "loss": 0.7019,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.085714285714287e-06,
+      "loss": 0.7339,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.114285714285715e-06,
+      "loss": 0.6643,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.142857142857143e-06,
+      "loss": 0.6927,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.171428571428573e-06,
+      "loss": 0.69,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.2e-06,
+      "loss": 0.7168,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.22857142857143e-06,
+      "loss": 0.7084,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.257142857142857e-06,
+      "loss": 0.7134,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.285714285714287e-06,
+      "loss": 0.6796,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.314285714285715e-06,
+      "loss": 0.7503,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.342857142857143e-06,
+      "loss": 0.7281,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.371428571428573e-06,
+      "loss": 0.6752,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.6628,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.428571428571429e-06,
+      "loss": 0.6904,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.457142857142859e-06,
+      "loss": 0.6498,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.485714285714287e-06,
+      "loss": 0.7177,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.514285714285715e-06,
+      "loss": 0.7633,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.542857142857145e-06,
+      "loss": 0.7073,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.6921,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.6e-06,
+      "loss": 0.7098,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.628571428571429e-06,
+      "loss": 0.6964,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.657142857142858e-06,
+      "loss": 0.661,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.685714285714287e-06,
+      "loss": 0.7188,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.714285714285715e-06,
+      "loss": 3.1241,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.742857142857144e-06,
+      "loss": 0.6733,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.771428571428572e-06,
+      "loss": 0.6861,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.8e-06,
+      "loss": 0.6856,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.82857142857143e-06,
+      "loss": 0.7139,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.857142857142858e-06,
+      "loss": 0.7331,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.885714285714286e-06,
+      "loss": 0.7211,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.914285714285716e-06,
+      "loss": 0.7067,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.942857142857142e-06,
+      "loss": 0.6631,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.971428571428572e-06,
+      "loss": 0.7084,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-06,
+      "loss": 0.6666,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.028571428571428e-06,
+      "loss": 0.7299,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.057142857142858e-06,
+      "loss": 0.6909,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.085714285714286e-06,
+      "loss": 0.6776,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.114285714285716e-06,
+      "loss": 0.6944,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 0.7051,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.171428571428572e-06,
+      "loss": 0.6923,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.6963,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.22857142857143e-06,
+      "loss": 0.6945,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.257142857142858e-06,
+      "loss": 0.6879,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.7093,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.314285714285714e-06,
+      "loss": 0.7059,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.342857142857144e-06,
+      "loss": 0.6696,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.371428571428572e-06,
+      "loss": 0.7402,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.4e-06,
+      "loss": 0.6631,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.42857142857143e-06,
+      "loss": 0.66,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.457142857142858e-06,
+      "loss": 0.6553,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.485714285714287e-06,
+      "loss": 0.6853,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.514285714285715e-06,
+      "loss": 0.6951,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.542857142857143e-06,
+      "loss": 0.6932,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.571428571428573e-06,
+      "loss": 0.6715,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.7041,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.62857142857143e-06,
+      "loss": 0.7001,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.657142857142859e-06,
+      "loss": 0.6713,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.685714285714285e-06,
+      "loss": 0.715,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 0.6906,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.742857142857143e-06,
+      "loss": 0.7311,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.771428571428571e-06,
+      "loss": 0.736,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.7162,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.828571428571429e-06,
+      "loss": 0.6655,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.857142857142859e-06,
+      "loss": 0.7024,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885714285714287e-06,
+      "loss": 0.6808,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.914285714285715e-06,
+      "loss": 0.6994,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.942857142857145e-06,
+      "loss": 0.72,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971428571428571e-06,
+      "loss": 0.6795,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.6888,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.002857142857143e-05,
+      "loss": 0.6928,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0057142857142859e-05,
+      "loss": 0.6984,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0085714285714288e-05,
+      "loss": 0.7251,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0114285714285715e-05,
+      "loss": 0.6913,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0142857142857143e-05,
+      "loss": 0.6542,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0171428571428573e-05,
+      "loss": 0.7292,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.02e-05,
+      "loss": 0.7341,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.022857142857143e-05,
+      "loss": 0.6753,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0257142857142858e-05,
+      "loss": 0.733,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 0.6979,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0314285714285715e-05,
+      "loss": 0.7273,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0342857142857143e-05,
+      "loss": 0.7086,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0371428571428572e-05,
+      "loss": 0.6882,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.04e-05,
+      "loss": 0.6469,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.042857142857143e-05,
+      "loss": 0.6859,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.045714285714286e-05,
+      "loss": 0.6966,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0485714285714286e-05,
+      "loss": 0.7376,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0514285714285714e-05,
+      "loss": 0.7045,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0542857142857144e-05,
+      "loss": 0.692,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0571428571428572e-05,
+      "loss": 0.7134,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 0.6778,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.062857142857143e-05,
+      "loss": 0.6788,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0657142857142856e-05,
+      "loss": 0.7043,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0685714285714286e-05,
+      "loss": 0.7169,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.7143,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0742857142857144e-05,
+      "loss": 0.7258,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0771428571428572e-05,
+      "loss": 0.6777,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 0.7033,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.082857142857143e-05,
+      "loss": 0.7273,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 0.6976,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0885714285714286e-05,
+      "loss": 0.713,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0914285714285716e-05,
+      "loss": 0.6927,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0942857142857144e-05,
+      "loss": 0.7027,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0971428571428573e-05,
+      "loss": 0.7464,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6789,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1028571428571428e-05,
+      "loss": 0.6981,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1057142857142858e-05,
+      "loss": 0.664,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1085714285714286e-05,
+      "loss": 0.7096,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1114285714285715e-05,
+      "loss": 0.6365,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1142857142857143e-05,
+      "loss": 0.6964,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1171428571428573e-05,
+      "loss": 0.6838,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.7016,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.122857142857143e-05,
+      "loss": 0.6973,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1257142857142857e-05,
+      "loss": 0.7061,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1285714285714287e-05,
+      "loss": 0.7118,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1314285714285715e-05,
+      "loss": 0.731,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1342857142857145e-05,
+      "loss": 0.6802,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1371428571428573e-05,
+      "loss": 0.6823,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.14e-05,
+      "loss": 0.6577,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.6645,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1457142857142857e-05,
+      "loss": 0.6688,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1485714285714287e-05,
+      "loss": 0.7314,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1514285714285715e-05,
+      "loss": 0.7206,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1542857142857145e-05,
+      "loss": 0.7292,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1571428571428573e-05,
+      "loss": 0.7082,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.16e-05,
+      "loss": 0.722,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1628571428571429e-05,
+      "loss": 0.6685,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1657142857142859e-05,
+      "loss": 0.697,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1685714285714287e-05,
+      "loss": 0.6789,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1714285714285716e-05,
+      "loss": 0.6867,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1742857142857144e-05,
+      "loss": 0.7096,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.177142857142857e-05,
+      "loss": 0.7121,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.18e-05,
+      "loss": 0.6933,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1828571428571429e-05,
+      "loss": 0.7229,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1857142857142858e-05,
+      "loss": 0.6866,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1885714285714286e-05,
+      "loss": 0.6322,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1914285714285716e-05,
+      "loss": 0.6801,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1942857142857144e-05,
+      "loss": 0.6864,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1971428571428572e-05,
+      "loss": 0.6538,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.7391,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.202857142857143e-05,
+      "loss": 0.6935,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2057142857142858e-05,
+      "loss": 0.6933,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2085714285714288e-05,
+      "loss": 0.6486,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2114285714285716e-05,
+      "loss": 0.6804,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.7133,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2171428571428572e-05,
+      "loss": 0.6914,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.22e-05,
+      "loss": 0.6955,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.222857142857143e-05,
+      "loss": 0.6808,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2257142857142858e-05,
+      "loss": 0.672,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2285714285714288e-05,
+      "loss": 0.6733,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2314285714285716e-05,
+      "loss": 0.6884,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2342857142857144e-05,
+      "loss": 0.7057,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2371428571428572e-05,
+      "loss": 0.6856,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.6868,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.242857142857143e-05,
+      "loss": 0.6583,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.245714285714286e-05,
+      "loss": 0.7362,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2485714285714287e-05,
+      "loss": 0.6506,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2514285714285714e-05,
+      "loss": 0.6798,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2542857142857144e-05,
+      "loss": 0.6284,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 0.661,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.6822,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.262857142857143e-05,
+      "loss": 0.6826,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2657142857142859e-05,
+      "loss": 0.6648,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2685714285714286e-05,
+      "loss": 0.6376,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2714285714285715e-05,
+      "loss": 0.6751,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2742857142857143e-05,
+      "loss": 0.6646,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2771428571428573e-05,
+      "loss": 0.683,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.6689,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2828571428571431e-05,
+      "loss": 0.73,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.7258,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2885714285714285e-05,
+      "loss": 0.6826,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2914285714285715e-05,
+      "loss": 0.6955,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2942857142857143e-05,
+      "loss": 0.7117,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2971428571428573e-05,
+      "loss": 0.6814,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.6805,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.302857142857143e-05,
+      "loss": 0.6233,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3057142857142857e-05,
+      "loss": 0.7186,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3085714285714287e-05,
+      "loss": 0.6773,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3114285714285715e-05,
+      "loss": 0.6682,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 0.6861,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3171428571428573e-05,
+      "loss": 0.7036,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.6839,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.322857142857143e-05,
+      "loss": 0.6759,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3257142857142857e-05,
+      "loss": 0.6923,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3285714285714287e-05,
+      "loss": 0.6806,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3314285714285715e-05,
+      "loss": 0.6759,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3342857142857144e-05,
+      "loss": 0.6698,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3371428571428572e-05,
+      "loss": 0.7038,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.6627,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3428571428571429e-05,
+      "loss": 0.7039,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3457142857142858e-05,
+      "loss": 2.1773,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3485714285714286e-05,
+      "loss": 0.6617,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3514285714285716e-05,
+      "loss": 0.6713,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3542857142857144e-05,
+      "loss": 0.7045,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.7275,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.7427,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3628571428571428e-05,
+      "loss": 0.7109,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3657142857142858e-05,
+      "loss": 0.7195,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3685714285714286e-05,
+      "loss": 0.7099,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 0.7197,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3742857142857144e-05,
+      "loss": 0.6804,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3771428571428574e-05,
+      "loss": 0.7139,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.38e-05,
+      "loss": 0.6723,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.382857142857143e-05,
+      "loss": 0.6969,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3857142857142858e-05,
+      "loss": 0.6871,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3885714285714288e-05,
+      "loss": 0.7016,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3914285714285716e-05,
+      "loss": 0.6391,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3942857142857145e-05,
+      "loss": 0.6896,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3971428571428573e-05,
+      "loss": 0.6958,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4e-05,
+      "loss": 0.6904,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.402857142857143e-05,
+      "loss": 0.7142,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4057142857142858e-05,
+      "loss": 0.6978,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4085714285714287e-05,
+      "loss": 0.6473,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4114285714285715e-05,
+      "loss": 0.6612,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4142857142857145e-05,
+      "loss": 0.6923,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4171428571428572e-05,
+      "loss": 0.6603,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 0.6485,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.422857142857143e-05,
+      "loss": 0.6979,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4257142857142859e-05,
+      "loss": 0.696,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.7209,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4314285714285717e-05,
+      "loss": 0.6646,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4342857142857145e-05,
+      "loss": 0.6654,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4371428571428571e-05,
+      "loss": 0.6952,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.6954,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4428571428571429e-05,
+      "loss": 0.6627,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4457142857142859e-05,
+      "loss": 0.6617,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4485714285714287e-05,
+      "loss": 0.6673,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4514285714285717e-05,
+      "loss": 0.673,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4542857142857143e-05,
+      "loss": 0.6896,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4571428571428573e-05,
+      "loss": 0.705,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.46e-05,
+      "loss": 0.6681,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.462857142857143e-05,
+      "loss": 0.6651,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4657142857142859e-05,
+      "loss": 0.648,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4685714285714288e-05,
+      "loss": 0.6648,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4714285714285716e-05,
+      "loss": 0.6493,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4742857142857143e-05,
+      "loss": 0.6744,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4771428571428573e-05,
+      "loss": 0.6859,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.48e-05,
+      "loss": 0.6474,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.482857142857143e-05,
+      "loss": 0.669,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 0.7058,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4885714285714288e-05,
+      "loss": 0.6697,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4914285714285715e-05,
+      "loss": 0.6271,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4942857142857143e-05,
+      "loss": 0.6923,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4971428571428572e-05,
+      "loss": 0.6838,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6217,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.502857142857143e-05,
+      "loss": 0.6955,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.505714285714286e-05,
+      "loss": 0.6741,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5085714285714288e-05,
+      "loss": 3.1619,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5114285714285714e-05,
+      "loss": 0.6924,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5142857142857144e-05,
+      "loss": 0.6941,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5171428571428572e-05,
+      "loss": 0.6673,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.6551,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.522857142857143e-05,
+      "loss": 0.667,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.525714285714286e-05,
+      "loss": 0.6718,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5285714285714286e-05,
+      "loss": 0.7008,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5314285714285716e-05,
+      "loss": 0.6709,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5342857142857142e-05,
+      "loss": 0.6639,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5371428571428572e-05,
+      "loss": 0.7184,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.54e-05,
+      "loss": 0.7268,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 0.6748,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.545714285714286e-05,
+      "loss": 0.6801,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5485714285714287e-05,
+      "loss": 0.6842,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5514285714285714e-05,
+      "loss": 0.6738,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5542857142857144e-05,
+      "loss": 0.721,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5571428571428573e-05,
+      "loss": 0.656,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.6783,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.562857142857143e-05,
+      "loss": 0.7229,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5657142857142856e-05,
+      "loss": 0.6682,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5685714285714286e-05,
+      "loss": 0.707,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.6762,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5742857142857145e-05,
+      "loss": 0.6385,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.577142857142857e-05,
+      "loss": 0.669,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.58e-05,
+      "loss": 0.686,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.582857142857143e-05,
+      "loss": 0.6845,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5857142857142857e-05,
+      "loss": 0.6563,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5885714285714287e-05,
+      "loss": 0.6477,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5914285714285717e-05,
+      "loss": 0.6944,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5942857142857143e-05,
+      "loss": 0.7022,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5971428571428573e-05,
+      "loss": 0.687,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6807,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.602857142857143e-05,
+      "loss": 0.687,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.605714285714286e-05,
+      "loss": 0.6737,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6085714285714285e-05,
+      "loss": 0.653,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6114285714285715e-05,
+      "loss": 0.6745,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6142857142857145e-05,
+      "loss": 0.7171,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6171428571428574e-05,
+      "loss": 0.676,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.62e-05,
+      "loss": 0.708,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.622857142857143e-05,
+      "loss": 0.6592,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6257142857142857e-05,
+      "loss": 0.6502,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6285714285714287e-05,
+      "loss": 0.6821,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6314285714285716e-05,
+      "loss": 0.6327,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6342857142857146e-05,
+      "loss": 0.6779,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6371428571428572e-05,
+      "loss": 0.6626,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.64e-05,
+      "loss": 0.703,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.6791,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.645714285714286e-05,
+      "loss": 0.6729,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6485714285714288e-05,
+      "loss": 2.8211,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6514285714285714e-05,
+      "loss": 0.6657,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6542857142857144e-05,
+      "loss": 0.664,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 0.6796,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.66e-05,
+      "loss": 0.6645,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.662857142857143e-05,
+      "loss": 0.6593,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.665714285714286e-05,
+      "loss": 0.6782,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6685714285714286e-05,
+      "loss": 0.6557,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6714285714285716e-05,
+      "loss": 0.6892,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6742857142857146e-05,
+      "loss": 0.7525,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6771428571428572e-05,
+      "loss": 4.3471,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.6602,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6828571428571428e-05,
+      "loss": 0.6763,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6857142857142858e-05,
+      "loss": 0.681,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6885714285714288e-05,
+      "loss": 0.7369,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6914285714285717e-05,
+      "loss": 0.72,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6942857142857144e-05,
+      "loss": 0.6801,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6971428571428574e-05,
+      "loss": 0.7167,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7e-05,
+      "loss": 0.7224,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.702857142857143e-05,
+      "loss": 0.6816,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.705714285714286e-05,
+      "loss": 0.6562,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.708571428571429e-05,
+      "loss": 0.7166,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7114285714285715e-05,
+      "loss": 0.7171,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.6703,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.717142857142857e-05,
+      "loss": 0.7146,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.72e-05,
+      "loss": 4.2215,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.722857142857143e-05,
+      "loss": 0.6711,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7257142857142857e-05,
+      "loss": 0.7227,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7285714285714287e-05,
+      "loss": 0.7076,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7314285714285717e-05,
+      "loss": 0.7196,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7342857142857143e-05,
+      "loss": 0.6919,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7371428571428573e-05,
+      "loss": 0.6327,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 0.6698,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.742857142857143e-05,
+      "loss": 0.6752,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.745714285714286e-05,
+      "loss": 0.6753,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.748571428571429e-05,
+      "loss": 0.6534,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7514285714285715e-05,
+      "loss": 0.6583,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7542857142857145e-05,
+      "loss": 0.6622,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.757142857142857e-05,
+      "loss": 0.7123,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.76e-05,
+      "loss": 0.6566,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.762857142857143e-05,
+      "loss": 0.7295,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.765714285714286e-05,
+      "loss": 0.6734,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7685714285714287e-05,
+      "loss": 0.6743,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 0.6909,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7742857142857143e-05,
+      "loss": 0.7063,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7771428571428573e-05,
+      "loss": 0.7033,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 0.7003,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7828571428571432e-05,
+      "loss": 0.6835,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.6775,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7885714285714285e-05,
+      "loss": 0.6966,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7914285714285715e-05,
+      "loss": 0.6526,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7942857142857144e-05,
+      "loss": 0.6937,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7971428571428574e-05,
+      "loss": 0.645,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6861,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.802857142857143e-05,
+      "loss": 0.6912,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8057142857142857e-05,
+      "loss": 0.7021,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8085714285714286e-05,
+      "loss": 0.7203,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8114285714285716e-05,
+      "loss": 0.6688,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8142857142857146e-05,
+      "loss": 0.6759,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8171428571428572e-05,
+      "loss": 0.674,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.6774,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.822857142857143e-05,
+      "loss": 0.6548,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8257142857142858e-05,
+      "loss": 0.6968,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 0.6729,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8314285714285714e-05,
+      "loss": 4.5579,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8342857142857144e-05,
+      "loss": 0.6527,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8371428571428574e-05,
+      "loss": 0.6587,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.6989,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842857142857143e-05,
+      "loss": 0.6835,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.845714285714286e-05,
+      "loss": 0.7152,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8485714285714286e-05,
+      "loss": 0.722,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8514285714285716e-05,
+      "loss": 0.6956,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8542857142857145e-05,
+      "loss": 0.6951,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.7268,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.86e-05,
+      "loss": 0.6965,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8628571428571428e-05,
+      "loss": 0.6878,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8657142857142858e-05,
+      "loss": 0.721,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8685714285714287e-05,
+      "loss": 0.6895,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8714285714285717e-05,
+      "loss": 0.67,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8742857142857143e-05,
+      "loss": 0.6594,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8771428571428573e-05,
+      "loss": 0.6706,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.88e-05,
+      "loss": 0.6808,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.882857142857143e-05,
+      "loss": 0.674,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 0.7026,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.888571428571429e-05,
+      "loss": 0.657,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8914285714285715e-05,
+      "loss": 0.6967,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8942857142857145e-05,
+      "loss": 0.6709,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8971428571428575e-05,
+      "loss": 0.6935,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6521,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.902857142857143e-05,
+      "loss": 0.6871,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9057142857142857e-05,
+      "loss": 0.7219,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9085714285714287e-05,
+      "loss": 0.6739,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9114285714285717e-05,
+      "loss": 0.7325,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9142857142857146e-05,
+      "loss": 0.6835,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9171428571428573e-05,
+      "loss": 0.699,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.6432,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.922857142857143e-05,
+      "loss": 0.6606,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.925714285714286e-05,
+      "loss": 0.6638,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.666,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9314285714285718e-05,
+      "loss": 0.6613,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9342857142857144e-05,
+      "loss": 0.6824,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.937142857142857e-05,
+      "loss": 0.7008,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.94e-05,
+      "loss": 0.672,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 0.6435,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.945714285714286e-05,
+      "loss": 0.6824,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9485714285714286e-05,
+      "loss": 0.6749,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9514285714285716e-05,
+      "loss": 0.6768,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9542857142857143e-05,
+      "loss": 0.6443,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9571428571428572e-05,
+      "loss": 3.2046,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.6565,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628571428571432e-05,
+      "loss": 0.6986,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657142857142858e-05,
+      "loss": 0.7224,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685714285714288e-05,
+      "loss": 0.6715,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9714285714285718e-05,
+      "loss": 0.659,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9742857142857144e-05,
+      "loss": 0.6684,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9771428571428574e-05,
+      "loss": 0.6799,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98e-05,
+      "loss": 0.6662,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982857142857143e-05,
+      "loss": 0.6556,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985714285714286e-05,
+      "loss": 0.6615,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.988571428571429e-05,
+      "loss": 0.6925,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9914285714285716e-05,
+      "loss": 0.6527,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9942857142857142e-05,
+      "loss": 0.6678,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9971428571428572e-05,
+      "loss": 0.6849,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.673,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999903468597e-05,
+      "loss": 2.6333,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999613874385e-05,
+      "loss": 0.6654,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999131217373e-05,
+      "loss": 0.695,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999998455497567e-05,
+      "loss": 0.6972,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997586714986e-05,
+      "loss": 0.6867,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999652486964e-05,
+      "loss": 0.6903,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999995269961555e-05,
+      "loss": 0.7034,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999382199075e-05,
+      "loss": 0.6719,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999992180957256e-05,
+      "loss": 0.6688,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999990346861104e-05,
+      "loss": 0.708,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999988319702332e-05,
+      "loss": 0.6866,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999986099480977e-05,
+      "loss": 0.749,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998368619708e-05,
+      "loss": 0.6763,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999981079850687e-05,
+      "loss": 0.649,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999978280441853e-05,
+      "loss": 0.683,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997528797063e-05,
+      "loss": 0.6948,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999972102437076e-05,
+      "loss": 0.6699,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996872384125e-05,
+      "loss": 0.6676,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996515218322e-05,
+      "loss": 0.6519,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999961387463053e-05,
+      "loss": 0.7083,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999957429680824e-05,
+      "loss": 0.6732,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995327883661e-05,
+      "loss": 0.6909,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999948934930487e-05,
+      "loss": 0.6725,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999944397962542e-05,
+      "loss": 0.6972,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999939667932863e-05,
+      "loss": 0.6944,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999934744841537e-05,
+      "loss": 0.7123,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999929628688664e-05,
+      "loss": 0.6293,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999924319474343e-05,
+      "loss": 0.6734,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999991881719867e-05,
+      "loss": 0.6643,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999991312186176e-05,
+      "loss": 0.6691,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999907233463714e-05,
+      "loss": 0.6459,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999901152004654e-05,
+      "loss": 0.6826,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999989487748469e-05,
+      "loss": 0.7341,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999888409903948e-05,
+      "loss": 0.6636,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999881749262556e-05,
+      "loss": 0.6878,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999874895560634e-05,
+      "loss": 0.6452,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999986784879832e-05,
+      "loss": 0.6772,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999986060897575e-05,
+      "loss": 0.6934,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999853176093064e-05,
+      "loss": 0.6419,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999845550150397e-05,
+      "loss": 0.6689,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999983773114791e-05,
+      "loss": 0.6691,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999829719085745e-05,
+      "loss": 0.6754,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999821513964056e-05,
+      "loss": 0.6943,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999813115783007e-05,
+      "loss": 0.7047,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999804524542756e-05,
+      "loss": 0.6778,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999979574024347e-05,
+      "loss": 0.7101,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999978676288532e-05,
+      "loss": 0.6869,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999977759246848e-05,
+      "loss": 0.6826,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999976822899312e-05,
+      "loss": 0.6508,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999758672459433e-05,
+      "loss": 0.6751,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999748922867592e-05,
+      "loss": 0.6599,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999973898021779e-05,
+      "loss": 0.6671,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999972884451022e-05,
+      "loss": 0.6687,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999718515745074e-05,
+      "loss": 0.6745,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999707993922555e-05,
+      "loss": 0.6765,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999697279042864e-05,
+      "loss": 0.6937,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999968637110621e-05,
+      "loss": 0.6655,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999675270112802e-05,
+      "loss": 0.6825,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999663976062855e-05,
+      "loss": 0.657,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999652488956585e-05,
+      "loss": 0.68,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999640808794218e-05,
+      "loss": 0.7028,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999628935575975e-05,
+      "loss": 0.6288,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999961686930209e-05,
+      "loss": 0.6342,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999960460997279e-05,
+      "loss": 0.6692,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999959215758832e-05,
+      "loss": 0.6866,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999579512148907e-05,
+      "loss": 0.6819,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999566673654808e-05,
+      "loss": 0.6742,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999553642106267e-05,
+      "loss": 0.6505,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999540417503532e-05,
+      "loss": 0.6403,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999526999846864e-05,
+      "loss": 0.6657,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999513389136517e-05,
+      "loss": 0.6552,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999499585372758e-05,
+      "loss": 0.6291,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999485588555852e-05,
+      "loss": 0.6413,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999471398686067e-05,
+      "loss": 0.669,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999945701576368e-05,
+      "loss": 0.6686,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999442439788964e-05,
+      "loss": 0.6779,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999427670762207e-05,
+      "loss": 0.6281,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999941270868369e-05,
+      "loss": 0.6425,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999397553553702e-05,
+      "loss": 0.6339,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999382205372538e-05,
+      "loss": 0.6997,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999366664140493e-05,
+      "loss": 0.6534,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999350929857865e-05,
+      "loss": 0.6669,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999933500252496e-05,
+      "loss": 0.6495,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999318882142083e-05,
+      "loss": 1.91,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999302568709548e-05,
+      "loss": 0.6302,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999928606222767e-05,
+      "loss": 0.6377,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999926936269677e-05,
+      "loss": 0.6725,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999252470117162e-05,
+      "loss": 0.6812,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999923538448918e-05,
+      "loss": 0.694,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999921810581315e-05,
+      "loss": 0.6565,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999920063408941e-05,
+      "loss": 0.6775,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999918296931829e-05,
+      "loss": 0.6974,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999165111500138e-05,
+      "loss": 0.6878,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999147060635292e-05,
+      "loss": 0.666,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999912881672411e-05,
+      "loss": 0.6821,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999110379766934e-05,
+      "loss": 0.6853,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999091749764127e-05,
+      "loss": 0.6076,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999072926716047e-05,
+      "loss": 0.6535,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999053910623056e-05,
+      "loss": 0.6397,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999903470148552e-05,
+      "loss": 0.7106,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999901529930381e-05,
+      "loss": 0.6761,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998995704078305e-05,
+      "loss": 0.6687,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999897591580938e-05,
+      "loss": 0.6562,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999895593449742e-05,
+      "loss": 0.6669,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998935760142805e-05,
+      "loss": 0.6905,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999891539274593e-05,
+      "loss": 0.7032,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998894832307184e-05,
+      "loss": 0.6912,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998874078826968e-05,
+      "loss": 0.6791,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998853132305683e-05,
+      "loss": 0.6951,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998831992743726e-05,
+      "loss": 0.6514,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998810660141514e-05,
+      "loss": 0.6375,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998789134499453e-05,
+      "loss": 0.6802,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998767415817963e-05,
+      "loss": 0.6866,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998745504097457e-05,
+      "loss": 0.6392,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998723399338364e-05,
+      "loss": 0.6373,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999870110154111e-05,
+      "loss": 0.6748,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999867861070612e-05,
+      "loss": 0.6933,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998655926833838e-05,
+      "loss": 0.6395,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998633049924693e-05,
+      "loss": 0.6829,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999860997997913e-05,
+      "loss": 0.6232,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998586716997595e-05,
+      "loss": 0.6311,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998563260980535e-05,
+      "loss": 0.6884,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998539611928407e-05,
+      "loss": 0.6969,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998515769841665e-05,
+      "loss": 0.6725,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998491734720767e-05,
+      "loss": 0.6756,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998467506566182e-05,
+      "loss": 0.673,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998443085378372e-05,
+      "loss": 0.6608,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998418471157815e-05,
+      "loss": 0.6908,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998393663904982e-05,
+      "loss": 0.6936,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999836866362035e-05,
+      "loss": 0.6699,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998343470304408e-05,
+      "loss": 0.678,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999831808395764e-05,
+      "loss": 0.6474,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998292504580528e-05,
+      "loss": 0.6624,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998266732173577e-05,
+      "loss": 0.6874,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999824076673728e-05,
+      "loss": 0.6791,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998214608272136e-05,
+      "loss": 0.6766,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998188256778656e-05,
+      "loss": 0.691,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999816171225735e-05,
+      "loss": 0.6763,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998134974708718e-05,
+      "loss": 0.641,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998108044133286e-05,
+      "loss": 0.6682,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999808092053157e-05,
+      "loss": 0.6559,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99980536039041e-05,
+      "loss": 0.6716,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998026094251397e-05,
+      "loss": 0.6658,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997998391573992e-05,
+      "loss": 0.6787,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997970495872425e-05,
+      "loss": 0.6806,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997942407147232e-05,
+      "loss": 0.6818,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999791412539895e-05,
+      "loss": 0.6932,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997885650628134e-05,
+      "loss": 0.6915,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999785698283533e-05,
+      "loss": 0.6352,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999782812202109e-05,
+      "loss": 0.6598,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999779906818597e-05,
+      "loss": 0.6515,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997769821330534e-05,
+      "loss": 0.69,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997740381455348e-05,
+      "loss": 0.6415,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999771074856098e-05,
+      "loss": 0.633,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997680922647994e-05,
+      "loss": 0.6409,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997650903716975e-05,
+      "loss": 0.6726,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997620691768502e-05,
+      "loss": 0.6311,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997590286803153e-05,
+      "loss": 0.6701,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999755968882152e-05,
+      "loss": 0.6403,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999752889782419e-05,
+      "loss": 0.6302,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999749791381176e-05,
+      "loss": 0.6817,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997466736784826e-05,
+      "loss": 0.6978,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999743536674399e-05,
+      "loss": 0.6469,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997403803689862e-05,
+      "loss": 0.6604,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997372047623048e-05,
+      "loss": 0.6516,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999734009854416e-05,
+      "loss": 0.6513,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997307956453817e-05,
+      "loss": 0.6751,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997275621352637e-05,
+      "loss": 0.651,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997243093241248e-05,
+      "loss": 0.6827,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997210372120276e-05,
+      "loss": 0.6602,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999717745799035e-05,
+      "loss": 0.6755,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999714435085211e-05,
+      "loss": 0.6765,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997111050706196e-05,
+      "loss": 0.6271,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997077557553247e-05,
+      "loss": 0.6616,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997043871393906e-05,
+      "loss": 0.6725,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999700999222883e-05,
+      "loss": 0.6646,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996975920058673e-05,
+      "loss": 0.6814,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996941654884093e-05,
+      "loss": 0.6376,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996907196705747e-05,
+      "loss": 0.6602,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996872545524304e-05,
+      "loss": 0.6689,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999683770134043e-05,
+      "loss": 0.638,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99968026641548e-05,
+      "loss": 0.6488,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996767433968092e-05,
+      "loss": 0.6933,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996732010780983e-05,
+      "loss": 0.615,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996696394594158e-05,
+      "loss": 0.6506,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996660585408304e-05,
+      "loss": 0.7361,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996624583224112e-05,
+      "loss": 0.6468,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999658838804228e-05,
+      "loss": 0.669,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996551999863505e-05,
+      "loss": 0.6867,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996515418688493e-05,
+      "loss": 0.6315,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996478644517938e-05,
+      "loss": 0.6416,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996441677352566e-05,
+      "loss": 0.6916,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996404517193078e-05,
+      "loss": 0.6816,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99963671640402e-05,
+      "loss": 0.659,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996329617894648e-05,
+      "loss": 0.6957,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996291878757152e-05,
+      "loss": 0.6573,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996253946628435e-05,
+      "loss": 0.6436,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999621582150923e-05,
+      "loss": 0.6796,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996177503400277e-05,
+      "loss": 0.6809,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996138992302314e-05,
+      "loss": 0.6599,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996100288216082e-05,
+      "loss": 0.6802,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996061391142332e-05,
+      "loss": 0.6159,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996022301081815e-05,
+      "loss": 0.6765,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999598301803528e-05,
+      "loss": 0.6753,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999594354200349e-05,
+      "loss": 0.6448,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995903872987207e-05,
+      "loss": 0.6766,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995864010987197e-05,
+      "loss": 0.6573,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999582395600423e-05,
+      "loss": 0.6441,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995783708039075e-05,
+      "loss": 0.6588,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995743267092514e-05,
+      "loss": 0.6452,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995702633165325e-05,
+      "loss": 0.6656,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995661806258298e-05,
+      "loss": 0.6722,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999562078637221e-05,
+      "loss": 0.6804,
+      "step": 913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995579573507863e-05,
+      "loss": 0.6665,
+      "step": 914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995538167666048e-05,
+      "loss": 0.6625,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999549656884757e-05,
+      "loss": 0.6812,
+      "step": 916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995454777053226e-05,
+      "loss": 0.6549,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995412792283822e-05,
+      "loss": 0.6909,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995370614540174e-05,
+      "loss": 0.6634,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999532824382309e-05,
+      "loss": 0.6561,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995285680133393e-05,
+      "loss": 0.6666,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995242923471904e-05,
+      "loss": 0.6721,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995199973839446e-05,
+      "loss": 0.6592,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995156831236854e-05,
+      "loss": 0.6236,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995113495664954e-05,
+      "loss": 0.6613,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995069967124587e-05,
+      "loss": 0.6731,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995026245616592e-05,
+      "loss": 0.6347,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994982331141813e-05,
+      "loss": 0.7042,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994938223701098e-05,
+      "loss": 0.7096,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994893923295296e-05,
+      "loss": 0.6712,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999484942992527e-05,
+      "loss": 0.6346,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994804743591865e-05,
+      "loss": 0.6838,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994759864295958e-05,
+      "loss": 0.6282,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999471479203841e-05,
+      "loss": 0.6802,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999466952682009e-05,
+      "loss": 0.692,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994624068641872e-05,
+      "loss": 0.6855,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994578417504637e-05,
+      "loss": 0.6137,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999453257340926e-05,
+      "loss": 0.6672,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994486536356636e-05,
+      "loss": 0.6658,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999444030634764e-05,
+      "loss": 0.6536,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999439388338318e-05,
+      "loss": 0.6427,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994347267464142e-05,
+      "loss": 0.6533,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994300458591426e-05,
+      "loss": 0.6767,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994253456765943e-05,
+      "loss": 0.6574,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999420626198859e-05,
+      "loss": 0.6156,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994158874260287e-05,
+      "loss": 0.6354,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994111293581947e-05,
+      "loss": 0.6204,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994063519954486e-05,
+      "loss": 0.6814,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994015553378828e-05,
+      "loss": 0.6341,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99939673938559e-05,
+      "loss": 0.6568,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993919041386623e-05,
+      "loss": 0.6725,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993870495971946e-05,
+      "loss": 0.619,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993821757612796e-05,
+      "loss": 0.6746,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993772826310115e-05,
+      "loss": 0.6191,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993723702064852e-05,
+      "loss": 0.6647,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999367438487795e-05,
+      "loss": 0.6797,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993624874750366e-05,
+      "loss": 0.6534,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999357517168305e-05,
+      "loss": 0.6567,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993525275676965e-05,
+      "loss": 0.6613,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993475186733078e-05,
+      "loss": 0.6582,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993424904852348e-05,
+      "loss": 0.6831,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999337443003575e-05,
+      "loss": 0.6811,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993323762284262e-05,
+      "loss": 0.6881,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993272901598858e-05,
+      "loss": 0.6628,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993221847980517e-05,
+      "loss": 0.6624,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993170601430233e-05,
+      "loss": 0.6548,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993119161948985e-05,
+      "loss": 0.6368,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993067529537775e-05,
+      "loss": 0.6536,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993015704197596e-05,
+      "loss": 0.6706,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999296368592945e-05,
+      "loss": 0.6251,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999291147473434e-05,
+      "loss": 0.664,
+      "step": 971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992859070613275e-05,
+      "loss": 0.6973,
+      "step": 972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992806473567263e-05,
+      "loss": 0.6377,
+      "step": 973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992753683597324e-05,
+      "loss": 0.6526,
+      "step": 974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992700700704478e-05,
+      "loss": 0.654,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992647524889747e-05,
+      "loss": 0.6537,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992594156154152e-05,
+      "loss": 0.6703,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999254059449873e-05,
+      "loss": 0.6245,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999248683992451e-05,
+      "loss": 0.6832,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992432892432534e-05,
+      "loss": 0.6611,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992378752023845e-05,
+      "loss": 0.7022,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999232441869948e-05,
+      "loss": 0.683,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992269892460497e-05,
+      "loss": 0.6632,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992215173307944e-05,
+      "loss": 0.6457,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999216026124288e-05,
+      "loss": 0.6747,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999210515626636e-05,
+      "loss": 0.6614,
+      "step": 986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992049858379453e-05,
+      "loss": 1.2856,
+      "step": 987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991994367583223e-05,
+      "loss": 0.6552,
+      "step": 988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991938683878746e-05,
+      "loss": 0.6987,
+      "step": 989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991882807267097e-05,
+      "loss": 0.6518,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991826737749348e-05,
+      "loss": 0.6813,
+      "step": 991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991770475326587e-05,
+      "loss": 0.6588,
+      "step": 992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991714019999897e-05,
+      "loss": 0.651,
+      "step": 993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991657371770373e-05,
+      "loss": 0.6871,
+      "step": 994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99916005306391e-05,
+      "loss": 0.6607,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991543496607185e-05,
+      "loss": 0.6379,
+      "step": 996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991486269675726e-05,
+      "loss": 0.7035,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991428849845828e-05,
+      "loss": 0.6281,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991371237118593e-05,
+      "loss": 0.6458,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999131343149514e-05,
+      "loss": 0.6632,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991255432976588e-05,
+      "loss": 0.6516,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999119724156405e-05,
+      "loss": 0.6819,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999113885725865e-05,
+      "loss": 0.6614,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991080280061518e-05,
+      "loss": 0.6729,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991021509973783e-05,
+      "loss": 0.6503,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990962546996583e-05,
+      "loss": 0.6711,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999090339113105e-05,
+      "loss": 0.6355,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990844042378334e-05,
+      "loss": 0.6524,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990784500739578e-05,
+      "loss": 0.6707,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990724766215925e-05,
+      "loss": 0.6684,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990664838808538e-05,
+      "loss": 0.6409,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999060471851857e-05,
+      "loss": 0.6562,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990544405347178e-05,
+      "loss": 0.6386,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999048389929553e-05,
+      "loss": 0.6771,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990423200364794e-05,
+      "loss": 0.6519,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990362308556146e-05,
+      "loss": 0.6873,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999030122387075e-05,
+      "loss": 0.6516,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99902399463098e-05,
+      "loss": 0.7007,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999017847587447e-05,
+      "loss": 0.6804,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990116812565944e-05,
+      "loss": 0.6946,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990054956385417e-05,
+      "loss": 0.6841,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989992907334088e-05,
+      "loss": 0.6245,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989930665413148e-05,
+      "loss": 0.6342,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99898682306238e-05,
+      "loss": 0.6566,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998980560296725e-05,
+      "loss": 0.642,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989742782444707e-05,
+      "loss": 0.6578,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989679769057386e-05,
+      "loss": 0.6665,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99896165628065e-05,
+      "loss": 0.6313,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989553163693275e-05,
+      "loss": 0.6288,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989489571718924e-05,
+      "loss": 0.6793,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989425786884688e-05,
+      "loss": 0.6455,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989361809191786e-05,
+      "loss": 0.6645,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989297638641463e-05,
+      "loss": 0.6236,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989233275234952e-05,
+      "loss": 0.6707,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99891687189735e-05,
+      "loss": 0.6632,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998910396985835e-05,
+      "loss": 0.6228,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989039027890757e-05,
+      "loss": 0.6456,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988973893071964e-05,
+      "loss": 0.6239,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998890856540324e-05,
+      "loss": 0.6851,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998884304488584e-05,
+      "loss": 0.7163,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998877733152103e-05,
+      "loss": 0.6567,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998871142531008e-05,
+      "loss": 0.6424,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988645326254262e-05,
+      "loss": 0.6693,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988579034354853e-05,
+      "loss": 0.6693,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998851254961313e-05,
+      "loss": 0.6674,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998844587203038e-05,
+      "loss": 0.6488,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988379001607888e-05,
+      "loss": 0.6464,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988311938346944e-05,
+      "loss": 0.6755,
+      "step": 1048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988244682248847e-05,
+      "loss": 0.7015,
+      "step": 1049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998817723331489e-05,
+      "loss": 0.6533,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988109591546378e-05,
+      "loss": 0.6453,
+      "step": 1051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988041756944618e-05,
+      "loss": 0.6885,
+      "step": 1052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998797372951092e-05,
+      "loss": 0.6221,
+      "step": 1053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987905509246595e-05,
+      "loss": 0.621,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998783709615296e-05,
+      "loss": 0.6562,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987768490231338e-05,
+      "loss": 0.6491,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998769969148305e-05,
+      "loss": 0.6202,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987630699909427e-05,
+      "loss": 0.6738,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987561515511802e-05,
+      "loss": 0.6864,
+      "step": 1059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987492138291508e-05,
+      "loss": 0.6536,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987422568249886e-05,
+      "loss": 0.6402,
+      "step": 1061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998735280538828e-05,
+      "loss": 0.6585,
+      "step": 1062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987282849708035e-05,
+      "loss": 0.6515,
+      "step": 1063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987212701210503e-05,
+      "loss": 0.6598,
+      "step": 1064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987142359897037e-05,
+      "loss": 0.6437,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987071825768993e-05,
+      "loss": 0.6784,
+      "step": 1066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987001098827735e-05,
+      "loss": 0.6055,
+      "step": 1067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986930179074636e-05,
+      "loss": 0.6956,
+      "step": 1068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998685906651105e-05,
+      "loss": 0.6626,
+      "step": 1069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986787761138363e-05,
+      "loss": 0.6451,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986716262957943e-05,
+      "loss": 0.6634,
+      "step": 1071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986644571971174e-05,
+      "loss": 0.6676,
+      "step": 1072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986572688179442e-05,
+      "loss": 0.6828,
+      "step": 1073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986500611584133e-05,
+      "loss": 3.0516,
+      "step": 1074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986428342186636e-05,
+      "loss": 0.6298,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986355879988352e-05,
+      "loss": 0.6699,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986283224990672e-05,
+      "loss": 0.6717,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998621037719501e-05,
+      "loss": 0.6526,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998613733660276e-05,
+      "loss": 0.6686,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998606410321534e-05,
+      "loss": 0.6555,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998599067703416e-05,
+      "loss": 0.663,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985917058060637e-05,
+      "loss": 0.6671,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99858432462962e-05,
+      "loss": 0.6396,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985769241742264e-05,
+      "loss": 0.6583,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985695044400265e-05,
+      "loss": 0.6389,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985620654271632e-05,
+      "loss": 0.6595,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985546071357806e-05,
+      "loss": 0.6437,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998547129566022e-05,
+      "loss": 0.6494,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998539632718032e-05,
+      "loss": 0.686,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985321165919552e-05,
+      "loss": 0.6135,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985245811879372e-05,
+      "loss": 0.6203,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998517026506123e-05,
+      "loss": 0.6648,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985094525466585e-05,
+      "loss": 0.6999,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985018593096903e-05,
+      "loss": 0.6519,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998494246795365e-05,
+      "loss": 0.64,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998486615003829e-05,
+      "loss": 0.67,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99847896393523e-05,
+      "loss": 0.679,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984712935897157e-05,
+      "loss": 0.7096,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984636039674342e-05,
+      "loss": 0.6621,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998455895068534e-05,
+      "loss": 0.6896,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984481668931635e-05,
+      "loss": 0.6465,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984404194414724e-05,
+      "loss": 0.619,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984326527136103e-05,
+      "loss": 0.6569,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984248667097273e-05,
+      "loss": 0.6832,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984170614299727e-05,
+      "loss": 0.6338,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984092368744984e-05,
+      "loss": 0.6842,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984013930434548e-05,
+      "loss": 0.6675,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983935299369934e-05,
+      "loss": 0.6962,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983856475552662e-05,
+      "loss": 0.695,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998377745898425e-05,
+      "loss": 0.6682,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983698249666232e-05,
+      "loss": 0.6206,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983618847600126e-05,
+      "loss": 0.655,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983539252787475e-05,
+      "loss": 0.6673,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983459465229808e-05,
+      "loss": 0.629,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983379484928668e-05,
+      "loss": 0.6614,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99832993118856e-05,
+      "loss": 0.6776,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983218946102152e-05,
+      "loss": 0.6329,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983138387579877e-05,
+      "loss": 0.6293,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983057636320326e-05,
+      "loss": 0.6483,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998297669232506e-05,
+      "loss": 0.6439,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982895555595644e-05,
+      "loss": 0.6481,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982814226133646e-05,
+      "loss": 0.6685,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982732703940627e-05,
+      "loss": 0.6495,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982650989018167e-05,
+      "loss": 0.6666,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982569081367844e-05,
+      "loss": 0.6344,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982486980991237e-05,
+      "loss": 0.6456,
+      "step": 1126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982404687889935e-05,
+      "loss": 0.6734,
+      "step": 1127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982322202065524e-05,
+      "loss": 0.6316,
+      "step": 1128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99822395235196e-05,
+      "loss": 0.6417,
+      "step": 1129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998215665225375e-05,
+      "loss": 0.6841,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982073588269585e-05,
+      "loss": 0.6585,
+      "step": 1131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981990331568698e-05,
+      "loss": 0.7493,
+      "step": 1132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981906882152708e-05,
+      "loss": 0.6731,
+      "step": 1133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981823240023217e-05,
+      "loss": 0.6485,
+      "step": 1134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981739405181845e-05,
+      "loss": 0.6311,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981655377630208e-05,
+      "loss": 0.6839,
+      "step": 1136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981571157369925e-05,
+      "loss": 0.6378,
+      "step": 1137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998148674440263e-05,
+      "loss": 0.655,
+      "step": 1138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981402138729946e-05,
+      "loss": 0.7004,
+      "step": 1139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998131734035351e-05,
+      "loss": 0.644,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981232349274956e-05,
+      "loss": 0.6785,
+      "step": 1141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998114716549593e-05,
+      "loss": 0.6114,
+      "step": 1142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981061789018072e-05,
+      "loss": 0.638,
+      "step": 1143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980976219843032e-05,
+      "loss": 0.6599,
+      "step": 1144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980890457972462e-05,
+      "loss": 0.6548,
+      "step": 1145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998080450340802e-05,
+      "loss": 0.6978,
+      "step": 1146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998071835615136e-05,
+      "loss": 0.6783,
+      "step": 1147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980632016204147e-05,
+      "loss": 0.6858,
+      "step": 1148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998054548356805e-05,
+      "loss": 0.6362,
+      "step": 1149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998045875824474e-05,
+      "loss": 0.672,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980371840235887e-05,
+      "loss": 0.6735,
+      "step": 1151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980284729543174e-05,
+      "loss": 0.6638,
+      "step": 1152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980197426168284e-05,
+      "loss": 0.6503,
+      "step": 1153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980109930112897e-05,
+      "loss": 0.6275,
+      "step": 1154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980022241378703e-05,
+      "loss": 0.6443,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979934359967398e-05,
+      "loss": 0.6812,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979846285880678e-05,
+      "loss": 0.6328,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997975801912024e-05,
+      "loss": 0.6429,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997966955968779e-05,
+      "loss": 0.6428,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979580907585042e-05,
+      "loss": 0.7221,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979492062813696e-05,
+      "loss": 0.6678,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979403025375476e-05,
+      "loss": 0.6276,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979313795272098e-05,
+      "loss": 0.6235,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997922437250529e-05,
+      "loss": 0.6634,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979134757076767e-05,
+      "loss": 0.6704,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979044948988267e-05,
+      "loss": 0.6496,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978954948241523e-05,
+      "loss": 0.6468,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997886475483827e-05,
+      "loss": 0.6248,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978774368780254e-05,
+      "loss": 0.6763,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978683790069215e-05,
+      "loss": 0.6431,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997859301870691e-05,
+      "loss": 0.6227,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997850205469508e-05,
+      "loss": 0.6641,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978410898035487e-05,
+      "loss": 0.599,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978319548729888e-05,
+      "loss": 0.6356,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978228006780056e-05,
+      "loss": 0.6489,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978136272187745e-05,
+      "loss": 0.6624,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978044344954738e-05,
+      "loss": 0.6059,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977952225082803e-05,
+      "loss": 0.632,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997785991257372e-05,
+      "loss": 0.6359,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977767407429268e-05,
+      "loss": 0.6403,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997767470965124e-05,
+      "loss": 0.6713,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977581819241418e-05,
+      "loss": 0.6552,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99774887362016e-05,
+      "loss": 0.642,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977395460533585e-05,
+      "loss": 0.6779,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977301992239167e-05,
+      "loss": 0.6664,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977208331320157e-05,
+      "loss": 0.6706,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997711447777836e-05,
+      "loss": 0.628,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997702043161559e-05,
+      "loss": 0.6432,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997692619283366e-05,
+      "loss": 3.4253,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997683176143439e-05,
+      "loss": 0.64,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976737137419602e-05,
+      "loss": 0.6447,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997664232079113e-05,
+      "loss": 0.6692,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976547311550796e-05,
+      "loss": 0.6252,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976452109700436e-05,
+      "loss": 0.6821,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976356715241892e-05,
+      "loss": 0.716,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976261128177e-05,
+      "loss": 0.7298,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976165348507614e-05,
+      "loss": 0.7156,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976069376235574e-05,
+      "loss": 0.6463,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997597321136274e-05,
+      "loss": 0.6997,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997587685389096e-05,
+      "loss": 0.688,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975780303822107e-05,
+      "loss": 0.6467,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997568356115803e-05,
+      "loss": 0.6638,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975586625900608e-05,
+      "loss": 0.6811,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997548949805171e-05,
+      "loss": 0.6527,
+      "step": 1204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975392177613207e-05,
+      "loss": 0.7021,
+      "step": 1205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975294664586982e-05,
+      "loss": 0.6214,
+      "step": 1206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975196958974918e-05,
+      "loss": 0.6748,
+      "step": 1207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99750990607789e-05,
+      "loss": 0.639,
+      "step": 1208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975000970000814e-05,
+      "loss": 0.6537,
+      "step": 1209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997490268664256e-05,
+      "loss": 0.6051,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974804210706034e-05,
+      "loss": 0.6463,
+      "step": 1211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974705542193132e-05,
+      "loss": 0.6475,
+      "step": 1212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997460668110577e-05,
+      "loss": 0.6059,
+      "step": 1213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974507627445845e-05,
+      "loss": 0.6727,
+      "step": 1214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974408381215277e-05,
+      "loss": 0.6654,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974308942415977e-05,
+      "loss": 0.6328,
+      "step": 1216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997420931104987e-05,
+      "loss": 0.6792,
+      "step": 1217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974109487118874e-05,
+      "loss": 0.6566,
+      "step": 1218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997400947062492e-05,
+      "loss": 0.6807,
+      "step": 1219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973909261569934e-05,
+      "loss": 0.6418,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973808859955858e-05,
+      "loss": 0.6046,
+      "step": 1221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973708265784627e-05,
+      "loss": 0.6644,
+      "step": 1222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973607479058183e-05,
+      "loss": 0.6613,
+      "step": 1223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973506499778473e-05,
+      "loss": 0.657,
+      "step": 1224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973405327947445e-05,
+      "loss": 0.6319,
+      "step": 1225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973303963567048e-05,
+      "loss": 0.6771,
+      "step": 1226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973202406639247e-05,
+      "loss": 0.669,
+      "step": 1227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973100657166e-05,
+      "loss": 0.6437,
+      "step": 1228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997299871514927e-05,
+      "loss": 0.6611,
+      "step": 1229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972896580591025e-05,
+      "loss": 0.6635,
+      "step": 1230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972794253493237e-05,
+      "loss": 0.6678,
+      "step": 1231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997269173385788e-05,
+      "loss": 0.7025,
+      "step": 1232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972589021686943e-05,
+      "loss": 0.6535,
+      "step": 1233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972486116982393e-05,
+      "loss": 0.6688,
+      "step": 1234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997238301974623e-05,
+      "loss": 0.6831,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997227972998044e-05,
+      "loss": 0.6966,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972176247687016e-05,
+      "loss": 0.6342,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997207257286795e-05,
+      "loss": 0.6649,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997196870552526e-05,
+      "loss": 0.6253,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971864645660933e-05,
+      "loss": 0.6536,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971760393276992e-05,
+      "loss": 0.6237,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971655948375443e-05,
+      "loss": 0.6677,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99715513109583e-05,
+      "loss": 0.6294,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997144648102759e-05,
+      "loss": 0.6645,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971341458585334e-05,
+      "loss": 0.6721,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997123624363356e-05,
+      "loss": 0.6963,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971130836174295e-05,
+      "loss": 0.6392,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971025236209582e-05,
+      "loss": 0.6172,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997091944374145e-05,
+      "loss": 0.6403,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997081345877195e-05,
+      "loss": 0.6443,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970707281303126e-05,
+      "loss": 0.6818,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970600911337024e-05,
+      "loss": 0.6496,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99704943488757e-05,
+      "loss": 0.6724,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970387593921212e-05,
+      "loss": 0.6728,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970280646475623e-05,
+      "loss": 0.6947,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970173506540993e-05,
+      "loss": 0.6693,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970066174119397e-05,
+      "loss": 0.6669,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99699586492129e-05,
+      "loss": 0.6523,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996985093182358e-05,
+      "loss": 0.6552,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996974302195352e-05,
+      "loss": 0.6677,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99696349196048e-05,
+      "loss": 0.6531,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996952662477951e-05,
+      "loss": 2.5447,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969418137479737e-05,
+      "loss": 1.5239,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996930945770758e-05,
+      "loss": 0.6091,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996920058546513e-05,
+      "loss": 0.6365,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969091520754494e-05,
+      "loss": 0.6205,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996898226357778e-05,
+      "loss": 0.6911,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968872813937094e-05,
+      "loss": 0.6431,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968763171834548e-05,
+      "loss": 0.6697,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968653337272262e-05,
+      "loss": 0.6481,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968543310252354e-05,
+      "loss": 0.6459,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968433090776948e-05,
+      "loss": 0.6292,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968322678848176e-05,
+      "loss": 0.6846,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968212074468164e-05,
+      "loss": 0.634,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996810127763905e-05,
+      "loss": 0.6668,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967990288362974e-05,
+      "loss": 0.6402,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967879106642076e-05,
+      "loss": 0.6282,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967767732478506e-05,
+      "loss": 0.6911,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967656165874413e-05,
+      "loss": 0.6284,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996754440683195e-05,
+      "loss": 0.6665,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967432455353273e-05,
+      "loss": 0.6219,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996732031144055e-05,
+      "loss": 0.6651,
+      "step": 1282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967207975095936e-05,
+      "loss": 0.6398,
+      "step": 1283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996709544632161e-05,
+      "loss": 0.6767,
+      "step": 1284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966982725119737e-05,
+      "loss": 0.6482,
+      "step": 1285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966869811492494e-05,
+      "loss": 0.6441,
+      "step": 1286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996675670544207e-05,
+      "loss": 0.6275,
+      "step": 1287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966643406970638e-05,
+      "loss": 0.6789,
+      "step": 1288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996652991608039e-05,
+      "loss": 0.6625,
+      "step": 1289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966416232773513e-05,
+      "loss": 0.6273,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966302357052208e-05,
+      "loss": 0.6366,
+      "step": 1291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966188288918672e-05,
+      "loss": 0.643,
+      "step": 1292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966074028375103e-05,
+      "loss": 0.6742,
+      "step": 1293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965959575423707e-05,
+      "loss": 0.6653,
+      "step": 1294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99658449300667e-05,
+      "loss": 0.6526,
+      "step": 1295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996573009230629e-05,
+      "loss": 0.6548,
+      "step": 1296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99656150621447e-05,
+      "loss": 0.6282,
+      "step": 1297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996549983958414e-05,
+      "loss": 0.6601,
+      "step": 1298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965384424626846e-05,
+      "loss": 0.6463,
+      "step": 1299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996526881727504e-05,
+      "loss": 0.6759,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965153017530958e-05,
+      "loss": 0.6849,
+      "step": 1301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965037025396827e-05,
+      "loss": 0.6263,
+      "step": 1302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964920840874896e-05,
+      "loss": 0.6399,
+      "step": 1303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964804463967403e-05,
+      "loss": 1.2964,
+      "step": 1304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99646878946766e-05,
+      "loss": 0.6856,
+      "step": 1305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996457113300473e-05,
+      "loss": 0.6496,
+      "step": 1306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964454178954053e-05,
+      "loss": 0.6106,
+      "step": 1307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996433703252682e-05,
+      "loss": 0.6742,
+      "step": 1308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964219693725303e-05,
+      "loss": 0.6459,
+      "step": 1309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996410216255176e-05,
+      "loss": 0.6509,
+      "step": 1310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963984439008462e-05,
+      "loss": 0.6138,
+      "step": 1311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963866523097683e-05,
+      "loss": 0.6437,
+      "step": 1312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99637484148217e-05,
+      "loss": 0.6815,
+      "step": 1313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963630114182788e-05,
+      "loss": 0.6856,
+      "step": 1314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963511621183236e-05,
+      "loss": 0.6474,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963392935825328e-05,
+      "loss": 0.6321,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963274058111364e-05,
+      "loss": 0.6828,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963154988043628e-05,
+      "loss": 0.6425,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963035725624423e-05,
+      "loss": 0.6607,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962916270856053e-05,
+      "loss": 0.6575,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962796623740822e-05,
+      "loss": 0.6504,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962676784281042e-05,
+      "loss": 0.6278,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996255675247903e-05,
+      "loss": 0.6615,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996243652833709e-05,
+      "loss": 0.7047,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996231611185756e-05,
+      "loss": 0.6227,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962195503042754e-05,
+      "loss": 0.6572,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962074701895006e-05,
+      "loss": 0.678,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961953708416643e-05,
+      "loss": 0.6709,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961832522610004e-05,
+      "loss": 0.8386,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996171114447743e-05,
+      "loss": 0.6824,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996158957402126e-05,
+      "loss": 0.6539,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961467811243845e-05,
+      "loss": 0.6182,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961345856147538e-05,
+      "loss": 0.6585,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961223708734687e-05,
+      "loss": 0.6133,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961101369007654e-05,
+      "loss": 0.6175,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960978836968797e-05,
+      "loss": 0.6391,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996085611262049e-05,
+      "loss": 0.6318,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960733195965095e-05,
+      "loss": 0.6275,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960610087004985e-05,
+      "loss": 0.62,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960486785742543e-05,
+      "loss": 0.6398,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960363292180144e-05,
+      "loss": 0.684,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960239606320172e-05,
+      "loss": 0.6441,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960115728165016e-05,
+      "loss": 0.6399,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959991657717073e-05,
+      "loss": 0.6588,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959867394978732e-05,
+      "loss": 0.619,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959742939952393e-05,
+      "loss": 0.6545,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959618292640458e-05,
+      "loss": 0.6701,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959493453045335e-05,
+      "loss": 0.657,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959368421169437e-05,
+      "loss": 0.6475,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959243197015172e-05,
+      "loss": 0.6563,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995911778058496e-05,
+      "loss": 0.6541,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958992171881228e-05,
+      "loss": 0.6408,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995886637090639e-05,
+      "loss": 0.6559,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958740377662886e-05,
+      "loss": 0.6572,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958614192153138e-05,
+      "loss": 0.6143,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958487814379593e-05,
+      "loss": 0.6611,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995836124434468e-05,
+      "loss": 0.6471,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958234482050853e-05,
+      "loss": 0.6845,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995810752750055e-05,
+      "loss": 0.6628,
+      "step": 1359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957980380696227e-05,
+      "loss": 0.6915,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995785304164034e-05,
+      "loss": 0.6697,
+      "step": 1361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995772551033534e-05,
+      "loss": 0.6247,
+      "step": 1362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99575977867837e-05,
+      "loss": 0.6714,
+      "step": 1363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957469870987878e-05,
+      "loss": 0.6538,
+      "step": 1364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957341762950346e-05,
+      "loss": 0.6804,
+      "step": 1365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995721346267358e-05,
+      "loss": 0.6425,
+      "step": 1366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995708497016005e-05,
+      "loss": 0.6264,
+      "step": 1367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956956285412243e-05,
+      "loss": 0.6424,
+      "step": 1368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956827408432642e-05,
+      "loss": 0.6299,
+      "step": 1369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956698339223735e-05,
+      "loss": 0.6809,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995656907778801e-05,
+      "loss": 0.6721,
+      "step": 1371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995643962412797e-05,
+      "loss": 0.6412,
+      "step": 1372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956309978246108e-05,
+      "loss": 0.612,
+      "step": 1373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956180140144927e-05,
+      "loss": 0.6013,
+      "step": 1374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995605010982694e-05,
+      "loss": 0.6844,
+      "step": 1375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995591988729465e-05,
+      "loss": 0.6618,
+      "step": 1376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955789472550577e-05,
+      "loss": 0.6249,
+      "step": 1377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955658865597237e-05,
+      "loss": 0.644,
+      "step": 1378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955528066437147e-05,
+      "loss": 0.6123,
+      "step": 1379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995539707507284e-05,
+      "loss": 0.6345,
+      "step": 1380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955265891506835e-05,
+      "loss": 0.6264,
+      "step": 1381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955134515741673e-05,
+      "loss": 0.6421,
+      "step": 1382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995500294777989e-05,
+      "loss": 0.6739,
+      "step": 1383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954871187624027e-05,
+      "loss": 0.634,
+      "step": 1384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995473923527662e-05,
+      "loss": 0.65,
+      "step": 1385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995460709074022e-05,
+      "loss": 0.6498,
+      "step": 1386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954474754017382e-05,
+      "loss": 0.6569,
+      "step": 1387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954342225110662e-05,
+      "loss": 0.645,
+      "step": 1388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995420950402261e-05,
+      "loss": 0.6611,
+      "step": 1389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954076590755794e-05,
+      "loss": 0.6361,
+      "step": 1390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953943485312782e-05,
+      "loss": 0.6284,
+      "step": 1391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995381018769614e-05,
+      "loss": 0.6314,
+      "step": 1392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953676697908443e-05,
+      "loss": 0.6654,
+      "step": 1393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995354301595227e-05,
+      "loss": 0.6716,
+      "step": 1394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953409141830194e-05,
+      "loss": 0.6522,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995327507554481e-05,
+      "loss": 0.6659,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99531408170987e-05,
+      "loss": 0.6904,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995300636649446e-05,
+      "loss": 0.6471,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952871723734685e-05,
+      "loss": 0.6905,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995273688882197e-05,
+      "loss": 0.609,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952601861758922e-05,
+      "loss": 0.638,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952466642548148e-05,
+      "loss": 0.6714,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952331231192256e-05,
+      "loss": 0.6323,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952195627693865e-05,
+      "loss": 0.619,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952059832055585e-05,
+      "loss": 0.6604,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951923844280047e-05,
+      "loss": 0.6464,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995178766436987e-05,
+      "loss": 0.6223,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995165129232769e-05,
+      "loss": 0.6314,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951514728156127e-05,
+      "loss": 0.6642,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995137797185783e-05,
+      "loss": 0.6193,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995124102343543e-05,
+      "loss": 0.6641,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951103882891583e-05,
+      "loss": 0.6945,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950966550228926e-05,
+      "loss": 0.6372,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950829025450116e-05,
+      "loss": 0.7012,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950691308557806e-05,
+      "loss": 0.6698,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995055339955465e-05,
+      "loss": 0.625,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995041529844332e-05,
+      "loss": 0.678,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950277005226476e-05,
+      "loss": 0.6247,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995013851990679e-05,
+      "loss": 0.6497,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949999842486933e-05,
+      "loss": 0.6409,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949860972969586e-05,
+      "loss": 0.6541,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994972191135743e-05,
+      "loss": 0.6871,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949582657653145e-05,
+      "loss": 0.6286,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949443211859427e-05,
+      "loss": 0.6548,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949303573978958e-05,
+      "loss": 0.6433,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949163744014446e-05,
+      "loss": 0.6277,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994902372196858e-05,
+      "loss": 0.6651,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994888350784407e-05,
+      "loss": 0.6841,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994874310164362e-05,
+      "loss": 0.664,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948602503369942e-05,
+      "loss": 0.6462,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994846171302575e-05,
+      "loss": 0.6474,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994832073061376e-05,
+      "loss": 0.6244,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948179556136696e-05,
+      "loss": 0.672,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948038189597283e-05,
+      "loss": 0.648,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947896630998253e-05,
+      "loss": 0.6485,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947754880342335e-05,
+      "loss": 0.6126,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947612937632267e-05,
+      "loss": 0.6454,
+      "step": 1437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994747080287079e-05,
+      "loss": 0.6607,
+      "step": 1438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994732847606065e-05,
+      "loss": 0.6198,
+      "step": 1439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947185957204588e-05,
+      "loss": 0.6572,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947043246305365e-05,
+      "loss": 0.6502,
+      "step": 1441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946900343365727e-05,
+      "loss": 0.7116,
+      "step": 1442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946757248388443e-05,
+      "loss": 0.6138,
+      "step": 1443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994661396137626e-05,
+      "loss": 0.6375,
+      "step": 1444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946470482331962e-05,
+      "loss": 0.5966,
+      "step": 1445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994632681125831e-05,
+      "loss": 0.6217,
+      "step": 1446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994618294815808e-05,
+      "loss": 0.6283,
+      "step": 1447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946038893034045e-05,
+      "loss": 0.6647,
+      "step": 1448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945894645888994e-05,
+      "loss": 0.6397,
+      "step": 1449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945750206725704e-05,
+      "loss": 0.6434,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994560557554697e-05,
+      "loss": 0.6236,
+      "step": 1451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945460752355577e-05,
+      "loss": 0.6538,
+      "step": 1452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994531573715433e-05,
+      "loss": 0.6472,
+      "step": 1453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994517052994602e-05,
+      "loss": 0.6223,
+      "step": 1454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994502513073346e-05,
+      "loss": 0.6252,
+      "step": 1455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944879539519447e-05,
+      "loss": 0.654,
+      "step": 1456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944733756306797e-05,
+      "loss": 0.6778,
+      "step": 1457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944587781098324e-05,
+      "loss": 0.6353,
+      "step": 1458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944441613896846e-05,
+      "loss": 0.6317,
+      "step": 1459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944295254705187e-05,
+      "loss": 0.6829,
+      "step": 1460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944148703526168e-05,
+      "loss": 0.6534,
+      "step": 1461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944001960362622e-05,
+      "loss": 0.6056,
+      "step": 1462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994385502521738e-05,
+      "loss": 0.7148,
+      "step": 1463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943707898093282e-05,
+      "loss": 0.6592,
+      "step": 1464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994356057899317e-05,
+      "loss": 0.6006,
+      "step": 1465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943413067919877e-05,
+      "loss": 0.6657,
+      "step": 1466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943265364876264e-05,
+      "loss": 0.6153,
+      "step": 1467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943117469865177e-05,
+      "loss": 0.6119,
+      "step": 1468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994296938288947e-05,
+      "loss": 0.6603,
+      "step": 1469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942821103952002e-05,
+      "loss": 0.6446,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994267263305564e-05,
+      "loss": 0.6628,
+      "step": 1471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942523970203248e-05,
+      "loss": 0.6419,
+      "step": 1472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994237511539769e-05,
+      "loss": 0.6527,
+      "step": 1473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994222606864185e-05,
+      "loss": 0.6965,
+      "step": 1474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99420768299386e-05,
+      "loss": 0.6492,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941927399290827e-05,
+      "loss": 0.6635,
+      "step": 1476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941777776701406e-05,
+      "loss": 0.6463,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941627962173232e-05,
+      "loss": 0.6859,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941477955709195e-05,
+      "loss": 0.6292,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941327757312194e-05,
+      "loss": 0.6596,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941177366985127e-05,
+      "loss": 0.6608,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941026784730898e-05,
+      "loss": 0.6342,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940876010552413e-05,
+      "loss": 0.6226,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940725044452583e-05,
+      "loss": 0.6587,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940573886434324e-05,
+      "loss": 0.6799,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940422536500555e-05,
+      "loss": 0.657,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940270994654196e-05,
+      "loss": 0.6273,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940119260898173e-05,
+      "loss": 0.6303,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939967335235412e-05,
+      "loss": 0.6558,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939815217668856e-05,
+      "loss": 0.6385,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939662908201432e-05,
+      "loss": 0.6056,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939510406836086e-05,
+      "loss": 0.6291,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939357713575757e-05,
+      "loss": 0.6329,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99392048284234e-05,
+      "loss": 0.6229,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993905175138196e-05,
+      "loss": 0.6279,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99388984824544e-05,
+      "loss": 0.626,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993874502164367e-05,
+      "loss": 0.6361,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993859136895274e-05,
+      "loss": 0.6465,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938437524384572e-05,
+      "loss": 0.6293,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938283487942137e-05,
+      "loss": 0.6711,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993812925962841e-05,
+      "loss": 0.6666,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937974839446368e-05,
+      "loss": 0.6434,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937820227398994e-05,
+      "loss": 0.6502,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937665423489272e-05,
+      "loss": 0.6369,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937510427720187e-05,
+      "loss": 0.6094,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937355240094738e-05,
+      "loss": 0.6425,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937199860615917e-05,
+      "loss": 0.677,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993704428928672e-05,
+      "loss": 0.676,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993688852611016e-05,
+      "loss": 0.6616,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993673257108924e-05,
+      "loss": 0.6543,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936576424226967e-05,
+      "loss": 0.6399,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993642008552636e-05,
+      "loss": 0.6638,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936263554990434e-05,
+      "loss": 0.6107,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936106832622217e-05,
+      "loss": 0.6371,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9935949918424727e-05,
+      "loss": 0.6532,
+      "step": 1515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935792812400997e-05,
+      "loss": 0.6355,
+      "step": 1516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935635514554062e-05,
+      "loss": 0.6566,
+      "step": 1517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935478024886955e-05,
+      "loss": 0.6319,
+      "step": 1518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935320343402722e-05,
+      "loss": 0.6422,
+      "step": 1519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99351624701044e-05,
+      "loss": 0.6413,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935004404995047e-05,
+      "loss": 0.6736,
+      "step": 1521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934846148077707e-05,
+      "loss": 0.628,
+      "step": 1522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934687699355435e-05,
+      "loss": 0.6757,
+      "step": 1523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993452905883129e-05,
+      "loss": 0.6758,
+      "step": 1524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934370226508338e-05,
+      "loss": 0.6616,
+      "step": 1525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934211202389646e-05,
+      "loss": 0.6373,
+      "step": 1526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934051986478283e-05,
+      "loss": 0.6497,
+      "step": 1527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993389257877732e-05,
+      "loss": 0.6565,
+      "step": 1528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933732979289833e-05,
+      "loss": 0.6752,
+      "step": 1529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933573188018913e-05,
+      "loss": 0.6488,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933413204967638e-05,
+      "loss": 0.6652,
+      "step": 1531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99332530301391e-05,
+      "loss": 0.6676,
+      "step": 1532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933092663536384e-05,
+      "loss": 0.672,
+      "step": 1533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932932105162594e-05,
+      "loss": 0.637,
+      "step": 1534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932771355020826e-05,
+      "loss": 0.6994,
+      "step": 1535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932610413114184e-05,
+      "loss": 0.6238,
+      "step": 1536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932449279445774e-05,
+      "loss": 0.6112,
+      "step": 1537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993228795401871e-05,
+      "loss": 0.6501,
+      "step": 1538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932126436836107e-05,
+      "loss": 0.6579,
+      "step": 1539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993196472790108e-05,
+      "loss": 0.6732,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931802827216754e-05,
+      "loss": 0.6372,
+      "step": 1541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931640734786252e-05,
+      "loss": 0.6464,
+      "step": 1542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931478450612706e-05,
+      "loss": 0.6272,
+      "step": 1543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931315974699245e-05,
+      "loss": 0.6635,
+      "step": 1544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931153307049008e-05,
+      "loss": 0.6489,
+      "step": 1545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930990447665138e-05,
+      "loss": 0.6455,
+      "step": 1546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930827396550776e-05,
+      "loss": 0.6236,
+      "step": 1547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993066415370907e-05,
+      "loss": 0.6712,
+      "step": 1548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930500719143175e-05,
+      "loss": 0.646,
+      "step": 1549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930337092856243e-05,
+      "loss": 0.6358,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993017327485143e-05,
+      "loss": 0.6437,
+      "step": 1551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930009265131912e-05,
+      "loss": 0.6272,
+      "step": 1552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929845063700838e-05,
+      "loss": 0.6704,
+      "step": 1553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992968067056139e-05,
+      "loss": 0.6588,
+      "step": 1554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929516085716736e-05,
+      "loss": 0.5638,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929351309170057e-05,
+      "loss": 0.6061,
+      "step": 1556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929186340924534e-05,
+      "loss": 0.6063,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929021180983347e-05,
+      "loss": 0.6544,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928855829349693e-05,
+      "loss": 0.6407,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928690286026754e-05,
+      "loss": 0.6226,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928524551017737e-05,
+      "loss": 0.6775,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928358624325833e-05,
+      "loss": 0.6434,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928192505954248e-05,
+      "loss": 0.6372,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928026195906192e-05,
+      "loss": 0.6229,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927859694184873e-05,
+      "loss": 0.5634,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992769300079351e-05,
+      "loss": 0.6272,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927526115735315e-05,
+      "loss": 0.6558,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992735903901351e-05,
+      "loss": 0.6321,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927191770631328e-05,
+      "loss": 0.6825,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927024310591992e-05,
+      "loss": 0.658,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926856658898734e-05,
+      "loss": 0.6166,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926688815554795e-05,
+      "loss": 0.6509,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926520780563413e-05,
+      "loss": 0.6089,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926352553927833e-05,
+      "loss": 0.6416,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926184135651302e-05,
+      "loss": 0.6409,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926015525737073e-05,
+      "loss": 0.6629,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99258467241884e-05,
+      "loss": 0.6647,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925677731008544e-05,
+      "loss": 0.6618,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925508546200762e-05,
+      "loss": 0.6202,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925339169768327e-05,
+      "loss": 0.6513,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925169601714508e-05,
+      "loss": 0.6471,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924999842042575e-05,
+      "loss": 0.6639,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924829890755804e-05,
+      "loss": 0.4837,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924659747857485e-05,
+      "loss": 0.6325,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924489413350894e-05,
+      "loss": 0.6418,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924318887239325e-05,
+      "loss": 0.6976,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924148169526068e-05,
+      "loss": 0.6552,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923977260214418e-05,
+      "loss": 0.6349,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923806159307677e-05,
+      "loss": 0.6562,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923634866809144e-05,
+      "loss": 0.6848,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923463382722132e-05,
+      "loss": 0.6376,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923291707049946e-05,
+      "loss": 0.6767,
+      "step": 1592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992311983979591e-05,
+      "loss": 0.6265,
+      "step": 1593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922947780963328e-05,
+      "loss": 0.7007,
+      "step": 1594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922775530555533e-05,
+      "loss": 0.6254,
+      "step": 1595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922603088575844e-05,
+      "loss": 0.6155,
+      "step": 1596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922430455027593e-05,
+      "loss": 0.6483,
+      "step": 1597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922257629914113e-05,
+      "loss": 0.6508,
+      "step": 1598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992208461323874e-05,
+      "loss": 0.6681,
+      "step": 1599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921911405004816e-05,
+      "loss": 0.6308,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921738005215687e-05,
+      "loss": 0.6536,
+      "step": 1601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992156441387469e-05,
+      "loss": 0.6416,
+      "step": 1602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921390630985188e-05,
+      "loss": 0.6507,
+      "step": 1603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921216656550534e-05,
+      "loss": 0.6163,
+      "step": 1604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992104249057408e-05,
+      "loss": 0.6089,
+      "step": 1605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920868133059196e-05,
+      "loss": 0.6485,
+      "step": 1606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920693584009246e-05,
+      "loss": 0.6591,
+      "step": 1607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920518843427595e-05,
+      "loss": 0.6636,
+      "step": 1608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920343911317626e-05,
+      "loss": 0.6614,
+      "step": 1609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992016878768271e-05,
+      "loss": 0.6315,
+      "step": 1610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919993472526227e-05,
+      "loss": 0.6649,
+      "step": 1611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919817965851566e-05,
+      "loss": 0.6718,
+      "step": 1612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919642267662113e-05,
+      "loss": 0.6181,
+      "step": 1613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919466377961258e-05,
+      "loss": 0.6484,
+      "step": 1614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99192902967524e-05,
+      "loss": 0.7034,
+      "step": 1615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991911402403894e-05,
+      "loss": 0.6101,
+      "step": 1616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918937559824276e-05,
+      "loss": 0.6236,
+      "step": 1617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918760904111818e-05,
+      "loss": 0.622,
+      "step": 1618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918584056904977e-05,
+      "loss": 0.6615,
+      "step": 1619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918407018207168e-05,
+      "loss": 0.5915,
+      "step": 1620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918229788021803e-05,
+      "loss": 0.637,
+      "step": 1621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991805236635231e-05,
+      "loss": 0.6289,
+      "step": 1622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917874753202114e-05,
+      "loss": 0.6525,
+      "step": 1623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991769694857464e-05,
+      "loss": 0.629,
+      "step": 1624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917518952473326e-05,
+      "loss": 0.6316,
+      "step": 1625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917340764901603e-05,
+      "loss": 0.6571,
+      "step": 1626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917162385862914e-05,
+      "loss": 0.6983,
+      "step": 1627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916983815360704e-05,
+      "loss": 0.6817,
+      "step": 1628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916805053398416e-05,
+      "loss": 0.6539,
+      "step": 1629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991662609997951e-05,
+      "loss": 0.6559,
+      "step": 1630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991644695510743e-05,
+      "loss": 0.6429,
+      "step": 1631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991626761878564e-05,
+      "loss": 0.6366,
+      "step": 1632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916088091017605e-05,
+      "loss": 0.6775,
+      "step": 1633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915908371806788e-05,
+      "loss": 0.6354,
+      "step": 1634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991572846115666e-05,
+      "loss": 0.6657,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915548359070687e-05,
+      "loss": 0.6851,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991536806555236e-05,
+      "loss": 0.6626,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915187580605147e-05,
+      "loss": 0.6602,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991500690423254e-05,
+      "loss": 0.6297,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914826036438025e-05,
+      "loss": 0.6464,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914644977225094e-05,
+      "loss": 0.6232,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914463726597238e-05,
+      "loss": 0.6702,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914282284557965e-05,
+      "loss": 0.6604,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914100651110774e-05,
+      "loss": 0.6585,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991391882625917e-05,
+      "loss": 0.6579,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913736810006664e-05,
+      "loss": 0.6518,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913554602356773e-05,
+      "loss": 0.6301,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991337220331301e-05,
+      "loss": 0.6397,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913189612878903e-05,
+      "loss": 0.6284,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913006831057967e-05,
+      "loss": 0.656,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912823857853742e-05,
+      "loss": 0.6623,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912640693269754e-05,
+      "loss": 0.652,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991245733730954e-05,
+      "loss": 0.6193,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991227378997664e-05,
+      "loss": 0.6625,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99120900512746e-05,
+      "loss": 0.6792,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911906121206964e-05,
+      "loss": 0.6861,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911721999777286e-05,
+      "loss": 0.6585,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911537686989116e-05,
+      "loss": 0.656,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991135318284602e-05,
+      "loss": 0.6027,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911168487351552e-05,
+      "loss": 0.6597,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910983600509283e-05,
+      "loss": 0.6793,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991079852232278e-05,
+      "loss": 0.6747,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991061325279562e-05,
+      "loss": 0.6403,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991042779193137e-05,
+      "loss": 0.6463,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910242139733625e-05,
+      "loss": 0.6111,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991005629620596e-05,
+      "loss": 0.6214,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909870261351964e-05,
+      "loss": 0.6593,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909684035175228e-05,
+      "loss": 0.6352,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.990949761767935e-05,
+      "loss": 0.6362,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909311008867926e-05,
+      "loss": 0.7098,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909124208744563e-05,
+      "loss": 0.6667,
+      "step": 1671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990893721731286e-05,
+      "loss": 0.6417,
+      "step": 1672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908750034576442e-05,
+      "loss": 0.6674,
+      "step": 1673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908562660538905e-05,
+      "loss": 0.6414,
+      "step": 1674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908375095203876e-05,
+      "loss": 0.6264,
+      "step": 1675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908187338574974e-05,
+      "loss": 0.6707,
+      "step": 1676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907999390655824e-05,
+      "loss": 0.6675,
+      "step": 1677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907811251450058e-05,
+      "loss": 0.6248,
+      "step": 1678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99076229209613e-05,
+      "loss": 0.6496,
+      "step": 1679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907434399193196e-05,
+      "loss": 0.6243,
+      "step": 1680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990724568614938e-05,
+      "loss": 0.6765,
+      "step": 1681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907056781833495e-05,
+      "loss": 0.6642,
+      "step": 1682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906867686249188e-05,
+      "loss": 0.6096,
+      "step": 1683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990667839940011e-05,
+      "loss": 0.6407,
+      "step": 1684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990648892128992e-05,
+      "loss": 0.6196,
+      "step": 1685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906299251922273e-05,
+      "loss": 0.6475,
+      "step": 1686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990610939130083e-05,
+      "loss": 0.6458,
+      "step": 1687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905919339429254e-05,
+      "loss": 0.6383,
+      "step": 1688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905729096311218e-05,
+      "loss": 0.6498,
+      "step": 1689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905538661950395e-05,
+      "loss": 0.6457,
+      "step": 1690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905348036350458e-05,
+      "loss": 0.6542,
+      "step": 1691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905157219515093e-05,
+      "loss": 0.6465,
+      "step": 1692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990496621144798e-05,
+      "loss": 0.6405,
+      "step": 1693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904775012152803e-05,
+      "loss": 0.6522,
+      "step": 1694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990458362163326e-05,
+      "loss": 0.6252,
+      "step": 1695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904392039893046e-05,
+      "loss": 0.6637,
+      "step": 1696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904200266935855e-05,
+      "loss": 0.6236,
+      "step": 1697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904008302765394e-05,
+      "loss": 0.6459,
+      "step": 1698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903816147385367e-05,
+      "loss": 0.6186,
+      "step": 1699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903623800799483e-05,
+      "loss": 0.5951,
+      "step": 1700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903431263011453e-05,
+      "loss": 0.6464,
+      "step": 1701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903238534025003e-05,
+      "loss": 0.6874,
+      "step": 1702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903045613843844e-05,
+      "loss": 0.6699,
+      "step": 1703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990285250247171e-05,
+      "loss": 0.5975,
+      "step": 1704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990265919991232e-05,
+      "loss": 0.6378,
+      "step": 1705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902465706169412e-05,
+      "loss": 0.6471,
+      "step": 1706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902272021246718e-05,
+      "loss": 0.6708,
+      "step": 1707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902078145147983e-05,
+      "loss": 0.6409,
+      "step": 1708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901884077876944e-05,
+      "loss": 0.6525,
+      "step": 1709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901689819437345e-05,
+      "loss": 0.6464,
+      "step": 1710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901495369832947e-05,
+      "loss": 0.6232,
+      "step": 1711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901300729067497e-05,
+      "loss": 0.6434,
+      "step": 1712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901105897144753e-05,
+      "loss": 0.6641,
+      "step": 1713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900910874068482e-05,
+      "loss": 0.5973,
+      "step": 1714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990071565984244e-05,
+      "loss": 0.6474,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99005202544704e-05,
+      "loss": 0.6434,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990032465795614e-05,
+      "loss": 0.6246,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990012887030343e-05,
+      "loss": 0.6517,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899932891516048e-05,
+      "loss": 0.6304,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899736721597787e-05,
+      "loss": 0.6304,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899540360552423e-05,
+      "loss": 0.6271,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899343808383754e-05,
+      "loss": 0.6396,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899147065095576e-05,
+      "loss": 0.6708,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989895013069168e-05,
+      "loss": 0.671,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898753005175874e-05,
+      "loss": 0.5874,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989855568855196e-05,
+      "loss": 0.6581,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898358180823754e-05,
+      "loss": 0.6297,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898160481995062e-05,
+      "loss": 0.6243,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897962592069705e-05,
+      "loss": 0.6274,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897764511051504e-05,
+      "loss": 0.6473,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897566238944275e-05,
+      "loss": 0.6463,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989736777575186e-05,
+      "loss": 0.632,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897169121478076e-05,
+      "loss": 0.5954,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989697027612677e-05,
+      "loss": 0.6245,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896771239701772e-05,
+      "loss": 0.6352,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896572012206933e-05,
+      "loss": 0.672,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896372593646095e-05,
+      "loss": 0.6823,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896172984023106e-05,
+      "loss": 0.6439,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989597318334182e-05,
+      "loss": 0.6327,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895773191606103e-05,
+      "loss": 0.6573,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895573008819804e-05,
+      "loss": 0.6523,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895372634986793e-05,
+      "loss": 0.6206,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989517207011094e-05,
+      "loss": 0.6306,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894971314196117e-05,
+      "loss": 0.6377,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894770367246197e-05,
+      "loss": 0.6247,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894569229265062e-05,
+      "loss": 0.6592,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989436790025659e-05,
+      "loss": 0.6424,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989416638022468e-05,
+      "loss": 0.6166,
+      "step": 1748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989396466917321e-05,
+      "loss": 0.6375,
+      "step": 1749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989376276710608e-05,
+      "loss": 0.6317,
+      "step": 1750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989356067402719e-05,
+      "loss": 0.6332,
+      "step": 1751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893358389940433e-05,
+      "loss": 0.6209,
+      "step": 1752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893155914849727e-05,
+      "loss": 0.6292,
+      "step": 1753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989295324875897e-05,
+      "loss": 0.634,
+      "step": 1754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989275039167208e-05,
+      "loss": 0.6202,
+      "step": 1755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989254734359297e-05,
+      "loss": 0.6331,
+      "step": 1756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892344104525566e-05,
+      "loss": 0.6224,
+      "step": 1757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892140674473784e-05,
+      "loss": 0.6487,
+      "step": 1758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989193705344156e-05,
+      "loss": 0.6279,
+      "step": 1759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989173324143282e-05,
+      "loss": 0.6626,
+      "step": 1760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98915292384515e-05,
+      "loss": 0.6045,
+      "step": 1761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891325044501538e-05,
+      "loss": 0.6621,
+      "step": 1762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891120659586875e-05,
+      "loss": 0.6385,
+      "step": 1763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890916083711463e-05,
+      "loss": 0.6557,
+      "step": 1764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890711316879243e-05,
+      "loss": 0.6916,
+      "step": 1765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890506359094172e-05,
+      "loss": 0.6281,
+      "step": 1766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989030121036021e-05,
+      "loss": 0.6495,
+      "step": 1767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890095870681312e-05,
+      "loss": 0.6551,
+      "step": 1768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889890340061448e-05,
+      "loss": 0.6709,
+      "step": 1769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988968461850458e-05,
+      "loss": 0.6355,
+      "step": 1770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889478706014687e-05,
+      "loss": 0.6693,
+      "step": 1771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988927260259574e-05,
+      "loss": 0.6481,
+      "step": 1772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889066308251715e-05,
+      "loss": 0.6494,
+      "step": 1773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98888598229866e-05,
+      "loss": 0.6381,
+      "step": 1774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988865314680438e-05,
+      "loss": 0.6374,
+      "step": 1775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888446279709046e-05,
+      "loss": 0.5887,
+      "step": 1776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988823922170459e-05,
+      "loss": 0.4352,
+      "step": 1777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888031972795014e-05,
+      "loss": 0.6079,
+      "step": 1778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988782453298431e-05,
+      "loss": 0.6859,
+      "step": 1779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988761690227649e-05,
+      "loss": 0.664,
+      "step": 1780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887409080675558e-05,
+      "loss": 0.6164,
+      "step": 1781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887201068185537e-05,
+      "loss": 0.6452,
+      "step": 1782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886992864810427e-05,
+      "loss": 0.6617,
+      "step": 1783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988678447055426e-05,
+      "loss": 0.6177,
+      "step": 1784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886575885421055e-05,
+      "loss": 0.648,
+      "step": 1785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988636710941484e-05,
+      "loss": 0.628,
+      "step": 1786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988615814253964e-05,
+      "loss": 0.6647,
+      "step": 1787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885948984799502e-05,
+      "loss": 0.6066,
+      "step": 1788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885739636198452e-05,
+      "loss": 0.6772,
+      "step": 1789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885530096740538e-05,
+      "loss": 0.5922,
+      "step": 1790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98853203664298e-05,
+      "loss": 0.6581,
+      "step": 1791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885110445270293e-05,
+      "loss": 0.6274,
+      "step": 1792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884900333266065e-05,
+      "loss": 0.6212,
+      "step": 1793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884690030421177e-05,
+      "loss": 0.6083,
+      "step": 1794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884479536739688e-05,
+      "loss": 0.679,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884268852225657e-05,
+      "loss": 0.5946,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884057976883156e-05,
+      "loss": 0.6153,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988384691071626e-05,
+      "loss": 0.641,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883635653729034e-05,
+      "loss": 0.6189,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883424205925567e-05,
+      "loss": 0.6589,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988321256730993e-05,
+      "loss": 0.5898,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883000737886222e-05,
+      "loss": 0.6634,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882788717658523e-05,
+      "loss": 0.6209,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882576506630927e-05,
+      "loss": 0.6213,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882364104807536e-05,
+      "loss": 0.6416,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988215151219245e-05,
+      "loss": 0.6597,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881938728789768e-05,
+      "loss": 0.6236,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881725754603606e-05,
+      "loss": 0.7011,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881512589638065e-05,
+      "loss": 0.6585,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988129923389727e-05,
+      "loss": 0.6309,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881085687385337e-05,
+      "loss": 0.6742,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880871950106386e-05,
+      "loss": 0.6624,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880658022064544e-05,
+      "loss": 0.6176,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988044390326395e-05,
+      "loss": 0.6514,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880229593708726e-05,
+      "loss": 0.6162,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880015093403015e-05,
+      "loss": 0.6215,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987980040235096e-05,
+      "loss": 0.6193,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879585520556702e-05,
+      "loss": 0.6318,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987937044802439e-05,
+      "loss": 0.6345,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879155184758175e-05,
+      "loss": 0.5925,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878939730762222e-05,
+      "loss": 0.6344,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987872408604068e-05,
+      "loss": 0.4283,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878508250597718e-05,
+      "loss": 0.5957,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878292224437497e-05,
+      "loss": 0.613,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878076007564197e-05,
+      "loss": 0.6213,
+      "step": 1825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877859599981984e-05,
+      "loss": 0.6369,
+      "step": 1826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987764300169504e-05,
+      "loss": 0.6027,
+      "step": 1827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877426212707546e-05,
+      "loss": 0.6474,
+      "step": 1828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877209233023688e-05,
+      "loss": 0.6242,
+      "step": 1829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876992062647652e-05,
+      "loss": 0.6402,
+      "step": 1830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876774701583636e-05,
+      "loss": 0.6246,
+      "step": 1831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987655714983583e-05,
+      "loss": 0.6614,
+      "step": 1832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987633940740844e-05,
+      "loss": 0.6088,
+      "step": 1833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987612147430567e-05,
+      "loss": 0.6223,
+      "step": 1834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987590335053172e-05,
+      "loss": 0.6011,
+      "step": 1835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987568503609081e-05,
+      "loss": 0.5983,
+      "step": 1836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987546653098715e-05,
+      "loss": 0.6702,
+      "step": 1837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875247835224957e-05,
+      "loss": 0.6669,
+      "step": 1838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875028948808457e-05,
+      "loss": 0.6409,
+      "step": 1839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874809871741877e-05,
+      "loss": 0.6461,
+      "step": 1840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987459060402944e-05,
+      "loss": 0.6472,
+      "step": 1841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874371145675387e-05,
+      "loss": 0.5826,
+      "step": 1842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874151496683948e-05,
+      "loss": 0.6303,
+      "step": 1843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987393165705937e-05,
+      "loss": 0.6345,
+      "step": 1844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987371162680589e-05,
+      "loss": 0.6391,
+      "step": 1845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873491405927764e-05,
+      "loss": 0.6064,
+      "step": 1846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987327099442924e-05,
+      "loss": 0.6318,
+      "step": 1847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987305039231457e-05,
+      "loss": 0.6633,
+      "step": 1848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987282959958802e-05,
+      "loss": 0.6296,
+      "step": 1849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872608616253846e-05,
+      "loss": 0.6551,
+      "step": 1850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872387442316316e-05,
+      "loss": 0.6549,
+      "step": 1851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872166077779705e-05,
+      "loss": 0.614,
+      "step": 1852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871944522648282e-05,
+      "loss": 0.617,
+      "step": 1853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871722776926322e-05,
+      "loss": 0.6529,
+      "step": 1854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987150084061811e-05,
+      "loss": 0.6065,
+      "step": 1855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871278713727932e-05,
+      "loss": 0.6672,
+      "step": 1856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871056396260075e-05,
+      "loss": 0.6067,
+      "step": 1857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987083388821883e-05,
+      "loss": 0.6767,
+      "step": 1858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870611189608494e-05,
+      "loss": 0.6202,
+      "step": 1859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870388300433366e-05,
+      "loss": 0.6834,
+      "step": 1860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987016522069775e-05,
+      "loss": 0.6449,
+      "step": 1861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869941950405952e-05,
+      "loss": 0.6529,
+      "step": 1862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869718489562282e-05,
+      "loss": 0.6243,
+      "step": 1863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869494838171052e-05,
+      "loss": 0.6184,
+      "step": 1864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869270996236586e-05,
+      "loss": 0.6209,
+      "step": 1865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869046963763203e-05,
+      "loss": 0.6342,
+      "step": 1866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868822740755224e-05,
+      "loss": 0.6681,
+      "step": 1867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868598327216982e-05,
+      "loss": 0.6662,
+      "step": 1868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868373723152814e-05,
+      "loss": 0.6589,
+      "step": 1869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868148928567046e-05,
+      "loss": 0.6683,
+      "step": 1870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867923943464023e-05,
+      "loss": 0.6384,
+      "step": 1871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986769876784809e-05,
+      "loss": 0.6397,
+      "step": 1872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867473401723595e-05,
+      "loss": 0.6348,
+      "step": 1873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867247845094882e-05,
+      "loss": 0.6573,
+      "step": 1874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867022097966314e-05,
+      "loss": 0.6175,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986679616034225e-05,
+      "loss": 0.6525,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986657003222704e-05,
+      "loss": 0.6484,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866343713625066e-05,
+      "loss": 0.6418,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866117204540685e-05,
+      "loss": 0.6387,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865890504978273e-05,
+      "loss": 0.6387,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986566361494221e-05,
+      "loss": 0.6585,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865436534436874e-05,
+      "loss": 0.5972,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865209263466646e-05,
+      "loss": 0.6162,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864981802035923e-05,
+      "loss": 0.5933,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864754150149086e-05,
+      "loss": 0.6612,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864526307810537e-05,
+      "loss": 0.6489,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864298275024673e-05,
+      "loss": 0.648,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986407005179589e-05,
+      "loss": 0.628,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986384163812861e-05,
+      "loss": 0.6393,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 0.6528,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986338423949616e-05,
+      "loss": 0.638,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863155254539828e-05,
+      "loss": 0.6475,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986292607916265e-05,
+      "loss": 0.6841,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986269671336905e-05,
+      "loss": 0.669,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862467157163458e-05,
+      "loss": 0.6374,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862237410550306e-05,
+      "loss": 0.6035,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862007473534026e-05,
+      "loss": 0.6868,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861777346119062e-05,
+      "loss": 0.6522,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861547028309855e-05,
+      "loss": 0.6087,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986131652011085e-05,
+      "loss": 0.6657,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98610858215265e-05,
+      "loss": 0.629,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860854932561254e-05,
+      "loss": 0.6472,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860623853219578e-05,
+      "loss": 0.6374,
+      "step": 1903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860392583505926e-05,
+      "loss": 0.7072,
+      "step": 1904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860161123424766e-05,
+      "loss": 0.6215,
+      "step": 1905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859929472980567e-05,
+      "loss": 0.6168,
+      "step": 1906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859697632177796e-05,
+      "loss": 0.6534,
+      "step": 1907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859465601020935e-05,
+      "loss": 0.4146,
+      "step": 1908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859233379514465e-05,
+      "loss": 0.6463,
+      "step": 1909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859000967662864e-05,
+      "loss": 0.6376,
+      "step": 1910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985876836547062e-05,
+      "loss": 0.6247,
+      "step": 1911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858535572942225e-05,
+      "loss": 0.605,
+      "step": 1912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858302590082177e-05,
+      "loss": 0.6248,
+      "step": 1913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858069416894964e-05,
+      "loss": 0.6641,
+      "step": 1914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98578360533851e-05,
+      "loss": 0.6425,
+      "step": 1915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985760249955708e-05,
+      "loss": 0.6393,
+      "step": 1916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857368755415424e-05,
+      "loss": 0.662,
+      "step": 1917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985713482096463e-05,
+      "loss": 0.6314,
+      "step": 1918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985690069620923e-05,
+      "loss": 0.6307,
+      "step": 1919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856666381153733e-05,
+      "loss": 0.6264,
+      "step": 1920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856431875802665e-05,
+      "loss": 0.6278,
+      "step": 1921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856197180160557e-05,
+      "loss": 0.6269,
+      "step": 1922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855962294231938e-05,
+      "loss": 0.6235,
+      "step": 1923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985572721802134e-05,
+      "loss": 0.6066,
+      "step": 1924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855491951533308e-05,
+      "loss": 0.6192,
+      "step": 1925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985525649477238e-05,
+      "loss": 0.6259,
+      "step": 1926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98550208477431e-05,
+      "loss": 0.6542,
+      "step": 1927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985478501045002e-05,
+      "loss": 0.6868,
+      "step": 1928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854548982897692e-05,
+      "loss": 0.6468,
+      "step": 1929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854312765090675e-05,
+      "loss": 0.6418,
+      "step": 1930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854076357033526e-05,
+      "loss": 0.646,
+      "step": 1931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853839758730812e-05,
+      "loss": 0.6081,
+      "step": 1932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853602970187104e-05,
+      "loss": 0.6144,
+      "step": 1933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853365991406965e-05,
+      "loss": 0.6177,
+      "step": 1934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853128822394976e-05,
+      "loss": 0.633,
+      "step": 1935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852891463155714e-05,
+      "loss": 0.6338,
+      "step": 1936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985265391369376e-05,
+      "loss": 0.6328,
+      "step": 1937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985241617401371e-05,
+      "loss": 0.6198,
+      "step": 1938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985217824412014e-05,
+      "loss": 0.6232,
+      "step": 1939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985194012401765e-05,
+      "loss": 0.6193,
+      "step": 1940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851701813710838e-05,
+      "loss": 0.6393,
+      "step": 1941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851463313204303e-05,
+      "loss": 0.6387,
+      "step": 1942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985122462250265e-05,
+      "loss": 0.6215,
+      "step": 1943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850985741610487e-05,
+      "loss": 0.6631,
+      "step": 1944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985074667053243e-05,
+      "loss": 0.5987,
+      "step": 1945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850507409273088e-05,
+      "loss": 0.6282,
+      "step": 1946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850267957837083e-05,
+      "loss": 0.6363,
+      "step": 1947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985002831622904e-05,
+      "loss": 0.6419,
+      "step": 1948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984978848445358e-05,
+      "loss": 0.644,
+      "step": 1949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984954846251534e-05,
+      "loss": 0.6245,
+      "step": 1950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849308250418953e-05,
+      "loss": 0.6587,
+      "step": 1951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984906784816905e-05,
+      "loss": 0.6608,
+      "step": 1952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984882725577028e-05,
+      "loss": 0.6076,
+      "step": 1953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848586473227286e-05,
+      "loss": 0.6639,
+      "step": 1954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848345500544712e-05,
+      "loss": 0.6276,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848104337727213e-05,
+      "loss": 0.6516,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984786298477945e-05,
+      "loss": 0.6404,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847621441706076e-05,
+      "loss": 0.6172,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847379708511756e-05,
+      "loss": 0.6253,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847137785201158e-05,
+      "loss": 0.6576,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984689567177895e-05,
+      "loss": 0.6593,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846653368249812e-05,
+      "loss": 0.6644,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984641087461842e-05,
+      "loss": 0.6275,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984616819088945e-05,
+      "loss": 0.625,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845925317067594e-05,
+      "loss": 0.6191,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845682253157537e-05,
+      "loss": 0.669,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984543899916397e-05,
+      "loss": 0.6648,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845195555091598e-05,
+      "loss": 0.5974,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844951920945113e-05,
+      "loss": 0.6578,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844708096729223e-05,
+      "loss": 0.6412,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984446408244863e-05,
+      "loss": 0.6438,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984421987810805e-05,
+      "loss": 0.6605,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843975483712198e-05,
+      "loss": 0.6844,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843730899265786e-05,
+      "loss": 0.6232,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843486124773546e-05,
+      "loss": 0.639,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843241160240194e-05,
+      "loss": 0.6759,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842996005670464e-05,
+      "loss": 0.6752,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842750661069092e-05,
+      "loss": 0.6049,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842505126440808e-05,
+      "loss": 0.609,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842259401790356e-05,
+      "loss": 0.6509,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984201348712248e-05,
+      "loss": 0.6463,
+      "step": 1981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984176738244193e-05,
+      "loss": 0.6157,
+      "step": 1982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841521087753447e-05,
+      "loss": 0.6683,
+      "step": 1983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841274603061798e-05,
+      "loss": 0.6351,
+      "step": 1984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841027928371738e-05,
+      "loss": 0.6611,
+      "step": 1985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984078106368803e-05,
+      "loss": 0.6249,
+      "step": 1986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984053400901544e-05,
+      "loss": 0.6432,
+      "step": 1987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840286764358732e-05,
+      "loss": 0.6072,
+      "step": 1988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984003932972269e-05,
+      "loss": 0.6384,
+      "step": 1989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839791705112077e-05,
+      "loss": 0.638,
+      "step": 1990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839543890531687e-05,
+      "loss": 0.6224,
+      "step": 1991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 0.6736,
+      "step": 1992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98390476914807e-05,
+      "loss": 0.6219,
+      "step": 1993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983879930701968e-05,
+      "loss": 0.6231,
+      "step": 1994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838550732608042e-05,
+      "loss": 0.6828,
+      "step": 1995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983830196825058e-05,
+      "loss": 0.6524,
+      "step": 1996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838053013952093e-05,
+      "loss": 0.5933,
+      "step": 1997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837803869717394e-05,
+      "loss": 0.3901,
+      "step": 1998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837554535551288e-05,
+      "loss": 0.6244,
+      "step": 1999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837305011458595e-05,
+      "loss": 0.6176,
+      "step": 2000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837055297444124e-05,
+      "loss": 0.6668,
+      "step": 2001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836805393512706e-05,
+      "loss": 0.632,
+      "step": 2002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836555299669157e-05,
+      "loss": 0.6466,
+      "step": 2003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836305015918306e-05,
+      "loss": 0.6098,
+      "step": 2004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983605454226499e-05,
+      "loss": 0.636,
+      "step": 2005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835803878714045e-05,
+      "loss": 0.654,
+      "step": 2006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835553025270308e-05,
+      "loss": 0.6519,
+      "step": 2007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835301981938617e-05,
+      "loss": 0.6492,
+      "step": 2008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835050748723826e-05,
+      "loss": 0.6465,
+      "step": 2009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834799325630784e-05,
+      "loss": 0.6326,
+      "step": 2010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983454771266434e-05,
+      "loss": 0.6367,
+      "step": 2011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834295909829358e-05,
+      "loss": 0.6152,
+      "step": 2012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834043917130698e-05,
+      "loss": 0.6109,
+      "step": 2013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833791734573225e-05,
+      "loss": 0.6261,
+      "step": 2014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833539362161804e-05,
+      "loss": 0.6122,
+      "step": 2015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833286799901314e-05,
+      "loss": 0.6201,
+      "step": 2016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833034047796624e-05,
+      "loss": 0.628,
+      "step": 2017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832781105852615e-05,
+      "loss": 0.6037,
+      "step": 2018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832527974074175e-05,
+      "loss": 0.6236,
+      "step": 2019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983227465246619e-05,
+      "loss": 0.6593,
+      "step": 2020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832021141033547e-05,
+      "loss": 0.6651,
+      "step": 2021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983176743978114e-05,
+      "loss": 0.6589,
+      "step": 2022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831513548713873e-05,
+      "loss": 0.6544,
+      "step": 2023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831259467836645e-05,
+      "loss": 0.648,
+      "step": 2024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831005197154356e-05,
+      "loss": 0.6346,
+      "step": 2025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830750736671923e-05,
+      "loss": 0.6306,
+      "step": 2026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830496086394255e-05,
+      "loss": 0.6493,
+      "step": 2027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830241246326265e-05,
+      "loss": 0.6314,
+      "step": 2028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982998621647288e-05,
+      "loss": 0.6242,
+      "step": 2029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 0.6526,
+      "step": 2030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829475587429613e-05,
+      "loss": 0.6273,
+      "step": 2031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829219988249587e-05,
+      "loss": 0.6184,
+      "step": 2032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982896419930388e-05,
+      "loss": 0.6465,
+      "step": 2033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982870822059743e-05,
+      "loss": 0.5841,
+      "step": 2034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982845205213518e-05,
+      "loss": 0.6348,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982819569392207e-05,
+      "loss": 0.6434,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982793914596306e-05,
+      "loss": 0.6338,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827682408263088e-05,
+      "loss": 0.6417,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827425480827127e-05,
+      "loss": 0.6309,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827168363660126e-05,
+      "loss": 0.6273,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826911056767052e-05,
+      "loss": 0.6085,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826653560152875e-05,
+      "loss": 0.6359,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982639587382256e-05,
+      "loss": 0.6522,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826137997781093e-05,
+      "loss": 0.6194,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825879932033444e-05,
+      "loss": 0.6315,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825621676584595e-05,
+      "loss": 0.5991,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825363231439532e-05,
+      "loss": 0.6417,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982510459660325e-05,
+      "loss": 0.6416,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982484577208074e-05,
+      "loss": 0.6549,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824586757876995e-05,
+      "loss": 0.6454,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824327553997023e-05,
+      "loss": 0.6036,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824068160445822e-05,
+      "loss": 0.6235,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98238085772284e-05,
+      "loss": 0.6474,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823548804349773e-05,
+      "loss": 0.6626,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982328884181495e-05,
+      "loss": 0.6216,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823028689628957e-05,
+      "loss": 0.6385,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822768347796812e-05,
+      "loss": 0.6224,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822507816323545e-05,
+      "loss": 0.6647,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982224709521418e-05,
+      "loss": 0.61,
+      "step": 2059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821986184473757e-05,
+      "loss": 0.6542,
+      "step": 2060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821725084107307e-05,
+      "loss": 0.6141,
+      "step": 2061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821463794119875e-05,
+      "loss": 0.6552,
+      "step": 2062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821202314516504e-05,
+      "loss": 0.6735,
+      "step": 2063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820940645302244e-05,
+      "loss": 0.6452,
+      "step": 2064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982067878648214e-05,
+      "loss": 0.6288,
+      "step": 2065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982041673806126e-05,
+      "loss": 0.6433,
+      "step": 2066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820154500044655e-05,
+      "loss": 0.6022,
+      "step": 2067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819892072437388e-05,
+      "loss": 0.6326,
+      "step": 2068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819629455244524e-05,
+      "loss": 0.5979,
+      "step": 2069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981936664847114e-05,
+      "loss": 0.6401,
+      "step": 2070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819103652122304e-05,
+      "loss": 0.6134,
+      "step": 2071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818840466203092e-05,
+      "loss": 0.6338,
+      "step": 2072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818577090718595e-05,
+      "loss": 0.6302,
+      "step": 2073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818313525673884e-05,
+      "loss": 0.6257,
+      "step": 2074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818049771074058e-05,
+      "loss": 0.7112,
+      "step": 2075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817785826924206e-05,
+      "loss": 0.6489,
+      "step": 2076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981752169322942e-05,
+      "loss": 0.6605,
+      "step": 2077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817257369994806e-05,
+      "loss": 0.6276,
+      "step": 2078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816992857225462e-05,
+      "loss": 0.6246,
+      "step": 2079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816728154926496e-05,
+      "loss": 0.6447,
+      "step": 2080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981646326310302e-05,
+      "loss": 0.6135,
+      "step": 2081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816198181760148e-05,
+      "loss": 0.6826,
+      "step": 2082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815932910902998e-05,
+      "loss": 0.6578,
+      "step": 2083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981566745053669e-05,
+      "loss": 0.6039,
+      "step": 2084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815401800666346e-05,
+      "loss": 0.3769,
+      "step": 2085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815135961297097e-05,
+      "loss": 0.6415,
+      "step": 2086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814869932434083e-05,
+      "loss": 0.6505,
+      "step": 2087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981460371408243e-05,
+      "loss": 0.3598,
+      "step": 2088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981433730624728e-05,
+      "loss": 0.6934,
+      "step": 2089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814070708933778e-05,
+      "loss": 0.6646,
+      "step": 2090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981380392214707e-05,
+      "loss": 0.5848,
+      "step": 2091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981353694589231e-05,
+      "loss": 0.6221,
+      "step": 2092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813269780174645e-05,
+      "loss": 0.6442,
+      "step": 2093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981300242499924e-05,
+      "loss": 0.6212,
+      "step": 2094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812734880371253e-05,
+      "loss": 0.6428,
+      "step": 2095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981246714629585e-05,
+      "loss": 0.6246,
+      "step": 2096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812199222778204e-05,
+      "loss": 0.6458,
+      "step": 2097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981193110982348e-05,
+      "loss": 0.6221,
+      "step": 2098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981166280743686e-05,
+      "loss": 0.6619,
+      "step": 2099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811394315623523e-05,
+      "loss": 0.628,
+      "step": 2100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981112563438865e-05,
+      "loss": 0.6573,
+      "step": 2101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810856763737428e-05,
+      "loss": 0.6411,
+      "step": 2102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981058770367505e-05,
+      "loss": 0.6559,
+      "step": 2103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810318454206715e-05,
+      "loss": 0.6459,
+      "step": 2104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810049015337614e-05,
+      "loss": 0.6226,
+      "step": 2105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980977938707295e-05,
+      "loss": 0.6086,
+      "step": 2106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980950956941793e-05,
+      "loss": 0.6255,
+      "step": 2107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809239562377764e-05,
+      "loss": 0.6168,
+      "step": 2108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808969365957666e-05,
+      "loss": 0.6509,
+      "step": 2109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808698980162847e-05,
+      "loss": 0.6292,
+      "step": 2110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808428404998532e-05,
+      "loss": 0.3799,
+      "step": 2111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808157640469943e-05,
+      "loss": 0.638,
+      "step": 2112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807886686582307e-05,
+      "loss": 0.5846,
+      "step": 2113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807615543340855e-05,
+      "loss": 0.676,
+      "step": 2114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807344210750826e-05,
+      "loss": 0.6488,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807072688817454e-05,
+      "loss": 0.6284,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806800977545984e-05,
+      "loss": 0.6382,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806529076941658e-05,
+      "loss": 0.6267,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806256987009726e-05,
+      "loss": 0.6372,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805984707755444e-05,
+      "loss": 0.6115,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805712239184064e-05,
+      "loss": 0.6045,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805439581300855e-05,
+      "loss": 0.6198,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805166734111073e-05,
+      "loss": 0.6498,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804893697619986e-05,
+      "loss": 0.6149,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980462047183287e-05,
+      "loss": 0.6201,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804347056754993e-05,
+      "loss": 0.6266,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804073452391642e-05,
+      "loss": 0.6244,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803799658748096e-05,
+      "loss": 0.6107,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803525675829638e-05,
+      "loss": 0.6253,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803251503641562e-05,
+      "loss": 0.6268,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802977142189155e-05,
+      "loss": 0.616,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980270259147772e-05,
+      "loss": 0.6209,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802427851512558e-05,
+      "loss": 0.6248,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802152922298967e-05,
+      "loss": 0.6215,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980187780384226e-05,
+      "loss": 0.6051,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980160249614775e-05,
+      "loss": 0.6617,
+      "step": 2136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801326999220743e-05,
+      "loss": 0.669,
+      "step": 2137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980105131306657e-05,
+      "loss": 0.6534,
+      "step": 2138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800775437690544e-05,
+      "loss": 0.6598,
+      "step": 2139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800499373097998e-05,
+      "loss": 0.6514,
+      "step": 2140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800223119294254e-05,
+      "loss": 0.6628,
+      "step": 2141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799946676284657e-05,
+      "loss": 0.6528,
+      "step": 2142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979967004407453e-05,
+      "loss": 0.5804,
+      "step": 2143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799393222669224e-05,
+      "loss": 0.6404,
+      "step": 2144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799116212074077e-05,
+      "loss": 0.6506,
+      "step": 2145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798839012294444e-05,
+      "loss": 0.6197,
+      "step": 2146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979856162333567e-05,
+      "loss": 0.5976,
+      "step": 2147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798284045203115e-05,
+      "loss": 0.6729,
+      "step": 2148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798006277902137e-05,
+      "loss": 0.6437,
+      "step": 2149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797728321438096e-05,
+      "loss": 0.6002,
+      "step": 2150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979745017581636e-05,
+      "loss": 0.6449,
+      "step": 2151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97971718410423e-05,
+      "loss": 0.6648,
+      "step": 2152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979689331712129e-05,
+      "loss": 0.6188,
+      "step": 2153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796614604058705e-05,
+      "loss": 0.5948,
+      "step": 2154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796335701859924e-05,
+      "loss": 0.6465,
+      "step": 2155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979605661053034e-05,
+      "loss": 0.612,
+      "step": 2156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979577733007533e-05,
+      "loss": 0.3542,
+      "step": 2157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795497860500296e-05,
+      "loss": 0.6469,
+      "step": 2158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795218201810627e-05,
+      "loss": 0.6298,
+      "step": 2159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794938354011725e-05,
+      "loss": 0.6964,
+      "step": 2160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979465831710899e-05,
+      "loss": 0.5995,
+      "step": 2161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794378091107834e-05,
+      "loss": 0.6752,
+      "step": 2162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979409767601366e-05,
+      "loss": 0.6182,
+      "step": 2163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793817071831887e-05,
+      "loss": 0.6088,
+      "step": 2164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979353627856793e-05,
+      "loss": 0.6239,
+      "step": 2165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793255296227214e-05,
+      "loss": 0.6483,
+      "step": 2166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792974124815157e-05,
+      "loss": 0.6125,
+      "step": 2167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792692764337194e-05,
+      "loss": 0.6279,
+      "step": 2168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792411214798754e-05,
+      "loss": 0.6498,
+      "step": 2169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979212947620527e-05,
+      "loss": 0.6094,
+      "step": 2170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791847548562187e-05,
+      "loss": 0.6439,
+      "step": 2171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979156543187494e-05,
+      "loss": 0.605,
+      "step": 2172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791283126148984e-05,
+      "loss": 0.6024,
+      "step": 2173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791000631389767e-05,
+      "loss": 0.6307,
+      "step": 2174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979071794760274e-05,
+      "loss": 0.6549,
+      "step": 2175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790435074793363e-05,
+      "loss": 0.6455,
+      "step": 2176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97901520129671e-05,
+      "loss": 0.6343,
+      "step": 2177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789868762129408e-05,
+      "loss": 0.6511,
+      "step": 2178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978958532228576e-05,
+      "loss": 0.6478,
+      "step": 2179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978930169344163e-05,
+      "loss": 0.6825,
+      "step": 2180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978901787560249e-05,
+      "loss": 0.6498,
+      "step": 2181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788733868773826e-05,
+      "loss": 0.6544,
+      "step": 2182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788449672961113e-05,
+      "loss": 0.6826,
+      "step": 2183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788165288169844e-05,
+      "loss": 0.5956,
+      "step": 2184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787880714405503e-05,
+      "loss": 0.6132,
+      "step": 2185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978759595167359e-05,
+      "loss": 0.5947,
+      "step": 2186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97873109999796e-05,
+      "loss": 0.6272,
+      "step": 2187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787025859329037e-05,
+      "loss": 0.6498,
+      "step": 2188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97867405297274e-05,
+      "loss": 0.6442,
+      "step": 2189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786455011180203e-05,
+      "loss": 0.6152,
+      "step": 2190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978616930369296e-05,
+      "loss": 0.6153,
+      "step": 2191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978588340727118e-05,
+      "loss": 0.625,
+      "step": 2192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978559732192039e-05,
+      "loss": 0.6198,
+      "step": 2193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785311047646107e-05,
+      "loss": 0.6504,
+      "step": 2194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978502458445386e-05,
+      "loss": 0.6287,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978473793234918e-05,
+      "loss": 0.6403,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784451091337602e-05,
+      "loss": 0.6339,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784164061424665e-05,
+      "loss": 0.6057,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783876842615907e-05,
+      "loss": 0.6436,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978358943491687e-05,
+      "loss": 0.6406,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783301838333116e-05,
+      "loss": 0.6624,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783014052870186e-05,
+      "loss": 0.617,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978272607853364e-05,
+      "loss": 0.6358,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978243791532903e-05,
+      "loss": 0.6629,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782149563261933e-05,
+      "loss": 0.5905,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781861022337903e-05,
+      "loss": 0.6377,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781572292562524e-05,
+      "loss": 0.6079,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781283373941356e-05,
+      "loss": 0.6035,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978099426647999e-05,
+      "loss": 0.5996,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780704970184e-05,
+      "loss": 0.6445,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978041548505897e-05,
+      "loss": 0.6339,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780125811110492e-05,
+      "loss": 0.6066,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977983594834416e-05,
+      "loss": 0.6558,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977954589676557e-05,
+      "loss": 0.6532,
+      "step": 2214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779255656380318e-05,
+      "loss": 0.6473,
+      "step": 2215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977896522719401e-05,
+      "loss": 0.6032,
+      "step": 2216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778674609212252e-05,
+      "loss": 0.641,
+      "step": 2217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778383802440654e-05,
+      "loss": 0.6051,
+      "step": 2218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778092806884833e-05,
+      "loss": 0.6684,
+      "step": 2219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977780162255041e-05,
+      "loss": 0.6394,
+      "step": 2220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777510249443e-05,
+      "loss": 0.5985,
+      "step": 2221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777218687568228e-05,
+      "loss": 0.6651,
+      "step": 2222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977692693693173e-05,
+      "loss": 0.6322,
+      "step": 2223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977663499753913e-05,
+      "loss": 0.6896,
+      "step": 2224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776342869396075e-05,
+      "loss": 0.6172,
+      "step": 2225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776050552508192e-05,
+      "loss": 0.6452,
+      "step": 2226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775758046881135e-05,
+      "loss": 0.6842,
+      "step": 2227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977546535252055e-05,
+      "loss": 0.6068,
+      "step": 2228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977517246943208e-05,
+      "loss": 0.663,
+      "step": 2229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774879397621387e-05,
+      "loss": 0.653,
+      "step": 2230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774586137094126e-05,
+      "loss": 0.645,
+      "step": 2231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774292687855964e-05,
+      "loss": 0.6479,
+      "step": 2232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773999049912556e-05,
+      "loss": 0.6157,
+      "step": 2233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773705223269578e-05,
+      "loss": 0.6565,
+      "step": 2234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773411207932707e-05,
+      "loss": 0.6578,
+      "step": 2235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977311700390761e-05,
+      "loss": 0.6036,
+      "step": 2236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772822611199972e-05,
+      "loss": 0.6408,
+      "step": 2237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772528029815476e-05,
+      "loss": 0.6414,
+      "step": 2238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977223325975981e-05,
+      "loss": 0.6442,
+      "step": 2239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771938301038664e-05,
+      "loss": 0.6437,
+      "step": 2240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977164315365773e-05,
+      "loss": 0.6807,
+      "step": 2241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977134781762271e-05,
+      "loss": 0.6355,
+      "step": 2242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771052292939304e-05,
+      "loss": 0.6744,
+      "step": 2243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977075657961322e-05,
+      "loss": 0.6469,
+      "step": 2244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770460677650167e-05,
+      "loss": 0.6049,
+      "step": 2245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977016458705585e-05,
+      "loss": 0.6692,
+      "step": 2246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769868307835996e-05,
+      "loss": 0.6113,
+      "step": 2247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769571839996318e-05,
+      "loss": 0.62,
+      "step": 2248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976927518354254e-05,
+      "loss": 0.3477,
+      "step": 2249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768978338480396e-05,
+      "loss": 0.6309,
+      "step": 2250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976868130481561e-05,
+      "loss": 0.6145,
+      "step": 2251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768384082553923e-05,
+      "loss": 0.6408,
+      "step": 2252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768086671701064e-05,
+      "loss": 0.6017,
+      "step": 2253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767789072262784e-05,
+      "loss": 0.6521,
+      "step": 2254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976749128424482e-05,
+      "loss": 0.6397,
+      "step": 2255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976719330765293e-05,
+      "loss": 0.642,
+      "step": 2256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766895142492865e-05,
+      "loss": 0.6097,
+      "step": 2257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766596788770373e-05,
+      "loss": 0.6606,
+      "step": 2258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766298246491225e-05,
+      "loss": 0.6449,
+      "step": 2259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765999515661177e-05,
+      "loss": 0.6354,
+      "step": 2260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765700596286e-05,
+      "loss": 0.6459,
+      "step": 2261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765401488371464e-05,
+      "loss": 0.6261,
+      "step": 2262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765102191923345e-05,
+      "loss": 0.6595,
+      "step": 2263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764802706947423e-05,
+      "loss": 0.6032,
+      "step": 2264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764503033449475e-05,
+      "loss": 0.6066,
+      "step": 2265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764203171435285e-05,
+      "loss": 0.6031,
+      "step": 2266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976390312091065e-05,
+      "loss": 0.6487,
+      "step": 2267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763602881881357e-05,
+      "loss": 0.6039,
+      "step": 2268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976330245435321e-05,
+      "loss": 0.6324,
+      "step": 2269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763001838331998e-05,
+      "loss": 0.6423,
+      "step": 2270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762701033823536e-05,
+      "loss": 0.6249,
+      "step": 2271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976240004083362e-05,
+      "loss": 0.6656,
+      "step": 2272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762098859368067e-05,
+      "loss": 0.6451,
+      "step": 2273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761797489432696e-05,
+      "loss": 0.638,
+      "step": 2274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976149593103332e-05,
+      "loss": 0.6442,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761194184175758e-05,
+      "loss": 0.6564,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760892248865843e-05,
+      "loss": 0.6143,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760590125109397e-05,
+      "loss": 0.6619,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976028781291226e-05,
+      "loss": 0.6467,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759985312280264e-05,
+      "loss": 0.6132,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975968262321925e-05,
+      "loss": 0.6365,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759379745735066e-05,
+      "loss": 0.6169,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975907667983355e-05,
+      "loss": 0.6099,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975877342552056e-05,
+      "loss": 0.6328,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975846998280195e-05,
+      "loss": 0.6199,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975816635168358e-05,
+      "loss": 0.6738,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757862532171308e-05,
+      "loss": 0.6512,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757558524271003e-05,
+      "loss": 0.6246,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757254327988527e-05,
+      "loss": 0.6293,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756949943329763e-05,
+      "loss": 0.6585,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975664537030058e-05,
+      "loss": 0.662,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756340608906865e-05,
+      "loss": 0.6526,
+      "step": 2292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756035659154495e-05,
+      "loss": 0.6498,
+      "step": 2293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975573052104936e-05,
+      "loss": 0.6541,
+      "step": 2294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975542519459735e-05,
+      "loss": 0.6324,
+      "step": 2295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975511967980437e-05,
+      "loss": 0.6185,
+      "step": 2296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97548139766763e-05,
+      "loss": 0.6739,
+      "step": 2297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754508085219057e-05,
+      "loss": 0.6371,
+      "step": 2298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754202005438538e-05,
+      "loss": 0.6795,
+      "step": 2299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975389573734066e-05,
+      "loss": 0.6509,
+      "step": 2300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753589280931328e-05,
+      "loss": 0.5875,
+      "step": 2301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753282636216462e-05,
+      "loss": 0.6519,
+      "step": 2302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975297580320198e-05,
+      "loss": 0.6046,
+      "step": 2303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975266878189381e-05,
+      "loss": 0.6438,
+      "step": 2304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975236157229788e-05,
+      "loss": 0.6059,
+      "step": 2305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975205417442011e-05,
+      "loss": 0.6387,
+      "step": 2306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751746588266454e-05,
+      "loss": 0.6139,
+      "step": 2307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975143881384283e-05,
+      "loss": 0.6237,
+      "step": 2308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751130851155196e-05,
+      "loss": 0.6646,
+      "step": 2309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975082270020949e-05,
+      "loss": 0.64,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750514361011658e-05,
+      "loss": 0.6299,
+      "step": 2311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975020583356766e-05,
+      "loss": 0.6342,
+      "step": 2312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749897117883453e-05,
+      "loss": 0.6271,
+      "step": 2313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974958821396499e-05,
+      "loss": 0.6507,
+      "step": 2314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.6341,
+      "step": 2315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748969841449166e-05,
+      "loss": 0.6161,
+      "step": 2316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748660372863744e-05,
+      "loss": 0.6579,
+      "step": 2317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748350716067944e-05,
+      "loss": 0.6516,
+      "step": 2318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748040871067752e-05,
+      "loss": 0.6775,
+      "step": 2319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747730837869145e-05,
+      "loss": 0.6292,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747420616478105e-05,
+      "loss": 0.63,
+      "step": 2321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747110206900628e-05,
+      "loss": 0.6296,
+      "step": 2322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746799609142703e-05,
+      "loss": 0.672,
+      "step": 2323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746488823210327e-05,
+      "loss": 0.6319,
+      "step": 2324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746177849109497e-05,
+      "loss": 0.6604,
+      "step": 2325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745866686846224e-05,
+      "loss": 0.6059,
+      "step": 2326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974555533642651e-05,
+      "loss": 0.6406,
+      "step": 2327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745243797856365e-05,
+      "loss": 0.6323,
+      "step": 2328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974493207114181e-05,
+      "loss": 0.6548,
+      "step": 2329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974462015628886e-05,
+      "loss": 0.6103,
+      "step": 2330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744308053303535e-05,
+      "loss": 0.6223,
+      "step": 2331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974399576219186e-05,
+      "loss": 0.6639,
+      "step": 2332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743683282959866e-05,
+      "loss": 0.6381,
+      "step": 2333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974337061561359e-05,
+      "loss": 0.6523,
+      "step": 2334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743057760159063e-05,
+      "loss": 0.6092,
+      "step": 2335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974274471660232e-05,
+      "loss": 0.6366,
+      "step": 2336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974243148494942e-05,
+      "loss": 0.626,
+      "step": 2337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742118065206397e-05,
+      "loss": 0.6207,
+      "step": 2338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741804457379308e-05,
+      "loss": 0.6525,
+      "step": 2339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741490661474202e-05,
+      "loss": 0.6261,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741176677497146e-05,
+      "loss": 0.63,
+      "step": 2341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740862505454194e-05,
+      "loss": 0.6598,
+      "step": 2342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740548145351416e-05,
+      "loss": 0.5884,
+      "step": 2343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974023359719488e-05,
+      "loss": 0.6399,
+      "step": 2344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739918860990657e-05,
+      "loss": 0.6781,
+      "step": 2345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973960393674483e-05,
+      "loss": 0.6227,
+      "step": 2346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739288824463466e-05,
+      "loss": 0.6439,
+      "step": 2347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973897352415266e-05,
+      "loss": 0.6487,
+      "step": 2348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738658035818495e-05,
+      "loss": 0.638,
+      "step": 2349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738342359467066e-05,
+      "loss": 0.6416,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973802649510446e-05,
+      "loss": 0.6325,
+      "step": 2351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737710442736783e-05,
+      "loss": 0.6093,
+      "step": 2352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737394202370132e-05,
+      "loss": 0.639,
+      "step": 2353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737077774010615e-05,
+      "loss": 0.6198,
+      "step": 2354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736761157664338e-05,
+      "loss": 0.6343,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736444353337415e-05,
+      "loss": 0.6458,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736127361035962e-05,
+      "loss": 0.6472,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97358101807661e-05,
+      "loss": 0.6343,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735492812533952e-05,
+      "loss": 0.6042,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735175256345647e-05,
+      "loss": 0.6465,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734857512207314e-05,
+      "loss": 0.6587,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734539580125085e-05,
+      "loss": 0.6122,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734221460105104e-05,
+      "loss": 0.6469,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733903152153507e-05,
+      "loss": 0.6689,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733584656276445e-05,
+      "loss": 0.6151,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973326597248006e-05,
+      "loss": 0.6413,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973294710077051e-05,
+      "loss": 0.6348,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973262804115395e-05,
+      "loss": 0.6535,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973230879363654e-05,
+      "loss": 0.6735,
+      "step": 2369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973198935822444e-05,
+      "loss": 0.6194,
+      "step": 2370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731669734923825e-05,
+      "loss": 0.6331,
+      "step": 2371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731349923740857e-05,
+      "loss": 0.6133,
+      "step": 2372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731029924681712e-05,
+      "loss": 0.6427,
+      "step": 2373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730709737752577e-05,
+      "loss": 0.6935,
+      "step": 2374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973038936295962e-05,
+      "loss": 0.5893,
+      "step": 2375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730068800309036e-05,
+      "loss": 0.6816,
+      "step": 2376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972974804980701e-05,
+      "loss": 0.6415,
+      "step": 2377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972942711145974e-05,
+      "loss": 0.6617,
+      "step": 2378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729105985273414e-05,
+      "loss": 0.6209,
+      "step": 2379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728784671254237e-05,
+      "loss": 0.6547,
+      "step": 2380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728463169408403e-05,
+      "loss": 0.6483,
+      "step": 2381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728141479742135e-05,
+      "loss": 0.6141,
+      "step": 2382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972781960226163e-05,
+      "loss": 0.6222,
+      "step": 2383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727497536973114e-05,
+      "loss": 0.6197,
+      "step": 2384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727175283882792e-05,
+      "loss": 0.5892,
+      "step": 2385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726852842996894e-05,
+      "loss": 0.6091,
+      "step": 2386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726530214321643e-05,
+      "loss": 0.6899,
+      "step": 2387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972620739786327e-05,
+      "loss": 0.6057,
+      "step": 2388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725884393628005e-05,
+      "loss": 0.6162,
+      "step": 2389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725561201622085e-05,
+      "loss": 0.6461,
+      "step": 2390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725237821851745e-05,
+      "loss": 0.6638,
+      "step": 2391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724914254323235e-05,
+      "loss": 0.6382,
+      "step": 2392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724590499042797e-05,
+      "loss": 0.6681,
+      "step": 2393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724266556016683e-05,
+      "loss": 0.6367,
+      "step": 2394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723942425251147e-05,
+      "loss": 0.6623,
+      "step": 2395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972361810675245e-05,
+      "loss": 0.6417,
+      "step": 2396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723293600526853e-05,
+      "loss": 0.6351,
+      "step": 2397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722968906580614e-05,
+      "loss": 0.6431,
+      "step": 2398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722644024920008e-05,
+      "loss": 0.6221,
+      "step": 2399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722318955551307e-05,
+      "loss": 0.6059,
+      "step": 2400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721993698480783e-05,
+      "loss": 0.6467,
+      "step": 2401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721668253714723e-05,
+      "loss": 0.6339,
+      "step": 2402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97213426212594e-05,
+      "loss": 0.6138,
+      "step": 2403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721016801121105e-05,
+      "loss": 0.6099,
+      "step": 2404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720690793306137e-05,
+      "loss": 0.6613,
+      "step": 2405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720364597820777e-05,
+      "loss": 0.6487,
+      "step": 2406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720038214671327e-05,
+      "loss": 0.6451,
+      "step": 2407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971971164386409e-05,
+      "loss": 0.6167,
+      "step": 2408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719384885405372e-05,
+      "loss": 0.6573,
+      "step": 2409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719057939301477e-05,
+      "loss": 0.6473,
+      "step": 2410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718730805558725e-05,
+      "loss": 0.6266,
+      "step": 2411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971840348418342e-05,
+      "loss": 0.631,
+      "step": 2412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971807597518189e-05,
+      "loss": 0.6567,
+      "step": 2413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717748278560456e-05,
+      "loss": 0.6189,
+      "step": 2414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717420394325445e-05,
+      "loss": 0.6383,
+      "step": 2415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717092322483183e-05,
+      "loss": 0.6097,
+      "step": 2416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971676406304001e-05,
+      "loss": 0.6232,
+      "step": 2417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716435616002264e-05,
+      "loss": 0.6267,
+      "step": 2418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716106981376278e-05,
+      "loss": 0.6178,
+      "step": 2419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715778159168402e-05,
+      "loss": 0.6119,
+      "step": 2420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715449149384985e-05,
+      "loss": 0.6681,
+      "step": 2421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715119952032384e-05,
+      "loss": 0.6312,
+      "step": 2422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714790567116943e-05,
+      "loss": 0.6114,
+      "step": 2423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714460994645023e-05,
+      "loss": 0.6204,
+      "step": 2424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714131234622996e-05,
+      "loss": 0.6301,
+      "step": 2425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713801287057223e-05,
+      "loss": 0.5969,
+      "step": 2426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971347115195407e-05,
+      "loss": 0.6536,
+      "step": 2427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713140829319915e-05,
+      "loss": 0.5959,
+      "step": 2428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971281031916114e-05,
+      "loss": 0.6156,
+      "step": 2429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712479621484117e-05,
+      "loss": 0.5987,
+      "step": 2430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971214873629524e-05,
+      "loss": 0.676,
+      "step": 2431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711817663600884e-05,
+      "loss": 0.6363,
+      "step": 2432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711486403407455e-05,
+      "loss": 0.6212,
+      "step": 2433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711154955721338e-05,
+      "loss": 0.6356,
+      "step": 2434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710823320548937e-05,
+      "loss": 0.6636,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971049149789665e-05,
+      "loss": 0.6213,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971015948777089e-05,
+      "loss": 0.5779,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709827290178064e-05,
+      "loss": 0.6574,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709494905124586e-05,
+      "loss": 0.6774,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709162332616872e-05,
+      "loss": 0.5923,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970882957266134e-05,
+      "loss": 0.6194,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708496625264423e-05,
+      "loss": 0.6216,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708163490432538e-05,
+      "loss": 0.604,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707830168172124e-05,
+      "loss": 0.6555,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707496658489613e-05,
+      "loss": 0.7125,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707162961391448e-05,
+      "loss": 0.6116,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706829076884064e-05,
+      "loss": 0.3366,
+      "step": 2447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706495004973914e-05,
+      "loss": 0.6261,
+      "step": 2448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706160745667446e-05,
+      "loss": 0.6432,
+      "step": 2449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705826298971112e-05,
+      "loss": 0.6776,
+      "step": 2450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705491664891368e-05,
+      "loss": 0.595,
+      "step": 2451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970515684343468e-05,
+      "loss": 0.6005,
+      "step": 2452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704821834607505e-05,
+      "loss": 0.3261,
+      "step": 2453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704486638416314e-05,
+      "loss": 0.6382,
+      "step": 2454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704151254867575e-05,
+      "loss": 0.6517,
+      "step": 2455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703815683967772e-05,
+      "loss": 0.5906,
+      "step": 2456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703479925723377e-05,
+      "loss": 0.6445,
+      "step": 2457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970314398014087e-05,
+      "loss": 0.6006,
+      "step": 2458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970280784722674e-05,
+      "loss": 0.6615,
+      "step": 2459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970247152698748e-05,
+      "loss": 0.617,
+      "step": 2460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970213501942958e-05,
+      "loss": 0.6339,
+      "step": 2461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970179832455953e-05,
+      "loss": 0.6414,
+      "step": 2462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701461442383842e-05,
+      "loss": 0.6612,
+      "step": 2463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701124372909012e-05,
+      "loss": 0.6406,
+      "step": 2464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700787116141553e-05,
+      "loss": 0.6408,
+      "step": 2465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700449672087975e-05,
+      "loss": 0.6146,
+      "step": 2466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700112040754784e-05,
+      "loss": 0.6215,
+      "step": 2467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969977422214851e-05,
+      "loss": 0.6077,
+      "step": 2468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699436216275672e-05,
+      "loss": 0.6404,
+      "step": 2469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969909802314279e-05,
+      "loss": 0.6313,
+      "step": 2470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96987596427564e-05,
+      "loss": 0.6422,
+      "step": 2471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698421075123032e-05,
+      "loss": 0.6685,
+      "step": 2472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698082320249223e-05,
+      "loss": 0.6372,
+      "step": 2473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697743378141512e-05,
+      "loss": 0.6244,
+      "step": 2474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697404248806445e-05,
+      "loss": 0.6072,
+      "step": 2475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697064932250563e-05,
+      "loss": 0.6237,
+      "step": 2476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696725428480428e-05,
+      "loss": 0.638,
+      "step": 2477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696385737502587e-05,
+      "loss": 0.633,
+      "step": 2478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696045859323597e-05,
+      "loss": 0.6284,
+      "step": 2479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695705793950025e-05,
+      "loss": 0.607,
+      "step": 2480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695365541388432e-05,
+      "loss": 0.6185,
+      "step": 2481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695025101645392e-05,
+      "loss": 0.6608,
+      "step": 2482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969468447472747e-05,
+      "loss": 0.6083,
+      "step": 2483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969434366064125e-05,
+      "loss": 0.6095,
+      "step": 2484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694002659393306e-05,
+      "loss": 0.6542,
+      "step": 2485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693661470990225e-05,
+      "loss": 0.6106,
+      "step": 2486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969332009543859e-05,
+      "loss": 0.5985,
+      "step": 2487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692978532745e-05,
+      "loss": 0.601,
+      "step": 2488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969263678291604e-05,
+      "loss": 0.6431,
+      "step": 2489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692294845958315e-05,
+      "loss": 0.6243,
+      "step": 2490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969195272187842e-05,
+      "loss": 0.6086,
+      "step": 2491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691610410682966e-05,
+      "loss": 0.6355,
+      "step": 2492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969126791237856e-05,
+      "loss": 0.6231,
+      "step": 2493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690925226971812e-05,
+      "loss": 0.639,
+      "step": 2494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969058235446934e-05,
+      "loss": 0.6704,
+      "step": 2495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690239294877766e-05,
+      "loss": 0.6265,
+      "step": 2496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689896048203708e-05,
+      "loss": 0.6101,
+      "step": 2497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689552614453794e-05,
+      "loss": 0.6542,
+      "step": 2498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689208993634658e-05,
+      "loss": 0.6708,
+      "step": 2499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968886518575293e-05,
+      "loss": 0.6241,
+      "step": 2500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688521190815253e-05,
+      "loss": 0.6855,
+      "step": 2501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968817700882826e-05,
+      "loss": 0.6072,
+      "step": 2502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687832639798604e-05,
+      "loss": 0.644,
+      "step": 2503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968748808373293e-05,
+      "loss": 0.5961,
+      "step": 2504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687143340637885e-05,
+      "loss": 0.6259,
+      "step": 2505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686798410520136e-05,
+      "loss": 0.6439,
+      "step": 2506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686453293386334e-05,
+      "loss": 0.6196,
+      "step": 2507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686107989243145e-05,
+      "loss": 0.638,
+      "step": 2508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685762498097233e-05,
+      "loss": 0.661,
+      "step": 2509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968541681995527e-05,
+      "loss": 0.6235,
+      "step": 2510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968507095482393e-05,
+      "loss": 0.6109,
+      "step": 2511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968472490270989e-05,
+      "loss": 0.6365,
+      "step": 2512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684378663619832e-05,
+      "loss": 0.6054,
+      "step": 2513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968403223756044e-05,
+      "loss": 0.6219,
+      "step": 2514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96836856245384e-05,
+      "loss": 0.6121,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683338824560403e-05,
+      "loss": 0.6405,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682991837633148e-05,
+      "loss": 0.6228,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682644663763336e-05,
+      "loss": 0.6222,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682297302957666e-05,
+      "loss": 0.6655,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681949755222846e-05,
+      "loss": 0.6193,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968160202056558e-05,
+      "loss": 0.5876,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968125409899259e-05,
+      "loss": 0.6602,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680905990510588e-05,
+      "loss": 0.6731,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680557695126294e-05,
+      "loss": 0.6348,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9680209212846437e-05,
+      "loss": 0.6392,
+      "step": 2525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967986054367774e-05,
+      "loss": 0.6475,
+      "step": 2526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967951168762694e-05,
+      "loss": 0.6228,
+      "step": 2527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679162644700766e-05,
+      "loss": 0.6102,
+      "step": 2528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967881341490596e-05,
+      "loss": 0.6298,
+      "step": 2529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678463998249264e-05,
+      "loss": 0.6235,
+      "step": 2530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678114394737424e-05,
+      "loss": 0.6166,
+      "step": 2531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677764604377186e-05,
+      "loss": 0.6229,
+      "step": 2532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967741462717531e-05,
+      "loss": 0.6377,
+      "step": 2533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967706446313855e-05,
+      "loss": 0.5891,
+      "step": 2534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676714112273664e-05,
+      "loss": 0.6121,
+      "step": 2535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676363574587414e-05,
+      "loss": 0.6243,
+      "step": 2536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676012850086574e-05,
+      "loss": 0.3215,
+      "step": 2537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675661938777914e-05,
+      "loss": 0.6379,
+      "step": 2538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675310840668205e-05,
+      "loss": 0.6366,
+      "step": 2539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674959555764227e-05,
+      "loss": 0.6144,
+      "step": 2540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674608084072765e-05,
+      "loss": 0.6304,
+      "step": 2541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96742564256006e-05,
+      "loss": 0.6182,
+      "step": 2542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673904580354523e-05,
+      "loss": 0.6552,
+      "step": 2543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673552548341327e-05,
+      "loss": 0.6276,
+      "step": 2544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967320032956781e-05,
+      "loss": 0.6437,
+      "step": 2545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967284792404077e-05,
+      "loss": 0.6416,
+      "step": 2546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967249533176701e-05,
+      "loss": 0.583,
+      "step": 2547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967214255275334e-05,
+      "loss": 0.6606,
+      "step": 2548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967178958700657e-05,
+      "loss": 0.618,
+      "step": 2549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671436434533514e-05,
+      "loss": 0.6133,
+      "step": 2550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671083095340987e-05,
+      "loss": 0.654,
+      "step": 2551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670729569435812e-05,
+      "loss": 0.6548,
+      "step": 2552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670375856824823e-05,
+      "loss": 0.63,
+      "step": 2553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670021957514836e-05,
+      "loss": 0.6393,
+      "step": 2554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966966787151269e-05,
+      "loss": 0.6262,
+      "step": 2555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966931359882522e-05,
+      "loss": 0.6485,
+      "step": 2556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668959139459264e-05,
+      "loss": 0.628,
+      "step": 2557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668604493421673e-05,
+      "loss": 0.6635,
+      "step": 2558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668249660719282e-05,
+      "loss": 0.6375,
+      "step": 2559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966789464135895e-05,
+      "loss": 0.6078,
+      "step": 2560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966753943534753e-05,
+      "loss": 0.6435,
+      "step": 2561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667184042691877e-05,
+      "loss": 0.6109,
+      "step": 2562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666828463398853e-05,
+      "loss": 0.6167,
+      "step": 2563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666472697475324e-05,
+      "loss": 0.6324,
+      "step": 2564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966611674492816e-05,
+      "loss": 0.6676,
+      "step": 2565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966576060576423e-05,
+      "loss": 0.6715,
+      "step": 2566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966540427999041e-05,
+      "loss": 0.6285,
+      "step": 2567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966504776761358e-05,
+      "loss": 0.6214,
+      "step": 2568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664691068640623e-05,
+      "loss": 0.5997,
+      "step": 2569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966433418307843e-05,
+      "loss": 0.6142,
+      "step": 2570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966397711093388e-05,
+      "loss": 0.6167,
+      "step": 2571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663619852213876e-05,
+      "loss": 0.6332,
+      "step": 2572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966326240692532e-05,
+      "loss": 0.6424,
+      "step": 2573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662904775075098e-05,
+      "loss": 0.6145,
+      "step": 2574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662546956670123e-05,
+      "loss": 0.6428,
+      "step": 2575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662188951717305e-05,
+      "loss": 0.6521,
+      "step": 2576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661830760223553e-05,
+      "loss": 0.641,
+      "step": 2577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661472382195782e-05,
+      "loss": 0.6456,
+      "step": 2578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966111381764091e-05,
+      "loss": 0.615,
+      "step": 2579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660755066565863e-05,
+      "loss": 0.5915,
+      "step": 2580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660396128977566e-05,
+      "loss": 0.636,
+      "step": 2581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660037004882948e-05,
+      "loss": 0.5782,
+      "step": 2582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965967769428894e-05,
+      "loss": 0.6278,
+      "step": 2583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659318197202483e-05,
+      "loss": 0.6308,
+      "step": 2584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658958513630518e-05,
+      "loss": 0.6235,
+      "step": 2585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658598643579983e-05,
+      "loss": 0.6713,
+      "step": 2586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658238587057832e-05,
+      "loss": 0.6125,
+      "step": 2587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657878344071014e-05,
+      "loss": 0.6494,
+      "step": 2588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657517914626484e-05,
+      "loss": 0.623,
+      "step": 2589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96571572987312e-05,
+      "loss": 0.6382,
+      "step": 2590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656796496392126e-05,
+      "loss": 0.6072,
+      "step": 2591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656435507616226e-05,
+      "loss": 0.6248,
+      "step": 2592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656074332410472e-05,
+      "loss": 0.6637,
+      "step": 2593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965571297078183e-05,
+      "loss": 0.596,
+      "step": 2594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655351422737285e-05,
+      "loss": 0.5855,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654989688283815e-05,
+      "loss": 0.6342,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654627767428398e-05,
+      "loss": 0.6322,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654265660178028e-05,
+      "loss": 0.633,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653903366539694e-05,
+      "loss": 0.6052,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653540886520387e-05,
+      "loss": 0.6505,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653178220127112e-05,
+      "loss": 0.6128,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652815367366867e-05,
+      "loss": 0.6298,
+      "step": 2602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652452328246654e-05,
+      "loss": 0.6069,
+      "step": 2603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652089102773487e-05,
+      "loss": 0.6456,
+      "step": 2604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965172569095438e-05,
+      "loss": 0.6181,
+      "step": 2605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965136209279634e-05,
+      "loss": 0.6362,
+      "step": 2606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96509983083064e-05,
+      "loss": 0.6242,
+      "step": 2607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965063433749157e-05,
+      "loss": 0.6302,
+      "step": 2608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650270180358887e-05,
+      "loss": 0.6885,
+      "step": 2609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649905836915373e-05,
+      "loss": 0.6376,
+      "step": 2610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964954130716807e-05,
+      "loss": 0.6197,
+      "step": 2611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649176591124012e-05,
+      "loss": 0.6071,
+      "step": 2612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964881168879024e-05,
+      "loss": 0.6062,
+      "step": 2613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96484466001738e-05,
+      "loss": 0.6497,
+      "step": 2614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648081325281737e-05,
+      "loss": 0.6078,
+      "step": 2615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647715864121108e-05,
+      "loss": 0.6604,
+      "step": 2616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647350216698967e-05,
+      "loss": 0.6651,
+      "step": 2617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646984383022373e-05,
+      "loss": 0.6184,
+      "step": 2618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646618363098388e-05,
+      "loss": 0.5771,
+      "step": 2619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964625215693408e-05,
+      "loss": 0.6634,
+      "step": 2620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645885764536522e-05,
+      "loss": 0.3179,
+      "step": 2621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964551918591278e-05,
+      "loss": 0.6397,
+      "step": 2622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645152421069936e-05,
+      "loss": 0.5965,
+      "step": 2623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644785470015073e-05,
+      "loss": 0.6487,
+      "step": 2624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964441833275527e-05,
+      "loss": 0.6646,
+      "step": 2625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964405100929762e-05,
+      "loss": 0.6584,
+      "step": 2626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964368349964921e-05,
+      "loss": 0.6501,
+      "step": 2627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964331580381714e-05,
+      "loss": 0.6293,
+      "step": 2628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642947921808507e-05,
+      "loss": 0.6454,
+      "step": 2629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964257985363041e-05,
+      "loss": 0.6153,
+      "step": 2630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642211599289962e-05,
+      "loss": 0.5982,
+      "step": 2631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641843158794266e-05,
+      "loss": 0.5935,
+      "step": 2632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964147453215044e-05,
+      "loss": 0.6395,
+      "step": 2633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641105719365595e-05,
+      "loss": 0.6471,
+      "step": 2634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640736720446857e-05,
+      "loss": 0.6462,
+      "step": 2635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964036753540135e-05,
+      "loss": 0.6364,
+      "step": 2636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639998164236198e-05,
+      "loss": 0.6352,
+      "step": 2637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639628606958535e-05,
+      "loss": 0.6134,
+      "step": 2638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639258863575496e-05,
+      "loss": 0.6239,
+      "step": 2639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638888934094212e-05,
+      "loss": 0.6381,
+      "step": 2640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638518818521833e-05,
+      "loss": 0.6364,
+      "step": 2641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638148516865507e-05,
+      "loss": 0.6654,
+      "step": 2642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637778029132377e-05,
+      "loss": 0.6363,
+      "step": 2643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637407355329595e-05,
+      "loss": 0.6269,
+      "step": 2644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963703649546432e-05,
+      "loss": 0.6425,
+      "step": 2645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636665449543715e-05,
+      "loss": 0.6226,
+      "step": 2646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636294217574938e-05,
+      "loss": 0.6241,
+      "step": 2647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635922799565156e-05,
+      "loss": 0.6467,
+      "step": 2648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963555119552155e-05,
+      "loss": 0.6063,
+      "step": 2649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635179405451277e-05,
+      "loss": 0.6324,
+      "step": 2650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963480742936153e-05,
+      "loss": 0.6569,
+      "step": 2651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634435267259483e-05,
+      "loss": 0.644,
+      "step": 2652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634062919152322e-05,
+      "loss": 0.6258,
+      "step": 2653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633690385047236e-05,
+      "loss": 0.6611,
+      "step": 2654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633317664951418e-05,
+      "loss": 0.6219,
+      "step": 2655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632944758872064e-05,
+      "loss": 0.6033,
+      "step": 2656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963257166681637e-05,
+      "loss": 0.6437,
+      "step": 2657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632198388791548e-05,
+      "loss": 0.6535,
+      "step": 2658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631824924804795e-05,
+      "loss": 0.607,
+      "step": 2659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631451274863326e-05,
+      "loss": 0.624,
+      "step": 2660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963107743897435e-05,
+      "loss": 0.6149,
+      "step": 2661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963070341714509e-05,
+      "loss": 0.6121,
+      "step": 2662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630329209382767e-05,
+      "loss": 0.6211,
+      "step": 2663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629954815694603e-05,
+      "loss": 0.6468,
+      "step": 2664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629580236087823e-05,
+      "loss": 0.5998,
+      "step": 2665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629205470569664e-05,
+      "loss": 0.6252,
+      "step": 2666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962883051914736e-05,
+      "loss": 0.6033,
+      "step": 2667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628455381828152e-05,
+      "loss": 0.6582,
+      "step": 2668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962808005861928e-05,
+      "loss": 0.5958,
+      "step": 2669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627704549527986e-05,
+      "loss": 0.6361,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627328854561526e-05,
+      "loss": 0.6549,
+      "step": 2671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962695297372715e-05,
+      "loss": 0.6329,
+      "step": 2672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626576907032117e-05,
+      "loss": 0.6416,
+      "step": 2673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626200654483688e-05,
+      "loss": 0.6324,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625824216089123e-05,
+      "loss": 0.6192,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625447591855695e-05,
+      "loss": 0.6185,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962507078179067e-05,
+      "loss": 0.6203,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624693785901325e-05,
+      "loss": 0.6241,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962431660419494e-05,
+      "loss": 0.6658,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9623939236678793e-05,
+      "loss": 0.6232,
+      "step": 2680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623561683360174e-05,
+      "loss": 0.6344,
+      "step": 2681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623183944246372e-05,
+      "loss": 0.648,
+      "step": 2682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622806019344675e-05,
+      "loss": 0.6433,
+      "step": 2683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962242790866238e-05,
+      "loss": 0.6254,
+      "step": 2684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622049612206795e-05,
+      "loss": 0.6515,
+      "step": 2685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621671129985215e-05,
+      "loss": 0.6318,
+      "step": 2686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962129246200495e-05,
+      "loss": 0.672,
+      "step": 2687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962091360827331e-05,
+      "loss": 0.5973,
+      "step": 2688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962053456879761e-05,
+      "loss": 0.6352,
+      "step": 2689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962015534358517e-05,
+      "loss": 0.6084,
+      "step": 2690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619775932643302e-05,
+      "loss": 0.6103,
+      "step": 2691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619396335979345e-05,
+      "loss": 0.6446,
+      "step": 2692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619016553600616e-05,
+      "loss": 0.628,
+      "step": 2693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618636585514456e-05,
+      "loss": 0.6482,
+      "step": 2694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961825643172819e-05,
+      "loss": 0.6392,
+      "step": 2695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961787609224917e-05,
+      "loss": 0.6349,
+      "step": 2696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961749556708473e-05,
+      "loss": 0.6115,
+      "step": 2697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617114856242222e-05,
+      "loss": 0.6406,
+      "step": 2698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616733959728992e-05,
+      "loss": 0.643,
+      "step": 2699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616352877552394e-05,
+      "loss": 0.6307,
+      "step": 2700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615971609719788e-05,
+      "loss": 0.6517,
+      "step": 2701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615590156238532e-05,
+      "loss": 0.6111,
+      "step": 2702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615208517115993e-05,
+      "loss": 0.6297,
+      "step": 2703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614826692359535e-05,
+      "loss": 0.6604,
+      "step": 2704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614444681976533e-05,
+      "loss": 0.6348,
+      "step": 2705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614062485974364e-05,
+      "loss": 0.6441,
+      "step": 2706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613680104360405e-05,
+      "loss": 0.6147,
+      "step": 2707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613297537142036e-05,
+      "loss": 0.6678,
+      "step": 2708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612914784326646e-05,
+      "loss": 0.6058,
+      "step": 2709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961253184592162e-05,
+      "loss": 0.5972,
+      "step": 2710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612148721934356e-05,
+      "loss": 0.6078,
+      "step": 2711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961176541237225e-05,
+      "loss": 0.5765,
+      "step": 2712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96113819172427e-05,
+      "loss": 0.614,
+      "step": 2713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961099823655311e-05,
+      "loss": 0.6307,
+      "step": 2714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610614370310892e-05,
+      "loss": 0.6313,
+      "step": 2715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610230318523453e-05,
+      "loss": 0.6114,
+      "step": 2716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609846081198207e-05,
+      "loss": 0.6165,
+      "step": 2717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609461658342575e-05,
+      "loss": 0.6374,
+      "step": 2718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609077049963975e-05,
+      "loss": 0.6518,
+      "step": 2719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608692256069836e-05,
+      "loss": 0.6232,
+      "step": 2720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608307276667587e-05,
+      "loss": 0.6538,
+      "step": 2721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607922111764655e-05,
+      "loss": 0.6254,
+      "step": 2722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607536761368484e-05,
+      "loss": 0.6034,
+      "step": 2723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607151225486507e-05,
+      "loss": 0.6422,
+      "step": 2724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606765504126174e-05,
+      "loss": 0.6385,
+      "step": 2725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606379597294924e-05,
+      "loss": 0.6412,
+      "step": 2726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605993505000212e-05,
+      "loss": 0.6035,
+      "step": 2727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605607227249494e-05,
+      "loss": 0.6251,
+      "step": 2728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605220764050222e-05,
+      "loss": 0.5923,
+      "step": 2729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960483411540986e-05,
+      "loss": 0.6444,
+      "step": 2730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604447281335878e-05,
+      "loss": 0.6409,
+      "step": 2731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960406026183573e-05,
+      "loss": 0.6451,
+      "step": 2732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603673056916907e-05,
+      "loss": 0.5842,
+      "step": 2733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603285666586867e-05,
+      "loss": 0.6307,
+      "step": 2734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96028980908531e-05,
+      "loss": 0.6082,
+      "step": 2735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602510329723087e-05,
+      "loss": 0.6311,
+      "step": 2736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960212238320431e-05,
+      "loss": 0.6693,
+      "step": 2737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601734251304262e-05,
+      "loss": 0.6379,
+      "step": 2738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601345934030435e-05,
+      "loss": 0.6154,
+      "step": 2739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960095743139033e-05,
+      "loss": 0.6466,
+      "step": 2740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600568743391438e-05,
+      "loss": 0.6074,
+      "step": 2741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600179870041278e-05,
+      "loss": 0.6417,
+      "step": 2742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599790811347342e-05,
+      "loss": 0.6568,
+      "step": 2743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959940156731715e-05,
+      "loss": 0.6175,
+      "step": 2744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599012137958214e-05,
+      "loss": 0.637,
+      "step": 2745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959862252327806e-05,
+      "loss": 0.6212,
+      "step": 2746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598232723284197e-05,
+      "loss": 0.6318,
+      "step": 2747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597842737984156e-05,
+      "loss": 0.6303,
+      "step": 2748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597452567385472e-05,
+      "loss": 0.6399,
+      "step": 2749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959706221149567e-05,
+      "loss": 0.6605,
+      "step": 2750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959667167032229e-05,
+      "loss": 0.6216,
+      "step": 2751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596280943872875e-05,
+      "loss": 0.652,
+      "step": 2752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595890032154962e-05,
+      "loss": 0.6398,
+      "step": 2753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.95954989351761e-05,
+      "loss": 0.6488,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595107652943842e-05,
+      "loss": 0.6243,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594716185465742e-05,
+      "loss": 0.6228,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594324532749353e-05,
+      "loss": 0.6512,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959393269480224e-05,
+      "loss": 0.6593,
+      "step": 2758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959354067163197e-05,
+      "loss": 0.6331,
+      "step": 2759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959314846324611e-05,
+      "loss": 0.5899,
+      "step": 2760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959275606965223e-05,
+      "loss": 0.6594,
+      "step": 2761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592363490857908e-05,
+      "loss": 0.6066,
+      "step": 2762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959197072687072e-05,
+      "loss": 0.6421,
+      "step": 2763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591577777698253e-05,
+      "loss": 0.6245,
+      "step": 2764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591184643348092e-05,
+      "loss": 0.6177,
+      "step": 2765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590791323827825e-05,
+      "loss": 0.6498,
+      "step": 2766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590397819145045e-05,
+      "loss": 0.6496,
+      "step": 2767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590004129307357e-05,
+      "loss": 0.6549,
+      "step": 2768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958961025432235e-05,
+      "loss": 0.6152,
+      "step": 2769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589216194197637e-05,
+      "loss": 0.6046,
+      "step": 2770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958882194894082e-05,
+      "loss": 0.594,
+      "step": 2771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588427518559517e-05,
+      "loss": 0.6285,
+      "step": 2772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588032903061334e-05,
+      "loss": 0.6108,
+      "step": 2773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95876381024539e-05,
+      "loss": 0.6,
+      "step": 2774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587243116744832e-05,
+      "loss": 0.5675,
+      "step": 2775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586847945941752e-05,
+      "loss": 0.6097,
+      "step": 2776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586452590052297e-05,
+      "loss": 0.6156,
+      "step": 2777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586057049084093e-05,
+      "loss": 0.6647,
+      "step": 2778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958566132304478e-05,
+      "loss": 0.6303,
+      "step": 2779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585265411941997e-05,
+      "loss": 0.6151,
+      "step": 2780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584869315783388e-05,
+      "loss": 0.5891,
+      "step": 2781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95844730345766e-05,
+      "loss": 0.6079,
+      "step": 2782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958407656832928e-05,
+      "loss": 0.6443,
+      "step": 2783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583679917049093e-05,
+      "loss": 0.6055,
+      "step": 2784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583283080743686e-05,
+      "loss": 0.6591,
+      "step": 2785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582886059420723e-05,
+      "loss": 0.605,
+      "step": 2786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582488853087872e-05,
+      "loss": 0.6305,
+      "step": 2787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95820914617528e-05,
+      "loss": 0.6163,
+      "step": 2788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581693885423177e-05,
+      "loss": 0.6518,
+      "step": 2789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581296124106682e-05,
+      "loss": 0.6472,
+      "step": 2790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958089817781099e-05,
+      "loss": 0.6426,
+      "step": 2791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580500046543793e-05,
+      "loss": 0.6498,
+      "step": 2792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580101730312768e-05,
+      "loss": 0.6275,
+      "step": 2793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579703229125607e-05,
+      "loss": 0.6349,
+      "step": 2794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579304542990005e-05,
+      "loss": 0.6319,
+      "step": 2795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578905671913663e-05,
+      "loss": 0.6381,
+      "step": 2796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578506615904275e-05,
+      "loss": 0.6715,
+      "step": 2797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578107374969547e-05,
+      "loss": 0.63,
+      "step": 2798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957770794911719e-05,
+      "loss": 0.6328,
+      "step": 2799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957730833835491e-05,
+      "loss": 0.5906,
+      "step": 2800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576908542690425e-05,
+      "loss": 0.6422,
+      "step": 2801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576508562131457e-05,
+      "loss": 0.6132,
+      "step": 2802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576108396685725e-05,
+      "loss": 0.6133,
+      "step": 2803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575708046360954e-05,
+      "loss": 0.6154,
+      "step": 2804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575307511164873e-05,
+      "loss": 0.5973,
+      "step": 2805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574906791105216e-05,
+      "loss": 0.6446,
+      "step": 2806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957450588618972e-05,
+      "loss": 0.6342,
+      "step": 2807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574104796426124e-05,
+      "loss": 0.6121,
+      "step": 2808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957370352182217e-05,
+      "loss": 0.6037,
+      "step": 2809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957330206238561e-05,
+      "loss": 0.5706,
+      "step": 2810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957290041812419e-05,
+      "loss": 0.6138,
+      "step": 2811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572498589045662e-05,
+      "loss": 0.6251,
+      "step": 2812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572096575157795e-05,
+      "loss": 0.6243,
+      "step": 2813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571694376468338e-05,
+      "loss": 0.6404,
+      "step": 2814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571291992985064e-05,
+      "loss": 0.6389,
+      "step": 2815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570889424715736e-05,
+      "loss": 0.6078,
+      "step": 2816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570486671668128e-05,
+      "loss": 0.5681,
+      "step": 2817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570083733850017e-05,
+      "loss": 0.6525,
+      "step": 2818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569680611269185e-05,
+      "loss": 0.6314,
+      "step": 2819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956927730393341e-05,
+      "loss": 0.6638,
+      "step": 2820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568873811850476e-05,
+      "loss": 0.6534,
+      "step": 2821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956847013502818e-05,
+      "loss": 0.6433,
+      "step": 2822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956806627347431e-05,
+      "loss": 0.6243,
+      "step": 2823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567662227196668e-05,
+      "loss": 0.6291,
+      "step": 2824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956725799620305e-05,
+      "loss": 0.6216,
+      "step": 2825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956685358050126e-05,
+      "loss": 0.6381,
+      "step": 2826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956644898009911e-05,
+      "loss": 0.6071,
+      "step": 2827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956604419500441e-05,
+      "loss": 0.5969,
+      "step": 2828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565639225224974e-05,
+      "loss": 0.6261,
+      "step": 2829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565234070768617e-05,
+      "loss": 0.6523,
+      "step": 2830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956482873164317e-05,
+      "loss": 0.6419,
+      "step": 2831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956442320785645e-05,
+      "loss": 0.6654,
+      "step": 2832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564017499416285e-05,
+      "loss": 0.6158,
+      "step": 2833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563611606330517e-05,
+      "loss": 0.6219,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563205528606977e-05,
+      "loss": 0.6264,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9562799266253503e-05,
+      "loss": 0.6208,
+      "step": 2836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562392819277943e-05,
+      "loss": 0.6607,
+      "step": 2837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956198618768814e-05,
+      "loss": 0.629,
+      "step": 2838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561579371491946e-05,
+      "loss": 0.6265,
+      "step": 2839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561172370697215e-05,
+      "loss": 0.6297,
+      "step": 2840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560765185311807e-05,
+      "loss": 0.6112,
+      "step": 2841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560357815343577e-05,
+      "loss": 0.646,
+      "step": 2842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559950260800396e-05,
+      "loss": 0.6465,
+      "step": 2843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559542521690133e-05,
+      "loss": 0.6217,
+      "step": 2844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559134598020653e-05,
+      "loss": 0.5822,
+      "step": 2845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558726489799835e-05,
+      "loss": 0.6459,
+      "step": 2846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558318197035562e-05,
+      "loss": 0.6339,
+      "step": 2847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557909719735712e-05,
+      "loss": 0.6356,
+      "step": 2848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557501057908173e-05,
+      "loss": 0.609,
+      "step": 2849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955709221156083e-05,
+      "loss": 0.6219,
+      "step": 2850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556683180701586e-05,
+      "loss": 0.5802,
+      "step": 2851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556273965338334e-05,
+      "loss": 0.633,
+      "step": 2852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555864565478967e-05,
+      "loss": 0.6335,
+      "step": 2853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555454981131395e-05,
+      "loss": 0.6295,
+      "step": 2854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555045212303528e-05,
+      "loss": 0.6545,
+      "step": 2855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554635259003275e-05,
+      "loss": 0.6346,
+      "step": 2856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554225121238548e-05,
+      "loss": 0.6202,
+      "step": 2857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955381479901727e-05,
+      "loss": 0.658,
+      "step": 2858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553404292347356e-05,
+      "loss": 0.6177,
+      "step": 2859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552993601236737e-05,
+      "loss": 0.5801,
+      "step": 2860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955258272569334e-05,
+      "loss": 0.634,
+      "step": 2861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95521716657251e-05,
+      "loss": 0.648,
+      "step": 2862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955176042133995e-05,
+      "loss": 0.6508,
+      "step": 2863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955134899254583e-05,
+      "loss": 0.6299,
+      "step": 2864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550937379350686e-05,
+      "loss": 0.6095,
+      "step": 2865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955052558176246e-05,
+      "loss": 0.6605,
+      "step": 2866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550113599789105e-05,
+      "loss": 0.5913,
+      "step": 2867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549701433438577e-05,
+      "loss": 0.6308,
+      "step": 2868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954928908271883e-05,
+      "loss": 0.6206,
+      "step": 2869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954887654763782e-05,
+      "loss": 0.5837,
+      "step": 2870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548463828203523e-05,
+      "loss": 0.6054,
+      "step": 2871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95480509244239e-05,
+      "loss": 0.6149,
+      "step": 2872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954763783630693e-05,
+      "loss": 0.6396,
+      "step": 2873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547224563860574e-05,
+      "loss": 0.6001,
+      "step": 2874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546811107092822e-05,
+      "loss": 0.6115,
+      "step": 2875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546397466011654e-05,
+      "loss": 0.6163,
+      "step": 2876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545983640625054e-05,
+      "loss": 0.6998,
+      "step": 2877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545569630941012e-05,
+      "loss": 0.6353,
+      "step": 2878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545155436967524e-05,
+      "loss": 0.5946,
+      "step": 2879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544741058712585e-05,
+      "loss": 0.5818,
+      "step": 2880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544326496184192e-05,
+      "loss": 0.6121,
+      "step": 2881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954391174939035e-05,
+      "loss": 0.6709,
+      "step": 2882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954349681833907e-05,
+      "loss": 0.6075,
+      "step": 2883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543081703038357e-05,
+      "loss": 0.6282,
+      "step": 2884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542666403496232e-05,
+      "loss": 0.5822,
+      "step": 2885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542250919720708e-05,
+      "loss": 0.6375,
+      "step": 2886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541835251719807e-05,
+      "loss": 0.6571,
+      "step": 2887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541419399501552e-05,
+      "loss": 0.6676,
+      "step": 2888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541003363073978e-05,
+      "loss": 0.6443,
+      "step": 2889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540587142445112e-05,
+      "loss": 0.6173,
+      "step": 2890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954017073762299e-05,
+      "loss": 0.6382,
+      "step": 2891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539754148615654e-05,
+      "loss": 0.6368,
+      "step": 2892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539337375431144e-05,
+      "loss": 0.6234,
+      "step": 2893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953892041807751e-05,
+      "loss": 0.6027,
+      "step": 2894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538503276562797e-05,
+      "loss": 0.6268,
+      "step": 2895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538085950895057e-05,
+      "loss": 0.6407,
+      "step": 2896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537668441082355e-05,
+      "loss": 0.6102,
+      "step": 2897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537250747132746e-05,
+      "loss": 0.6554,
+      "step": 2898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536832869054296e-05,
+      "loss": 0.6231,
+      "step": 2899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536414806855074e-05,
+      "loss": 0.635,
+      "step": 2900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535996560543148e-05,
+      "loss": 0.6596,
+      "step": 2901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953557813012659e-05,
+      "loss": 0.6354,
+      "step": 2902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953515951561349e-05,
+      "loss": 0.6318,
+      "step": 2903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534740717011917e-05,
+      "loss": 0.6167,
+      "step": 2904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953432173432996e-05,
+      "loss": 0.6069,
+      "step": 2905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533902567575712e-05,
+      "loss": 0.6286,
+      "step": 2906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533483216757265e-05,
+      "loss": 0.6022,
+      "step": 2907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533063681882712e-05,
+      "loss": 0.6415,
+      "step": 2908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532643962960154e-05,
+      "loss": 0.6244,
+      "step": 2909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532224059997693e-05,
+      "loss": 0.6308,
+      "step": 2910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953180397300344e-05,
+      "loss": 0.6289,
+      "step": 2911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95313837019855e-05,
+      "loss": 0.3277,
+      "step": 2912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953096324695199e-05,
+      "loss": 0.6606,
+      "step": 2913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530542607911027e-05,
+      "loss": 0.681,
+      "step": 2914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953012178487073e-05,
+      "loss": 0.6203,
+      "step": 2915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529700777839226e-05,
+      "loss": 0.6228,
+      "step": 2916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529279586824645e-05,
+      "loss": 0.6694,
+      "step": 2917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952885821183511e-05,
+      "loss": 0.671,
+      "step": 2918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528436652878765e-05,
+      "loss": 0.6362,
+      "step": 2919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952801490996375e-05,
+      "loss": 0.6551,
+      "step": 2920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527592983098196e-05,
+      "loss": 0.5861,
+      "step": 2921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952717087229026e-05,
+      "loss": 0.6295,
+      "step": 2922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526748577548085e-05,
+      "loss": 0.6308,
+      "step": 2923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952632609887983e-05,
+      "loss": 0.6136,
+      "step": 2924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525903436293644e-05,
+      "loss": 0.6171,
+      "step": 2925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952548058979769e-05,
+      "loss": 0.6174,
+      "step": 2926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525057559400134e-05,
+      "loss": 0.6579,
+      "step": 2927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524634345109143e-05,
+      "loss": 0.6082,
+      "step": 2928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524210946932886e-05,
+      "loss": 0.624,
+      "step": 2929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523787364879532e-05,
+      "loss": 0.6029,
+      "step": 2930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523363598957274e-05,
+      "loss": 0.613,
+      "step": 2931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522939649174276e-05,
+      "loss": 0.6431,
+      "step": 2932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522515515538732e-05,
+      "loss": 0.6697,
+      "step": 2933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952209119805883e-05,
+      "loss": 0.6066,
+      "step": 2934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952166669674276e-05,
+      "loss": 0.6408,
+      "step": 2935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952124201159872e-05,
+      "loss": 0.6008,
+      "step": 2936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520817142634905e-05,
+      "loss": 0.6025,
+      "step": 2937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952039208985952e-05,
+      "loss": 0.6494,
+      "step": 2938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519966853280774e-05,
+      "loss": 0.6223,
+      "step": 2939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951954143290687e-05,
+      "loss": 0.6259,
+      "step": 2940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519115828746033e-05,
+      "loss": 0.6482,
+      "step": 2941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518690040806465e-05,
+      "loss": 0.6576,
+      "step": 2942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518264069096395e-05,
+      "loss": 0.6214,
+      "step": 2943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517837913624048e-05,
+      "loss": 0.6426,
+      "step": 2944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517411574397647e-05,
+      "loss": 0.6285,
+      "step": 2945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516985051425425e-05,
+      "loss": 0.6132,
+      "step": 2946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951655834471562e-05,
+      "loss": 0.6123,
+      "step": 2947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951613145427646e-05,
+      "loss": 0.5909,
+      "step": 2948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515704380116197e-05,
+      "loss": 0.3229,
+      "step": 2949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515277122243076e-05,
+      "loss": 0.6303,
+      "step": 2950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514849680665338e-05,
+      "loss": 0.6249,
+      "step": 2951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514422055391243e-05,
+      "loss": 0.577,
+      "step": 2952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951399424642904e-05,
+      "loss": 0.6159,
+      "step": 2953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513566253786995e-05,
+      "loss": 0.6157,
+      "step": 2954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513138077473362e-05,
+      "loss": 0.6746,
+      "step": 2955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512709717496417e-05,
+      "loss": 0.5972,
+      "step": 2956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951228117386443e-05,
+      "loss": 0.6281,
+      "step": 2957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511852446585667e-05,
+      "loss": 0.594,
+      "step": 2958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951142353566841e-05,
+      "loss": 0.6174,
+      "step": 2959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951099444112094e-05,
+      "loss": 0.6266,
+      "step": 2960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.6328,
+      "step": 2961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510135701168494e-05,
+      "loss": 0.5697,
+      "step": 2962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95097060557801e-05,
+      "loss": 0.6186,
+      "step": 2963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509276226794648e-05,
+      "loss": 0.6344,
+      "step": 2964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950884621422044e-05,
+      "loss": 0.6102,
+      "step": 2965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508416018065778e-05,
+      "loss": 0.6452,
+      "step": 2966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507985638338963e-05,
+      "loss": 0.6444,
+      "step": 2967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507555075048308e-05,
+      "loss": 0.6092,
+      "step": 2968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507124328202123e-05,
+      "loss": 0.6262,
+      "step": 2969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506693397808724e-05,
+      "loss": 0.3145,
+      "step": 2970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506262283876437e-05,
+      "loss": 0.663,
+      "step": 2971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505830986413578e-05,
+      "loss": 0.6179,
+      "step": 2972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505399505428478e-05,
+      "loss": 0.6394,
+      "step": 2973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504967840929462e-05,
+      "loss": 0.6227,
+      "step": 2974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950453599292487e-05,
+      "loss": 0.6232,
+      "step": 2975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504103961423032e-05,
+      "loss": 0.644,
+      "step": 2976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95036717464323e-05,
+      "loss": 0.6131,
+      "step": 2977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503239347961006e-05,
+      "loss": 0.6443,
+      "step": 2978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502806766017507e-05,
+      "loss": 0.6121,
+      "step": 2979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502374000610152e-05,
+      "loss": 0.6099,
+      "step": 2980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501941051747292e-05,
+      "loss": 0.6591,
+      "step": 2981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501507919437295e-05,
+      "loss": 0.6578,
+      "step": 2982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501074603688514e-05,
+      "loss": 0.6336,
+      "step": 2983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950064110450932e-05,
+      "loss": 0.6141,
+      "step": 2984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950020742190808e-05,
+      "loss": 0.6425,
+      "step": 2985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499773555893163e-05,
+      "loss": 0.6287,
+      "step": 2986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499339506472954e-05,
+      "loss": 0.6219,
+      "step": 2987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498905273655828e-05,
+      "loss": 0.5792,
+      "step": 2988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949847085745017e-05,
+      "loss": 0.6551,
+      "step": 2989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498036257864365e-05,
+      "loss": 0.6301,
+      "step": 2990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497601474906803e-05,
+      "loss": 0.6406,
+      "step": 2991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497166508585877e-05,
+      "loss": 0.5862,
+      "step": 2992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496731358909992e-05,
+      "loss": 0.6596,
+      "step": 2993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949629602588754e-05,
+      "loss": 0.6032,
+      "step": 2994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495860509526935e-05,
+      "loss": 0.6443,
+      "step": 2995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495424809836575e-05,
+      "loss": 0.5735,
+      "step": 2996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494988926824877e-05,
+      "loss": 0.6386,
+      "step": 2997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494552860500256e-05,
+      "loss": 0.595,
+      "step": 2998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494116610871133e-05,
+      "loss": 0.5913,
+      "step": 2999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949368017794593e-05,
+      "loss": 0.6456,
+      "step": 3000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493243561733063e-05,
+      "loss": 0.629,
+      "step": 3001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492806762240975e-05,
+      "loss": 0.6357,
+      "step": 3002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492369779478094e-05,
+      "loss": 0.6947,
+      "step": 3003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491932613452854e-05,
+      "loss": 0.5978,
+      "step": 3004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94914952641737e-05,
+      "loss": 0.6532,
+      "step": 3005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491057731649072e-05,
+      "loss": 0.6363,
+      "step": 3006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490620015887415e-05,
+      "loss": 0.6071,
+      "step": 3007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490182116897183e-05,
+      "loss": 0.6046,
+      "step": 3008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489744034686827e-05,
+      "loss": 0.6402,
+      "step": 3009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489305769264813e-05,
+      "loss": 0.5877,
+      "step": 3010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488867320639594e-05,
+      "loss": 0.622,
+      "step": 3011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948842868881964e-05,
+      "loss": 0.66,
+      "step": 3012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487989873813412e-05,
+      "loss": 0.5933,
+      "step": 3013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948755087562939e-05,
+      "loss": 0.6339,
+      "step": 3014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487111694276043e-05,
+      "loss": 0.6296,
+      "step": 3015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486672329761857e-05,
+      "loss": 0.6006,
+      "step": 3016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486232782095306e-05,
+      "loss": 0.6229,
+      "step": 3017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485793051284884e-05,
+      "loss": 0.5997,
+      "step": 3018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485353137339077e-05,
+      "loss": 0.631,
+      "step": 3019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484913040266377e-05,
+      "loss": 0.6007,
+      "step": 3020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948447276007528e-05,
+      "loss": 0.5841,
+      "step": 3021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484032296774293e-05,
+      "loss": 0.6272,
+      "step": 3022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948359165037191e-05,
+      "loss": 0.6283,
+      "step": 3023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483150820876644e-05,
+      "loss": 0.6049,
+      "step": 3024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482709808297006e-05,
+      "loss": 0.6609,
+      "step": 3025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948226861264151e-05,
+      "loss": 0.3037,
+      "step": 3026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948182723391867e-05,
+      "loss": 0.6482,
+      "step": 3027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481385672137014e-05,
+      "loss": 0.6341,
+      "step": 3028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948094392730506e-05,
+      "loss": 0.653,
+      "step": 3029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948050199943134e-05,
+      "loss": 0.635,
+      "step": 3030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480059888524386e-05,
+      "loss": 0.6344,
+      "step": 3031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479617594592733e-05,
+      "loss": 0.6595,
+      "step": 3032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947917511764492e-05,
+      "loss": 0.626,
+      "step": 3033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947873245768949e-05,
+      "loss": 0.6491,
+      "step": 3034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947828961473499e-05,
+      "loss": 0.6255,
+      "step": 3035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477846588789967e-05,
+      "loss": 0.641,
+      "step": 3036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477403379862976e-05,
+      "loss": 0.662,
+      "step": 3037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476959987962575e-05,
+      "loss": 0.6435,
+      "step": 3038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947651641309732e-05,
+      "loss": 0.621,
+      "step": 3039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476072655275782e-05,
+      "loss": 0.6051,
+      "step": 3040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947562871450652e-05,
+      "loss": 0.6299,
+      "step": 3041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947518459079811e-05,
+      "loss": 0.6467,
+      "step": 3042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474740284159125e-05,
+      "loss": 0.6468,
+      "step": 3043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474295794598142e-05,
+      "loss": 0.648,
+      "step": 3044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473851122123743e-05,
+      "loss": 0.6302,
+      "step": 3045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473406266744518e-05,
+      "loss": 0.6443,
+      "step": 3046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472961228469046e-05,
+      "loss": 0.6752,
+      "step": 3047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472516007305928e-05,
+      "loss": 0.5897,
+      "step": 3048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472070603263747e-05,
+      "loss": 0.6474,
+      "step": 3049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471625016351117e-05,
+      "loss": 0.6577,
+      "step": 3050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471179246576634e-05,
+      "loss": 0.613,
+      "step": 3051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470733293948903e-05,
+      "loss": 0.5994,
+      "step": 3052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470287158476537e-05,
+      "loss": 0.627,
+      "step": 3053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469840840168146e-05,
+      "loss": 0.5767,
+      "step": 3054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469394339032348e-05,
+      "loss": 0.6259,
+      "step": 3055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468947655077763e-05,
+      "loss": 0.6205,
+      "step": 3056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468500788313017e-05,
+      "loss": 0.6129,
+      "step": 3057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946805373874673e-05,
+      "loss": 0.654,
+      "step": 3058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467606506387544e-05,
+      "loss": 0.6515,
+      "step": 3059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467159091244087e-05,
+      "loss": 0.6181,
+      "step": 3060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466711493324993e-05,
+      "loss": 0.6568,
+      "step": 3061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946626371263891e-05,
+      "loss": 0.5919,
+      "step": 3062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465815749194482e-05,
+      "loss": 0.5742,
+      "step": 3063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465367603000358e-05,
+      "loss": 0.6294,
+      "step": 3064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464919274065187e-05,
+      "loss": 0.6163,
+      "step": 3065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464470762397626e-05,
+      "loss": 0.6318,
+      "step": 3066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464022068006336e-05,
+      "loss": 0.6233,
+      "step": 3067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463573190899977e-05,
+      "loss": 0.6668,
+      "step": 3068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463124131087214e-05,
+      "loss": 0.6444,
+      "step": 3069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946267488857672e-05,
+      "loss": 0.6318,
+      "step": 3070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462225463377166e-05,
+      "loss": 0.6194,
+      "step": 3071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946177585549723e-05,
+      "loss": 0.6617,
+      "step": 3072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946132606494559e-05,
+      "loss": 0.6231,
+      "step": 3073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460876091730937e-05,
+      "loss": 0.6043,
+      "step": 3074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946042593586195e-05,
+      "loss": 0.626,
+      "step": 3075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945997559734732e-05,
+      "loss": 0.598,
+      "step": 3076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459525076195747e-05,
+      "loss": 0.6092,
+      "step": 3077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459074372415925e-05,
+      "loss": 0.6632,
+      "step": 3078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458623486016558e-05,
+      "loss": 0.6393,
+      "step": 3079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 0.6422,
+      "step": 3080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457721165394007e-05,
+      "loss": 0.6603,
+      "step": 3081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945726973118824e-05,
+      "loss": 0.6543,
+      "step": 3082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456818114397772e-05,
+      "loss": 0.6081,
+      "step": 3083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456366315031315e-05,
+      "loss": 0.6126,
+      "step": 3084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455914333097597e-05,
+      "loss": 0.6218,
+      "step": 3085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455462168605337e-05,
+      "loss": 0.669,
+      "step": 3086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945500982156327e-05,
+      "loss": 0.634,
+      "step": 3087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945455729198013e-05,
+      "loss": 0.6936,
+      "step": 3088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945410457986465e-05,
+      "loss": 0.6287,
+      "step": 3089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453651685225567e-05,
+      "loss": 0.6355,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453198608071634e-05,
+      "loss": 0.6168,
+      "step": 3091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452745348411594e-05,
+      "loss": 0.6326,
+      "step": 3092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452291906254196e-05,
+      "loss": 0.6133,
+      "step": 3093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94518382816082e-05,
+      "loss": 0.6122,
+      "step": 3094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451384474482352e-05,
+      "loss": 0.6251,
+      "step": 3095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450930484885423e-05,
+      "loss": 0.6166,
+      "step": 3096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450476312826178e-05,
+      "loss": 0.6148,
+      "step": 3097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945002195831338e-05,
+      "loss": 0.6474,
+      "step": 3098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944956742135581e-05,
+      "loss": 0.6475,
+      "step": 3099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944911270196223e-05,
+      "loss": 0.6373,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944865780014143e-05,
+      "loss": 0.6139,
+      "step": 3101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448202715902188e-05,
+      "loss": 0.6343,
+      "step": 3102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944774744925329e-05,
+      "loss": 0.607,
+      "step": 3103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447292000203526e-05,
+      "loss": 0.6117,
+      "step": 3104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446836368761692e-05,
+      "loss": 0.6137,
+      "step": 3105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944638055493658e-05,
+      "loss": 0.6037,
+      "step": 3106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944592455873699e-05,
+      "loss": 0.6351,
+      "step": 3107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944546838017173e-05,
+      "loss": 0.6334,
+      "step": 3108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445012019249604e-05,
+      "loss": 0.5969,
+      "step": 3109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444555475979425e-05,
+      "loss": 0.6519,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444098750369998e-05,
+      "loss": 0.6495,
+      "step": 3111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443641842430153e-05,
+      "loss": 0.6146,
+      "step": 3112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944318475216871e-05,
+      "loss": 0.6106,
+      "step": 3113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442727479594486e-05,
+      "loss": 0.6192,
+      "step": 3114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442270024716313e-05,
+      "loss": 0.6437,
+      "step": 3115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441812387543022e-05,
+      "loss": 0.6003,
+      "step": 3116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441354568083447e-05,
+      "loss": 0.6699,
+      "step": 3117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440896566346433e-05,
+      "loss": 0.635,
+      "step": 3118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944043838234082e-05,
+      "loss": 0.6056,
+      "step": 3119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439980016075447e-05,
+      "loss": 0.6045,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943952146755917e-05,
+      "loss": 0.6091,
+      "step": 3121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439062736800843e-05,
+      "loss": 0.5891,
+      "step": 3122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943860382380932e-05,
+      "loss": 0.6463,
+      "step": 3123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438144728593458e-05,
+      "loss": 0.6445,
+      "step": 3124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437685451162122e-05,
+      "loss": 0.5842,
+      "step": 3125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437225991524184e-05,
+      "loss": 0.629,
+      "step": 3126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436766349688506e-05,
+      "loss": 0.6151,
+      "step": 3127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436306525663967e-05,
+      "loss": 0.5972,
+      "step": 3128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435846519459445e-05,
+      "loss": 0.6314,
+      "step": 3129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435386331083822e-05,
+      "loss": 0.6076,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434925960545978e-05,
+      "loss": 0.5985,
+      "step": 3131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434465407854803e-05,
+      "loss": 0.6635,
+      "step": 3132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943400467301919e-05,
+      "loss": 0.6154,
+      "step": 3133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433543756048033e-05,
+      "loss": 0.6598,
+      "step": 3134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433082656950232e-05,
+      "loss": 0.6006,
+      "step": 3135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432621375734686e-05,
+      "loss": 0.6345,
+      "step": 3136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432159912410302e-05,
+      "loss": 0.6562,
+      "step": 3137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943169826698599e-05,
+      "loss": 0.6034,
+      "step": 3138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431236439470664e-05,
+      "loss": 0.6316,
+      "step": 3139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430774429873235e-05,
+      "loss": 0.6505,
+      "step": 3140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430312238202628e-05,
+      "loss": 0.6182,
+      "step": 3141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429849864467764e-05,
+      "loss": 0.6292,
+      "step": 3142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942938730867757e-05,
+      "loss": 0.6371,
+      "step": 3143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942892457084098e-05,
+      "loss": 0.6219,
+      "step": 3144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942846165096692e-05,
+      "loss": 0.6256,
+      "step": 3145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942799854906433e-05,
+      "loss": 0.6048,
+      "step": 3146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427535265142152e-05,
+      "loss": 0.6295,
+      "step": 3147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427071799209335e-05,
+      "loss": 0.6074,
+      "step": 3148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426608151274816e-05,
+      "loss": 0.611,
+      "step": 3149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426144321347557e-05,
+      "loss": 0.6159,
+      "step": 3150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425680309436506e-05,
+      "loss": 0.621,
+      "step": 3151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425216115550624e-05,
+      "loss": 0.6557,
+      "step": 3152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424751739698874e-05,
+      "loss": 0.6069,
+      "step": 3153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942428718189022e-05,
+      "loss": 0.6188,
+      "step": 3154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942382244213363e-05,
+      "loss": 0.6234,
+      "step": 3155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423357520438073e-05,
+      "loss": 0.6397,
+      "step": 3156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942289241681253e-05,
+      "loss": 0.6131,
+      "step": 3157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422427131265983e-05,
+      "loss": 0.6427,
+      "step": 3158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421961663807413e-05,
+      "loss": 0.5973,
+      "step": 3159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.94214960144458e-05,
+      "loss": 0.626,
+      "step": 3160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421030183190138e-05,
+      "loss": 0.6835,
+      "step": 3161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420564170049426e-05,
+      "loss": 0.6495,
+      "step": 3162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420097975032655e-05,
+      "loss": 0.6099,
+      "step": 3163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419631598148827e-05,
+      "loss": 0.6179,
+      "step": 3164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941916503940694e-05,
+      "loss": 0.6106,
+      "step": 3165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418698298816014e-05,
+      "loss": 0.5567,
+      "step": 3166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941823137638505e-05,
+      "loss": 0.6338,
+      "step": 3167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941776427212307e-05,
+      "loss": 0.6004,
+      "step": 3168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417296986039084e-05,
+      "loss": 0.6281,
+      "step": 3169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941682951814212e-05,
+      "loss": 0.6369,
+      "step": 3170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416361868441193e-05,
+      "loss": 0.5818,
+      "step": 3171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415894036945346e-05,
+      "loss": 0.6283,
+      "step": 3172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415426023663602e-05,
+      "loss": 0.6306,
+      "step": 3173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414957828605e-05,
+      "loss": 0.6273,
+      "step": 3174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414489451778577e-05,
+      "loss": 0.6306,
+      "step": 3175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414020893193375e-05,
+      "loss": 0.6348,
+      "step": 3176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413552152858444e-05,
+      "loss": 0.6079,
+      "step": 3177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413083230782833e-05,
+      "loss": 0.64,
+      "step": 3178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412614126975588e-05,
+      "loss": 0.6006,
+      "step": 3179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412144841445776e-05,
+      "loss": 0.6239,
+      "step": 3180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411675374202447e-05,
+      "loss": 0.6276,
+      "step": 3181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941120572525467e-05,
+      "loss": 0.6173,
+      "step": 3182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410735894611516e-05,
+      "loss": 0.6081,
+      "step": 3183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410265882282048e-05,
+      "loss": 0.6107,
+      "step": 3184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940979568827535e-05,
+      "loss": 0.6322,
+      "step": 3185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409325312600486e-05,
+      "loss": 0.6157,
+      "step": 3186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408854755266547e-05,
+      "loss": 0.6505,
+      "step": 3187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940838401628262e-05,
+      "loss": 0.6248,
+      "step": 3188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407913095657785e-05,
+      "loss": 0.6137,
+      "step": 3189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407441993401137e-05,
+      "loss": 0.6129,
+      "step": 3190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406970709521773e-05,
+      "loss": 0.6392,
+      "step": 3191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406499244028788e-05,
+      "loss": 0.6094,
+      "step": 3192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406027596931288e-05,
+      "loss": 0.6338,
+      "step": 3193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940555576823838e-05,
+      "loss": 0.6024,
+      "step": 3194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405083757959168e-05,
+      "loss": 0.6086,
+      "step": 3195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940461156610277e-05,
+      "loss": 0.6105,
+      "step": 3196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404139192678294e-05,
+      "loss": 0.623,
+      "step": 3197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940366663769487e-05,
+      "loss": 0.6114,
+      "step": 3198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403193901161614e-05,
+      "loss": 0.6295,
+      "step": 3199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402720983087656e-05,
+      "loss": 0.6242,
+      "step": 3200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402247883482127e-05,
+      "loss": 0.6399,
+      "step": 3201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940177460235416e-05,
+      "loss": 0.6026,
+      "step": 3202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940130113971289e-05,
+      "loss": 0.6139,
+      "step": 3203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400827495567464e-05,
+      "loss": 0.6213,
+      "step": 3204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400353669927018e-05,
+      "loss": 0.6454,
+      "step": 3205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399879662800702e-05,
+      "loss": 0.5962,
+      "step": 3206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399405474197673e-05,
+      "loss": 0.6227,
+      "step": 3207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939893110412708e-05,
+      "loss": 0.625,
+      "step": 3208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398456552598085e-05,
+      "loss": 0.6007,
+      "step": 3209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939798181961985e-05,
+      "loss": 0.619,
+      "step": 3210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397506905201534e-05,
+      "loss": 0.6091,
+      "step": 3211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397031809352313e-05,
+      "loss": 0.6694,
+      "step": 3212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396556532081353e-05,
+      "loss": 0.6467,
+      "step": 3213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396081073397838e-05,
+      "loss": 0.6299,
+      "step": 3214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395605433310937e-05,
+      "loss": 0.6653,
+      "step": 3215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9395129611829844e-05,
+      "loss": 0.5867,
+      "step": 3216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394653608963737e-05,
+      "loss": 0.6185,
+      "step": 3217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939417742472181e-05,
+      "loss": 0.6002,
+      "step": 3218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393701059113253e-05,
+      "loss": 0.6033,
+      "step": 3219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939322451214727e-05,
+      "loss": 0.6221,
+      "step": 3220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939274778383305e-05,
+      "loss": 0.6036,
+      "step": 3221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939227087417981e-05,
+      "loss": 0.6342,
+      "step": 3222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391793783196747e-05,
+      "loss": 0.6224,
+      "step": 3223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939131651089307e-05,
+      "loss": 0.6072,
+      "step": 3224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390839057278005e-05,
+      "loss": 0.6081,
+      "step": 3225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390361422360765e-05,
+      "loss": 0.6199,
+      "step": 3226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938988360615057e-05,
+      "loss": 0.6234,
+      "step": 3227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938940560865664e-05,
+      "loss": 0.6343,
+      "step": 3228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938892742988821e-05,
+      "loss": 0.6492,
+      "step": 3229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938844906985451e-05,
+      "loss": 0.6315,
+      "step": 3230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387970528564777e-05,
+      "loss": 0.6252,
+      "step": 3231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387491806028244e-05,
+      "loss": 0.6244,
+      "step": 3232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387012902254165e-05,
+      "loss": 0.624,
+      "step": 3233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386533817251772e-05,
+      "loss": 0.6188,
+      "step": 3234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386054551030323e-05,
+      "loss": 0.6033,
+      "step": 3235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938557510359907e-05,
+      "loss": 0.5829,
+      "step": 3236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385095474967262e-05,
+      "loss": 0.6486,
+      "step": 3237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938461566514417e-05,
+      "loss": 0.6607,
+      "step": 3238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938413567413905e-05,
+      "loss": 0.648,
+      "step": 3239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383655501961173e-05,
+      "loss": 0.6251,
+      "step": 3240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383175148619807e-05,
+      "loss": 0.6136,
+      "step": 3241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382694614124227e-05,
+      "loss": 0.6148,
+      "step": 3242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382213898483705e-05,
+      "loss": 0.621,
+      "step": 3243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381733001707532e-05,
+      "loss": 0.6252,
+      "step": 3244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938125192380498e-05,
+      "loss": 0.644,
+      "step": 3245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380770664785352e-05,
+      "loss": 0.6216,
+      "step": 3246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380289224657923e-05,
+      "loss": 0.592,
+      "step": 3247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379807603432e-05,
+      "loss": 0.6112,
+      "step": 3248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379325801116878e-05,
+      "loss": 0.6145,
+      "step": 3249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378843817721856e-05,
+      "loss": 0.6407,
+      "step": 3250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378361653256238e-05,
+      "loss": 0.6528,
+      "step": 3251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377879307729342e-05,
+      "loss": 0.6098,
+      "step": 3252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937739678115047e-05,
+      "loss": 0.6056,
+      "step": 3253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376914073528944e-05,
+      "loss": 0.622,
+      "step": 3254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937643118487408e-05,
+      "loss": 0.6191,
+      "step": 3255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93759481151952e-05,
+      "loss": 0.6303,
+      "step": 3256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375464864501636e-05,
+      "loss": 0.5995,
+      "step": 3257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937498143280271e-05,
+      "loss": 0.6241,
+      "step": 3258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374497820107763e-05,
+      "loss": 0.3,
+      "step": 3259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374014026426126e-05,
+      "loss": 0.6024,
+      "step": 3260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373530051767142e-05,
+      "loss": 0.6469,
+      "step": 3261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373045896140152e-05,
+      "loss": 0.6339,
+      "step": 3262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937256155955451e-05,
+      "loss": 0.624,
+      "step": 3263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937207704201956e-05,
+      "loss": 0.5919,
+      "step": 3264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371592343544655e-05,
+      "loss": 0.5867,
+      "step": 3265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937110746413916e-05,
+      "loss": 0.6315,
+      "step": 3266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937062240381243e-05,
+      "loss": 0.6034,
+      "step": 3267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370137162573833e-05,
+      "loss": 0.6046,
+      "step": 3268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369651740432737e-05,
+      "loss": 0.658,
+      "step": 3269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369166137398513e-05,
+      "loss": 0.6007,
+      "step": 3270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368680353480535e-05,
+      "loss": 0.6291,
+      "step": 3271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368194388688185e-05,
+      "loss": 0.5977,
+      "step": 3272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936770824303084e-05,
+      "loss": 0.6584,
+      "step": 3273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936722191651789e-05,
+      "loss": 0.6122,
+      "step": 3274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936673540915872e-05,
+      "loss": 0.6453,
+      "step": 3275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366248720962727e-05,
+      "loss": 0.619,
+      "step": 3276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365761851939308e-05,
+      "loss": 0.6209,
+      "step": 3277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365274802097857e-05,
+      "loss": 0.6095,
+      "step": 3278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364787571447782e-05,
+      "loss": 0.6829,
+      "step": 3279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936430015999849e-05,
+      "loss": 0.5921,
+      "step": 3280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363812567759383e-05,
+      "loss": 0.6096,
+      "step": 3281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363324794739885e-05,
+      "loss": 0.6156,
+      "step": 3282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936283684094941e-05,
+      "loss": 0.6673,
+      "step": 3283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362348706397374e-05,
+      "loss": 0.6554,
+      "step": 3284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361860391093207e-05,
+      "loss": 0.6141,
+      "step": 3285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936137189504633e-05,
+      "loss": 0.6609,
+      "step": 3286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936088321826618e-05,
+      "loss": 0.6244,
+      "step": 3287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936039436076219e-05,
+      "loss": 0.6219,
+      "step": 3288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359905322543798e-05,
+      "loss": 0.6281,
+      "step": 3289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359416103620444e-05,
+      "loss": 0.6351,
+      "step": 3290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358926704001574e-05,
+      "loss": 0.6806,
+      "step": 3291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358437123696635e-05,
+      "loss": 0.6314,
+      "step": 3292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357947362715078e-05,
+      "loss": 0.6394,
+      "step": 3293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357457421066364e-05,
+      "loss": 0.6548,
+      "step": 3294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356967298759953e-05,
+      "loss": 0.6526,
+      "step": 3295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93564769958053e-05,
+      "loss": 0.6556,
+      "step": 3296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355986512211873e-05,
+      "loss": 0.6343,
+      "step": 3297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355495847989143e-05,
+      "loss": 0.6492,
+      "step": 3298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355005003146582e-05,
+      "loss": 0.6207,
+      "step": 3299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935451397769367e-05,
+      "loss": 0.6202,
+      "step": 3300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935402277163988e-05,
+      "loss": 0.6451,
+      "step": 3301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93535313849947e-05,
+      "loss": 0.6682,
+      "step": 3302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935303981776762e-05,
+      "loss": 0.6283,
+      "step": 3303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352548069968123e-05,
+      "loss": 0.6309,
+      "step": 3304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352056141605705e-05,
+      "loss": 0.6018,
+      "step": 3305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351564032689865e-05,
+      "loss": 0.6159,
+      "step": 3306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351071743230102e-05,
+      "loss": 0.6115,
+      "step": 3307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350579273235925e-05,
+      "loss": 0.626,
+      "step": 3308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935008662271684e-05,
+      "loss": 0.6281,
+      "step": 3309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934959379168235e-05,
+      "loss": 0.6102,
+      "step": 3310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349100780141977e-05,
+      "loss": 0.6093,
+      "step": 3311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348607588105244e-05,
+      "loss": 0.6531,
+      "step": 3312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348114215581662e-05,
+      "loss": 0.622,
+      "step": 3313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347620662580765e-05,
+      "loss": 0.6473,
+      "step": 3314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347126929112077e-05,
+      "loss": 0.629,
+      "step": 3315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934663301518513e-05,
+      "loss": 0.6352,
+      "step": 3316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346138920809464e-05,
+      "loss": 0.6257,
+      "step": 3317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934564464599461e-05,
+      "loss": 0.6611,
+      "step": 3318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345150190750118e-05,
+      "loss": 0.6114,
+      "step": 3319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934465555508553e-05,
+      "loss": 0.6312,
+      "step": 3320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344160739010403e-05,
+      "loss": 0.6449,
+      "step": 3321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934366574253428e-05,
+      "loss": 0.6268,
+      "step": 3322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934317056566672e-05,
+      "loss": 0.652,
+      "step": 3323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342675208417287e-05,
+      "loss": 0.6229,
+      "step": 3324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342179670795544e-05,
+      "loss": 0.6274,
+      "step": 3325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341683952811054e-05,
+      "loss": 0.6151,
+      "step": 3326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934118805447339e-05,
+      "loss": 0.5772,
+      "step": 3327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340691975792128e-05,
+      "loss": 0.6067,
+      "step": 3328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340195716776842e-05,
+      "loss": 0.6327,
+      "step": 3329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933969927743711e-05,
+      "loss": 0.6169,
+      "step": 3330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933920265778253e-05,
+      "loss": 0.5819,
+      "step": 3331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338705857822672e-05,
+      "loss": 0.6199,
+      "step": 3332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933820887756714e-05,
+      "loss": 0.6394,
+      "step": 3333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933771171702552e-05,
+      "loss": 0.6171,
+      "step": 3334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337214376207417e-05,
+      "loss": 0.6071,
+      "step": 3335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336716855122435e-05,
+      "loss": 0.5939,
+      "step": 3336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933621915378017e-05,
+      "loss": 0.6479,
+      "step": 3337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933572127219024e-05,
+      "loss": 0.6005,
+      "step": 3338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933522321036225e-05,
+      "loss": 0.5935,
+      "step": 3339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334724968305818e-05,
+      "loss": 0.6336,
+      "step": 3340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334226546030567e-05,
+      "loss": 0.6226,
+      "step": 3341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333727943546115e-05,
+      "loss": 0.6114,
+      "step": 3342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933322916086209e-05,
+      "loss": 0.6001,
+      "step": 3343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332730197988118e-05,
+      "loss": 0.6335,
+      "step": 3344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933223105493384e-05,
+      "loss": 0.59,
+      "step": 3345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331731731708888e-05,
+      "loss": 0.661,
+      "step": 3346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93312322283229e-05,
+      "loss": 0.6254,
+      "step": 3347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330732544785525e-05,
+      "loss": 0.639,
+      "step": 3348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330232681106404e-05,
+      "loss": 0.6056,
+      "step": 3349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932973263729519e-05,
+      "loss": 0.5872,
+      "step": 3350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329232413361537e-05,
+      "loss": 0.6311,
+      "step": 3351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328732009315107e-05,
+      "loss": 0.5698,
+      "step": 3352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328231425165552e-05,
+      "loss": 0.6049,
+      "step": 3353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327730660922543e-05,
+      "loss": 0.6523,
+      "step": 3354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327229716595745e-05,
+      "loss": 0.6213,
+      "step": 3355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932672859219483e-05,
+      "loss": 0.6506,
+      "step": 3356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326227287729478e-05,
+      "loss": 0.6202,
+      "step": 3357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932572580320936e-05,
+      "loss": 0.6326,
+      "step": 3358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325224138644157e-05,
+      "loss": 0.6342,
+      "step": 3359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 0.6366,
+      "step": 3360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324220269417255e-05,
+      "loss": 0.6003,
+      "step": 3361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323718064774936e-05,
+      "loss": 0.6279,
+      "step": 3362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323215680126294e-05,
+      "loss": 0.6437,
+      "step": 3363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322713115481036e-05,
+      "loss": 0.6282,
+      "step": 3364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322210370848856e-05,
+      "loss": 0.6249,
+      "step": 3365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321707446239468e-05,
+      "loss": 0.6101,
+      "step": 3366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321204341662576e-05,
+      "loss": 0.6089,
+      "step": 3367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320701057127892e-05,
+      "loss": 0.6312,
+      "step": 3368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932019759264514e-05,
+      "loss": 0.6294,
+      "step": 3369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319693948224035e-05,
+      "loss": 0.622,
+      "step": 3370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319190123874302e-05,
+      "loss": 0.6196,
+      "step": 3371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318686119605664e-05,
+      "loss": 0.6153,
+      "step": 3372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318181935427854e-05,
+      "loss": 0.6776,
+      "step": 3373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317677571350608e-05,
+      "loss": 0.6595,
+      "step": 3374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317173027383662e-05,
+      "loss": 0.623,
+      "step": 3375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316668303536757e-05,
+      "loss": 0.6491,
+      "step": 3376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316163399819638e-05,
+      "loss": 0.6194,
+      "step": 3377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931565831624205e-05,
+      "loss": 0.669,
+      "step": 3378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315153052813743e-05,
+      "loss": 0.615,
+      "step": 3379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931464760954448e-05,
+      "loss": 0.5972,
+      "step": 3380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314141986444014e-05,
+      "loss": 0.619,
+      "step": 3381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93136361835221e-05,
+      "loss": 0.6079,
+      "step": 3382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313130200788518e-05,
+      "loss": 0.5984,
+      "step": 3383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312624038253025e-05,
+      "loss": 0.6135,
+      "step": 3384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312117695925398e-05,
+      "loss": 0.6647,
+      "step": 3385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931161117381541e-05,
+      "loss": 0.6178,
+      "step": 3386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311104471932843e-05,
+      "loss": 0.5891,
+      "step": 3387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310597590287476e-05,
+      "loss": 0.6306,
+      "step": 3388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93100905288891e-05,
+      "loss": 0.6187,
+      "step": 3389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309583287747496e-05,
+      "loss": 0.6009,
+      "step": 3390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309075866872464e-05,
+      "loss": 0.6014,
+      "step": 3391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93085682662738e-05,
+      "loss": 0.6454,
+      "step": 3392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308060485961304e-05,
+      "loss": 0.6253,
+      "step": 3393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307552525944776e-05,
+      "loss": 0.6408,
+      "step": 3394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307044386234027e-05,
+      "loss": 0.6522,
+      "step": 3395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306536066838862e-05,
+      "loss": 0.6315,
+      "step": 3396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93060275677691e-05,
+      "loss": 0.61,
+      "step": 3397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9305518889034552e-05,
+      "loss": 0.5801,
+      "step": 3398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930501003064505e-05,
+      "loss": 0.6086,
+      "step": 3399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304500992610404e-05,
+      "loss": 0.6201,
+      "step": 3400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303991774940454e-05,
+      "loss": 0.592,
+      "step": 3401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930348237764502e-05,
+      "loss": 0.6283,
+      "step": 3402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302972800733945e-05,
+      "loss": 0.618,
+      "step": 3403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302463044217067e-05,
+      "loss": 0.6039,
+      "step": 3404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930195310810422e-05,
+      "loss": 0.6045,
+      "step": 3405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301442992405257e-05,
+      "loss": 0.6204,
+      "step": 3406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930093269713002e-05,
+      "loss": 0.624,
+      "step": 3407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9300422222288367e-05,
+      "loss": 0.6484,
+      "step": 3408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299911567890148e-05,
+      "loss": 0.6631,
+      "step": 3409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299400733945226e-05,
+      "loss": 0.6262,
+      "step": 3410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929888972046346e-05,
+      "loss": 0.6358,
+      "step": 3411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929837852745472e-05,
+      "loss": 0.6019,
+      "step": 3412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929786715492887e-05,
+      "loss": 0.6281,
+      "step": 3413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297355602895785e-05,
+      "loss": 0.6213,
+      "step": 3414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296843871365343e-05,
+      "loss": 0.6184,
+      "step": 3415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296331960347424e-05,
+      "loss": 0.6425,
+      "step": 3416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295819869851904e-05,
+      "loss": 0.5947,
+      "step": 3417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295307599888676e-05,
+      "loss": 0.5818,
+      "step": 3418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929479515046763e-05,
+      "loss": 0.6218,
+      "step": 3419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929428252159866e-05,
+      "loss": 0.6222,
+      "step": 3420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929376971329166e-05,
+      "loss": 0.6268,
+      "step": 3421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929325672555653e-05,
+      "loss": 0.6693,
+      "step": 3422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292743558403177e-05,
+      "loss": 0.6377,
+      "step": 3423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292230211841504e-05,
+      "loss": 0.6494,
+      "step": 3424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929171668588143e-05,
+      "loss": 0.6692,
+      "step": 3425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929120298053286e-05,
+      "loss": 0.626,
+      "step": 3426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290689095805716e-05,
+      "loss": 0.6582,
+      "step": 3427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929017503170992e-05,
+      "loss": 0.5982,
+      "step": 3428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289660788255396e-05,
+      "loss": 0.6421,
+      "step": 3429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928914636545207e-05,
+      "loss": 0.6238,
+      "step": 3430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288631763309875e-05,
+      "loss": 0.6152,
+      "step": 3431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928811698183875e-05,
+      "loss": 0.6172,
+      "step": 3432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287602021048623e-05,
+      "loss": 0.6085,
+      "step": 3433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928708688094945e-05,
+      "loss": 0.6446,
+      "step": 3434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286571561551165e-05,
+      "loss": 0.6207,
+      "step": 3435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928605606286372e-05,
+      "loss": 0.6953,
+      "step": 3436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285540384897073e-05,
+      "loss": 0.63,
+      "step": 3437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928502452766117e-05,
+      "loss": 0.6141,
+      "step": 3438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284508491165983e-05,
+      "loss": 0.6457,
+      "step": 3439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283992275421462e-05,
+      "loss": 0.602,
+      "step": 3440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928347588043758e-05,
+      "loss": 0.6041,
+      "step": 3441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282959306224304e-05,
+      "loss": 0.668,
+      "step": 3442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928244255279161e-05,
+      "loss": 0.5826,
+      "step": 3443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928192562014947e-05,
+      "loss": 0.626,
+      "step": 3444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928140850830787e-05,
+      "loss": 0.6021,
+      "step": 3445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928089121727679e-05,
+      "loss": 0.6176,
+      "step": 3446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9280373747066216e-05,
+      "loss": 0.671,
+      "step": 3447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927985609768614e-05,
+      "loss": 0.6241,
+      "step": 3448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927933826914656e-05,
+      "loss": 0.6358,
+      "step": 3449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278820261457465e-05,
+      "loss": 0.6511,
+      "step": 3450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278302074628858e-05,
+      "loss": 0.6281,
+      "step": 3451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927778370867075e-05,
+      "loss": 0.588,
+      "step": 3452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927726516359314e-05,
+      "loss": 0.6242,
+      "step": 3453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276746439406046e-05,
+      "loss": 0.6276,
+      "step": 3454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927622753611948e-05,
+      "loss": 0.6174,
+      "step": 3455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927570845374346e-05,
+      "loss": 0.6562,
+      "step": 3456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275189192288003e-05,
+      "loss": 0.6248,
+      "step": 3457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274669751763142e-05,
+      "loss": 0.6006,
+      "step": 3458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92741501321789e-05,
+      "loss": 0.6009,
+      "step": 3459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927363033354531e-05,
+      "loss": 0.6309,
+      "step": 3460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273110355872413e-05,
+      "loss": 0.6088,
+      "step": 3461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927259019917024e-05,
+      "loss": 0.6018,
+      "step": 3462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272069863448835e-05,
+      "loss": 0.6566,
+      "step": 3463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271549348718247e-05,
+      "loss": 0.6576,
+      "step": 3464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927102865498852e-05,
+      "loss": 0.6576,
+      "step": 3465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270507782269713e-05,
+      "loss": 0.6209,
+      "step": 3466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926998673057188e-05,
+      "loss": 0.6408,
+      "step": 3467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269465499905076e-05,
+      "loss": 0.641,
+      "step": 3468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926894409027937e-05,
+      "loss": 0.5952,
+      "step": 3469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268422501704825e-05,
+      "loss": 0.5945,
+      "step": 3470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267900734191515e-05,
+      "loss": 0.6487,
+      "step": 3471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267378787749505e-05,
+      "loss": 0.6365,
+      "step": 3472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926685666238888e-05,
+      "loss": 0.6472,
+      "step": 3473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266334358119718e-05,
+      "loss": 0.6473,
+      "step": 3474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92658118749521e-05,
+      "loss": 0.608,
+      "step": 3475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265289212896117e-05,
+      "loss": 0.6488,
+      "step": 3476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926476637196186e-05,
+      "loss": 0.5755,
+      "step": 3477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264243352159418e-05,
+      "loss": 0.6003,
+      "step": 3478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263720153498892e-05,
+      "loss": 0.6154,
+      "step": 3479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263196775990388e-05,
+      "loss": 0.6117,
+      "step": 3480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262673219644e-05,
+      "loss": 0.6302,
+      "step": 3481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262149484469843e-05,
+      "loss": 0.6054,
+      "step": 3482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261625570478025e-05,
+      "loss": 0.632,
+      "step": 3483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261101477678664e-05,
+      "loss": 0.6268,
+      "step": 3484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260577206081876e-05,
+      "loss": 0.6194,
+      "step": 3485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260052755697785e-05,
+      "loss": 0.6252,
+      "step": 3486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259528126536515e-05,
+      "loss": 0.6109,
+      "step": 3487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259003318608192e-05,
+      "loss": 0.6235,
+      "step": 3488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925847833192295e-05,
+      "loss": 0.5877,
+      "step": 3489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257953166490928e-05,
+      "loss": 0.6404,
+      "step": 3490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925742782232226e-05,
+      "loss": 0.6358,
+      "step": 3491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925690229942709e-05,
+      "loss": 0.6295,
+      "step": 3492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9256376597815565e-05,
+      "loss": 0.59,
+      "step": 3493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925585071749783e-05,
+      "loss": 0.6543,
+      "step": 3494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255324658484048e-05,
+      "loss": 0.6194,
+      "step": 3495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254798420784367e-05,
+      "loss": 0.626,
+      "step": 3496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254272004408945e-05,
+      "loss": 0.6678,
+      "step": 3497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925374540936795e-05,
+      "loss": 0.6065,
+      "step": 3498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253218635671545e-05,
+      "loss": 0.5822,
+      "step": 3499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252691683329907e-05,
+      "loss": 0.6085,
+      "step": 3500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92521645523532e-05,
+      "loss": 0.622,
+      "step": 3501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251637242751608e-05,
+      "loss": 0.6258,
+      "step": 3502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925110975453531e-05,
+      "loss": 0.622,
+      "step": 3503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250582087714486e-05,
+      "loss": 0.61,
+      "step": 3504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925005424229933e-05,
+      "loss": 0.627,
+      "step": 3505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249526218300022e-05,
+      "loss": 0.5592,
+      "step": 3506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924899801572677e-05,
+      "loss": 0.6116,
+      "step": 3507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248469634589758e-05,
+      "loss": 0.602,
+      "step": 3508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247941074899198e-05,
+      "loss": 0.6523,
+      "step": 3509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924741233666529e-05,
+      "loss": 0.623,
+      "step": 3510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924688341989824e-05,
+      "loss": 0.6324,
+      "step": 3511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246354324608264e-05,
+      "loss": 0.6461,
+      "step": 3512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245825050805575e-05,
+      "loss": 0.634,
+      "step": 3513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245295598500385e-05,
+      "loss": 0.6469,
+      "step": 3514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244765967702928e-05,
+      "loss": 0.5797,
+      "step": 3515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924423615842342e-05,
+      "loss": 0.599,
+      "step": 3516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924370617067209e-05,
+      "loss": 0.6005,
+      "step": 3517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243176004459178e-05,
+      "loss": 0.6395,
+      "step": 3518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924264565979491e-05,
+      "loss": 0.6358,
+      "step": 3519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924211513668953e-05,
+      "loss": 0.5917,
+      "step": 3520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924158443515328e-05,
+      "loss": 0.6185,
+      "step": 3521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241053555196405e-05,
+      "loss": 0.6368,
+      "step": 3522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240522496829154e-05,
+      "loss": 0.6093,
+      "step": 3523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923999126006178e-05,
+      "loss": 0.6046,
+      "step": 3524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923945984490454e-05,
+      "loss": 0.5683,
+      "step": 3525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238928251367693e-05,
+      "loss": 0.5832,
+      "step": 3526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92383964794615e-05,
+      "loss": 0.3096,
+      "step": 3527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237864529196233e-05,
+      "loss": 0.6289,
+      "step": 3528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923733240058216e-05,
+      "loss": 0.6275,
+      "step": 3529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923680009362955e-05,
+      "loss": 0.5961,
+      "step": 3530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236267608348682e-05,
+      "loss": 0.6014,
+      "step": 3531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923573494474984e-05,
+      "loss": 0.5784,
+      "step": 3532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9235202102843305e-05,
+      "loss": 0.5761,
+      "step": 3533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234669082639366e-05,
+      "loss": 0.6164,
+      "step": 3534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923413588414831e-05,
+      "loss": 0.6142,
+      "step": 3535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923360250738043e-05,
+      "loss": 0.6424,
+      "step": 3536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233068952346034e-05,
+      "loss": 0.5889,
+      "step": 3537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232535219055407e-05,
+      "loss": 0.6746,
+      "step": 3538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923200130751887e-05,
+      "loss": 0.2975,
+      "step": 3539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923146721774672e-05,
+      "loss": 0.5781,
+      "step": 3540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230932949749265e-05,
+      "loss": 0.6044,
+      "step": 3541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230398503536834e-05,
+      "loss": 0.6036,
+      "step": 3542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229863879119733e-05,
+      "loss": 0.6156,
+      "step": 3543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229329076508287e-05,
+      "loss": 0.6729,
+      "step": 3544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228794095712824e-05,
+      "loss": 0.5842,
+      "step": 3545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922825893674367e-05,
+      "loss": 0.6129,
+      "step": 3546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227723599611156e-05,
+      "loss": 0.6229,
+      "step": 3547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227188084325618e-05,
+      "loss": 0.5922,
+      "step": 3548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226652390897396e-05,
+      "loss": 0.6024,
+      "step": 3549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922611651933683e-05,
+      "loss": 0.622,
+      "step": 3550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225580469654266e-05,
+      "loss": 0.6038,
+      "step": 3551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922504424186006e-05,
+      "loss": 0.6087,
+      "step": 3552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224507835964555e-05,
+      "loss": 0.6065,
+      "step": 3553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922397125197811e-05,
+      "loss": 0.5726,
+      "step": 3554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922343448991109e-05,
+      "loss": 0.6634,
+      "step": 3555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922289754977385e-05,
+      "loss": 0.6247,
+      "step": 3556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222360431576756e-05,
+      "loss": 0.6045,
+      "step": 3557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221823135330185e-05,
+      "loss": 0.6235,
+      "step": 3558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922128566104451e-05,
+      "loss": 0.6282,
+      "step": 3559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220748008730096e-05,
+      "loss": 0.624,
+      "step": 3560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922021017839734e-05,
+      "loss": 0.5988,
+      "step": 3561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219672170056608e-05,
+      "loss": 0.6515,
+      "step": 3562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219133983718302e-05,
+      "loss": 0.612,
+      "step": 3563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92185956193928e-05,
+      "loss": 0.61,
+      "step": 3564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921805707709051e-05,
+      "loss": 0.632,
+      "step": 3565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217518356821814e-05,
+      "loss": 0.6424,
+      "step": 3566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216979458597124e-05,
+      "loss": 0.284,
+      "step": 3567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216440382426835e-05,
+      "loss": 0.6035,
+      "step": 3568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215901128321364e-05,
+      "loss": 0.6237,
+      "step": 3569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215361696291114e-05,
+      "loss": 0.638,
+      "step": 3570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214822086346506e-05,
+      "loss": 0.6487,
+      "step": 3571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214282298497952e-05,
+      "loss": 0.6229,
+      "step": 3572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213742332755877e-05,
+      "loss": 0.6587,
+      "step": 3573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213202189130707e-05,
+      "loss": 0.6414,
+      "step": 3574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212661867632866e-05,
+      "loss": 0.6212,
+      "step": 3575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212121368272784e-05,
+      "loss": 0.6224,
+      "step": 3576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211580691060906e-05,
+      "loss": 0.595,
+      "step": 3577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211039836007658e-05,
+      "loss": 0.6075,
+      "step": 3578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921049880312349e-05,
+      "loss": 0.6231,
+      "step": 3579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209957592418848e-05,
+      "loss": 0.6335,
+      "step": 3580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209416203904174e-05,
+      "loss": 0.6282,
+      "step": 3581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208874637589926e-05,
+      "loss": 0.5892,
+      "step": 3582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208332893486556e-05,
+      "loss": 0.6192,
+      "step": 3583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207790971604526e-05,
+      "loss": 0.6132,
+      "step": 3584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92072488719543e-05,
+      "loss": 0.6022,
+      "step": 3585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206706594546337e-05,
+      "loss": 0.5982,
+      "step": 3586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206164139391114e-05,
+      "loss": 0.6254,
+      "step": 3587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205621506499097e-05,
+      "loss": 0.3091,
+      "step": 3588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205078695880768e-05,
+      "loss": 0.6279,
+      "step": 3589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204535707546602e-05,
+      "loss": 0.6219,
+      "step": 3590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920399254150709e-05,
+      "loss": 0.6383,
+      "step": 3591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9203449197772707e-05,
+      "loss": 0.6211,
+      "step": 3592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920290567635395e-05,
+      "loss": 0.598,
+      "step": 3593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202361977261313e-05,
+      "loss": 0.5992,
+      "step": 3594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920181810050529e-05,
+      "loss": 0.5923,
+      "step": 3595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201274046096384e-05,
+      "loss": 0.6313,
+      "step": 3596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200729814045097e-05,
+      "loss": 0.6013,
+      "step": 3597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920018540436193e-05,
+      "loss": 0.5923,
+      "step": 3598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199640817057404e-05,
+      "loss": 0.6314,
+      "step": 3599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919909605214203e-05,
+      "loss": 0.6118,
+      "step": 3600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198551109626323e-05,
+      "loss": 0.6423,
+      "step": 3601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198005989520804e-05,
+      "loss": 0.6409,
+      "step": 3602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197460691835997e-05,
+      "loss": 0.6091,
+      "step": 3603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196915216582432e-05,
+      "loss": 0.6086,
+      "step": 3604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919636956377064e-05,
+      "loss": 0.5947,
+      "step": 3605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919582373341115e-05,
+      "loss": 0.5849,
+      "step": 3606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919527772551451e-05,
+      "loss": 0.651,
+      "step": 3607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194731540091252e-05,
+      "loss": 0.5961,
+      "step": 3608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194185177151925e-05,
+      "loss": 0.5912,
+      "step": 3609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193638636707076e-05,
+      "loss": 0.5919,
+      "step": 3610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9193091918767256e-05,
+      "loss": 0.6722,
+      "step": 3611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192545023343025e-05,
+      "loss": 0.6078,
+      "step": 3612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9191997950444933e-05,
+      "loss": 0.6092,
+      "step": 3613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919145070008355e-05,
+      "loss": 0.5966,
+      "step": 3614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190903272269438e-05,
+      "loss": 0.6471,
+      "step": 3615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190355667013166e-05,
+      "loss": 0.6757,
+      "step": 3616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918980788432531e-05,
+      "loss": 0.6109,
+      "step": 3617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189259924216433e-05,
+      "loss": 0.6197,
+      "step": 3618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188711786697132e-05,
+      "loss": 0.6134,
+      "step": 3619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188163471777974e-05,
+      "loss": 0.6263,
+      "step": 3620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187614979469555e-05,
+      "loss": 0.5851,
+      "step": 3621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918706630978246e-05,
+      "loss": 0.6104,
+      "step": 3622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186517462727282e-05,
+      "loss": 0.6059,
+      "step": 3623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918596843831462e-05,
+      "loss": 0.6183,
+      "step": 3624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918541923655507e-05,
+      "loss": 0.6145,
+      "step": 3625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184869857459233e-05,
+      "loss": 0.592,
+      "step": 3626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918432030103772e-05,
+      "loss": 0.5858,
+      "step": 3627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183770567301142e-05,
+      "loss": 0.6069,
+      "step": 3628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918322065626011e-05,
+      "loss": 0.6121,
+      "step": 3629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182670567925237e-05,
+      "loss": 0.6065,
+      "step": 3630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918212030230715e-05,
+      "loss": 0.5847,
+      "step": 3631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918156985941647e-05,
+      "loss": 0.6155,
+      "step": 3632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181019239263823e-05,
+      "loss": 0.5978,
+      "step": 3633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918046844185984e-05,
+      "loss": 0.606,
+      "step": 3634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179917467215153e-05,
+      "loss": 0.5929,
+      "step": 3635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91793663153404e-05,
+      "loss": 0.6179,
+      "step": 3636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178814986246226e-05,
+      "loss": 0.6166,
+      "step": 3637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178263479943272e-05,
+      "loss": 0.6046,
+      "step": 3638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177711796442184e-05,
+      "loss": 0.6068,
+      "step": 3639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177159935753612e-05,
+      "loss": 0.6433,
+      "step": 3640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176607897888217e-05,
+      "loss": 0.6018,
+      "step": 3641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917605568285665e-05,
+      "loss": 0.6363,
+      "step": 3642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175503290669576e-05,
+      "loss": 0.5991,
+      "step": 3643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917495072133766e-05,
+      "loss": 0.6081,
+      "step": 3644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174397974871563e-05,
+      "loss": 0.6349,
+      "step": 3645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173845051281966e-05,
+      "loss": 0.6365,
+      "step": 3646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917329195057954e-05,
+      "loss": 0.6563,
+      "step": 3647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917273867277496e-05,
+      "loss": 0.6267,
+      "step": 3648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172185217878912e-05,
+      "loss": 0.6157,
+      "step": 3649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171631585902084e-05,
+      "loss": 0.5953,
+      "step": 3650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171077776855157e-05,
+      "loss": 0.5969,
+      "step": 3651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170523790748826e-05,
+      "loss": 0.6136,
+      "step": 3652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169969627593788e-05,
+      "loss": 0.5857,
+      "step": 3653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169415287400743e-05,
+      "loss": 0.5915,
+      "step": 3654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916886077018039e-05,
+      "loss": 0.6516,
+      "step": 3655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168306075943436e-05,
+      "loss": 0.5901,
+      "step": 3656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167751204700588e-05,
+      "loss": 0.6439,
+      "step": 3657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916719615646256e-05,
+      "loss": 0.6321,
+      "step": 3658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166640931240067e-05,
+      "loss": 0.616,
+      "step": 3659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166085529043834e-05,
+      "loss": 0.3141,
+      "step": 3660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165529949884578e-05,
+      "loss": 0.6266,
+      "step": 3661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164974193773026e-05,
+      "loss": 0.6001,
+      "step": 3662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164418260719907e-05,
+      "loss": 0.6368,
+      "step": 3663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163862150735958e-05,
+      "loss": 0.6194,
+      "step": 3664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163305863831908e-05,
+      "loss": 0.6584,
+      "step": 3665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162749400018502e-05,
+      "loss": 0.6142,
+      "step": 3666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162192759306486e-05,
+      "loss": 0.5902,
+      "step": 3667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91616359417066e-05,
+      "loss": 0.645,
+      "step": 3668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91610789472296e-05,
+      "loss": 0.5766,
+      "step": 3669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160521775886237e-05,
+      "loss": 0.6589,
+      "step": 3670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159964427687265e-05,
+      "loss": 0.6728,
+      "step": 3671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915940690264345e-05,
+      "loss": 0.6467,
+      "step": 3672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915884920076555e-05,
+      "loss": 0.6128,
+      "step": 3673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158291322064334e-05,
+      "loss": 0.6101,
+      "step": 3674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157733266550577e-05,
+      "loss": 0.6216,
+      "step": 3675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157175034235047e-05,
+      "loss": 0.5962,
+      "step": 3676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156616625128522e-05,
+      "loss": 0.5918,
+      "step": 3677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915605803924179e-05,
+      "loss": 0.625,
+      "step": 3678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155499276585626e-05,
+      "loss": 0.579,
+      "step": 3679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154940337170824e-05,
+      "loss": 0.6211,
+      "step": 3680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915438122100817e-05,
+      "loss": 0.2814,
+      "step": 3681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915382192810846e-05,
+      "loss": 0.6319,
+      "step": 3682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153262458482496e-05,
+      "loss": 0.6503,
+      "step": 3683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152702812141078e-05,
+      "loss": 0.5789,
+      "step": 3684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152142989095007e-05,
+      "loss": 0.6122,
+      "step": 3685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915158298935509e-05,
+      "loss": 0.6257,
+      "step": 3686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151022812932143e-05,
+      "loss": 0.6306,
+      "step": 3687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915046245983698e-05,
+      "loss": 0.6196,
+      "step": 3688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914990193008042e-05,
+      "loss": 0.6337,
+      "step": 3689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149341223673282e-05,
+      "loss": 0.6627,
+      "step": 3690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914878034062639e-05,
+      "loss": 0.6132,
+      "step": 3691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914821928095058e-05,
+      "loss": 0.642,
+      "step": 3692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147658044656676e-05,
+      "loss": 0.5986,
+      "step": 3693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147096631755516e-05,
+      "loss": 0.6523,
+      "step": 3694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914653504225794e-05,
+      "loss": 0.6207,
+      "step": 3695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145973276174793e-05,
+      "loss": 0.6296,
+      "step": 3696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145411333516914e-05,
+      "loss": 0.5961,
+      "step": 3697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144849214295152e-05,
+      "loss": 0.6084,
+      "step": 3698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144286918520367e-05,
+      "loss": 0.6247,
+      "step": 3699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143724446203407e-05,
+      "loss": 0.6212,
+      "step": 3700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143161797355142e-05,
+      "loss": 0.6148,
+      "step": 3701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914259897198642e-05,
+      "loss": 0.537,
+      "step": 3702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914203597010812e-05,
+      "loss": 0.6266,
+      "step": 3703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91414727917311e-05,
+      "loss": 0.6185,
+      "step": 3704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914090943686624e-05,
+      "loss": 0.6119,
+      "step": 3705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140345905524418e-05,
+      "loss": 0.5883,
+      "step": 3706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139782197716508e-05,
+      "loss": 0.6856,
+      "step": 3707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139218313453396e-05,
+      "loss": 0.6019,
+      "step": 3708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913865425274597e-05,
+      "loss": 0.6176,
+      "step": 3709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138090015605117e-05,
+      "loss": 0.622,
+      "step": 3710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137525602041735e-05,
+      "loss": 0.6292,
+      "step": 3711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136961012066713e-05,
+      "loss": 0.6283,
+      "step": 3712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136396245690957e-05,
+      "loss": 0.6224,
+      "step": 3713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913583130292537e-05,
+      "loss": 0.6055,
+      "step": 3714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135266183780854e-05,
+      "loss": 0.6159,
+      "step": 3715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134700888268325e-05,
+      "loss": 0.6534,
+      "step": 3716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91341354163987e-05,
+      "loss": 0.6277,
+      "step": 3717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133569768182885e-05,
+      "loss": 0.2939,
+      "step": 3718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913300394363181e-05,
+      "loss": 0.5929,
+      "step": 3719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132437942756393e-05,
+      "loss": 0.6064,
+      "step": 3720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131871765567568e-05,
+      "loss": 0.6083,
+      "step": 3721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131305412076258e-05,
+      "loss": 0.6649,
+      "step": 3722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91307388822934e-05,
+      "loss": 0.5772,
+      "step": 3723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130172176229937e-05,
+      "loss": 0.6728,
+      "step": 3724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91296052938968e-05,
+      "loss": 0.6556,
+      "step": 3725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129038235304946e-05,
+      "loss": 0.6501,
+      "step": 3726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128471000465315e-05,
+      "loss": 0.6261,
+      "step": 3727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127903589388854e-05,
+      "loss": 0.6209,
+      "step": 3728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127336002086524e-05,
+      "loss": 0.283,
+      "step": 3729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126768238569283e-05,
+      "loss": 0.6103,
+      "step": 3730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912620029884809e-05,
+      "loss": 0.6163,
+      "step": 3731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125632182933912e-05,
+      "loss": 0.6158,
+      "step": 3732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125063890837717e-05,
+      "loss": 0.6538,
+      "step": 3733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124495422570476e-05,
+      "loss": 0.5934,
+      "step": 3734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912392677814316e-05,
+      "loss": 0.6233,
+      "step": 3735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123357957566752e-05,
+      "loss": 0.603,
+      "step": 3736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122788960852234e-05,
+      "loss": 0.6231,
+      "step": 3737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122219788010587e-05,
+      "loss": 0.614,
+      "step": 3738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121650439052806e-05,
+      "loss": 0.5781,
+      "step": 3739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912108091398988e-05,
+      "loss": 0.5833,
+      "step": 3740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91205112128328e-05,
+      "loss": 0.6255,
+      "step": 3741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119941335592572e-05,
+      "loss": 0.6262,
+      "step": 3742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119371282280197e-05,
+      "loss": 0.6251,
+      "step": 3743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118801052906674e-05,
+      "loss": 0.6297,
+      "step": 3744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911823064748302e-05,
+      "loss": 0.5857,
+      "step": 3745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117660066020244e-05,
+      "loss": 0.6454,
+      "step": 3746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117089308529363e-05,
+      "loss": 0.6009,
+      "step": 3747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116518375021393e-05,
+      "loss": 0.6457,
+      "step": 3748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115947265507356e-05,
+      "loss": 0.6179,
+      "step": 3749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115375979998284e-05,
+      "loss": 0.6191,
+      "step": 3750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114804518505204e-05,
+      "loss": 0.6183,
+      "step": 3751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114232881039147e-05,
+      "loss": 0.6429,
+      "step": 3752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911366106761115e-05,
+      "loss": 0.5809,
+      "step": 3753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113089078232253e-05,
+      "loss": 0.6142,
+      "step": 3754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112516912913497e-05,
+      "loss": 0.6402,
+      "step": 3755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111944571665934e-05,
+      "loss": 0.6113,
+      "step": 3756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911137205450061e-05,
+      "loss": 0.6395,
+      "step": 3757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110799361428573e-05,
+      "loss": 0.6395,
+      "step": 3758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110226492460886e-05,
+      "loss": 0.6086,
+      "step": 3759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109653447608607e-05,
+      "loss": 0.6051,
+      "step": 3760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109080226882805e-05,
+      "loss": 0.6182,
+      "step": 3761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108506830294534e-05,
+      "loss": 0.6352,
+      "step": 3762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107933257854875e-05,
+      "loss": 0.6584,
+      "step": 3763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91073595095749e-05,
+      "loss": 0.5944,
+      "step": 3764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106785585465687e-05,
+      "loss": 0.5681,
+      "step": 3765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910621148553831e-05,
+      "loss": 0.6461,
+      "step": 3766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105637209803854e-05,
+      "loss": 0.6513,
+      "step": 3767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105062758273416e-05,
+      "loss": 0.6209,
+      "step": 3768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910448813095807e-05,
+      "loss": 0.6353,
+      "step": 3769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103913327868924e-05,
+      "loss": 0.6511,
+      "step": 3770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910333834901707e-05,
+      "loss": 0.6074,
+      "step": 3771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102763194413606e-05,
+      "loss": 0.638,
+      "step": 3772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102187864069645e-05,
+      "loss": 0.5868,
+      "step": 3773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101612357996283e-05,
+      "loss": 0.6248,
+      "step": 3774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910103667620464e-05,
+      "loss": 0.6101,
+      "step": 3775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100460818705826e-05,
+      "loss": 0.6362,
+      "step": 3776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909988478551096e-05,
+      "loss": 0.5846,
+      "step": 3777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909930857663116e-05,
+      "loss": 0.6484,
+      "step": 3778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098732192077557e-05,
+      "loss": 0.6132,
+      "step": 3779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098155631861272e-05,
+      "loss": 0.5778,
+      "step": 3780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909757889599344e-05,
+      "loss": 0.5724,
+      "step": 3781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909700198448519e-05,
+      "loss": 0.6043,
+      "step": 3782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909642489734767e-05,
+      "loss": 0.6013,
+      "step": 3783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095847634592015e-05,
+      "loss": 0.62,
+      "step": 3784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909527019622937e-05,
+      "loss": 0.6413,
+      "step": 3785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094692582270885e-05,
+      "loss": 0.6259,
+      "step": 3786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909411479272771e-05,
+      "loss": 0.5744,
+      "step": 3787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093536827611e-05,
+      "loss": 0.6206,
+      "step": 3788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092958686931916e-05,
+      "loss": 0.6124,
+      "step": 3789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092380370701617e-05,
+      "loss": 0.5954,
+      "step": 3790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091801878931268e-05,
+      "loss": 0.6033,
+      "step": 3791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091223211632036e-05,
+      "loss": 0.6535,
+      "step": 3792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090644368815096e-05,
+      "loss": 0.6273,
+      "step": 3793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909006535049163e-05,
+      "loss": 0.658,
+      "step": 3794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90894861566728e-05,
+      "loss": 0.6422,
+      "step": 3795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088906787369802e-05,
+      "loss": 0.6154,
+      "step": 3796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088327242593813e-05,
+      "loss": 0.6378,
+      "step": 3797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087747522356032e-05,
+      "loss": 0.6004,
+      "step": 3798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087167626667638e-05,
+      "loss": 0.5911,
+      "step": 3799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908658755553984e-05,
+      "loss": 0.6202,
+      "step": 3800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086007308983825e-05,
+      "loss": 0.6068,
+      "step": 3801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90854268870108e-05,
+      "loss": 0.6589,
+      "step": 3802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084846289631976e-05,
+      "loss": 0.6393,
+      "step": 3803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084265516858556e-05,
+      "loss": 0.6326,
+      "step": 3804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083684568701757e-05,
+      "loss": 0.593,
+      "step": 3805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908310344517279e-05,
+      "loss": 0.6444,
+      "step": 3806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082522146282877e-05,
+      "loss": 0.5839,
+      "step": 3807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908194067204324e-05,
+      "loss": 0.6251,
+      "step": 3808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081359022465107e-05,
+      "loss": 0.6242,
+      "step": 3809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080777197559707e-05,
+      "loss": 0.6396,
+      "step": 3810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908019519733827e-05,
+      "loss": 0.5928,
+      "step": 3811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079613021812032e-05,
+      "loss": 0.5935,
+      "step": 3812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079030670992237e-05,
+      "loss": 0.6088,
+      "step": 3813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078448144890127e-05,
+      "loss": 0.5844,
+      "step": 3814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077865443516948e-05,
+      "loss": 0.6294,
+      "step": 3815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077282566883947e-05,
+      "loss": 0.6069,
+      "step": 3816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076699515002382e-05,
+      "loss": 0.6449,
+      "step": 3817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076116287883502e-05,
+      "loss": 0.6622,
+      "step": 3818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075532885538575e-05,
+      "loss": 0.658,
+      "step": 3819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074949307978862e-05,
+      "loss": 0.6462,
+      "step": 3820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074365555215625e-05,
+      "loss": 0.6036,
+      "step": 3821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907378162726014e-05,
+      "loss": 0.6258,
+      "step": 3822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907319752412368e-05,
+      "loss": 0.6441,
+      "step": 3823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907261324581752e-05,
+      "loss": 0.5717,
+      "step": 3824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072028792352938e-05,
+      "loss": 0.6038,
+      "step": 3825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9071444163741222e-05,
+      "loss": 0.5971,
+      "step": 3826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070859359993656e-05,
+      "loss": 0.6335,
+      "step": 3827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907027438112153e-05,
+      "loss": 0.6281,
+      "step": 3828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906968922713614e-05,
+      "loss": 0.6044,
+      "step": 3829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069103898048788e-05,
+      "loss": 0.6203,
+      "step": 3830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9068518393870762e-05,
+      "loss": 0.5972,
+      "step": 3831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067932714613375e-05,
+      "loss": 0.6171,
+      "step": 3832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067346860287934e-05,
+      "loss": 0.5693,
+      "step": 3833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066760830905747e-05,
+      "loss": 0.6265,
+      "step": 3834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906617462647813e-05,
+      "loss": 0.6328,
+      "step": 3835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065588247016396e-05,
+      "loss": 0.618,
+      "step": 3836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906500169253187e-05,
+      "loss": 0.6367,
+      "step": 3837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064414963035876e-05,
+      "loss": 0.6243,
+      "step": 3838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906382805853974e-05,
+      "loss": 0.6229,
+      "step": 3839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063240979054796e-05,
+      "loss": 0.5961,
+      "step": 3840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062653724592374e-05,
+      "loss": 0.6373,
+      "step": 3841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906206629516381e-05,
+      "loss": 0.6566,
+      "step": 3842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061478690780454e-05,
+      "loss": 0.5917,
+      "step": 3843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060890911453647e-05,
+      "loss": 0.6225,
+      "step": 3844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060302957194732e-05,
+      "loss": 0.6339,
+      "step": 3845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9059714828015067e-05,
+      "loss": 0.6359,
+      "step": 3846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9059126523926e-05,
+      "loss": 0.6151,
+      "step": 3847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905853804493889e-05,
+      "loss": 0.6516,
+      "step": 3848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057949391065103e-05,
+      "loss": 0.6388,
+      "step": 3849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057360562315997e-05,
+      "loss": 0.5809,
+      "step": 3850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905677155870295e-05,
+      "loss": 0.6098,
+      "step": 3851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056182380237325e-05,
+      "loss": 0.5967,
+      "step": 3852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055593026930496e-05,
+      "loss": 0.6312,
+      "step": 3853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055003498793848e-05,
+      "loss": 0.636,
+      "step": 3854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905441379583876e-05,
+      "loss": 0.5929,
+      "step": 3855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053823918076616e-05,
+      "loss": 0.6232,
+      "step": 3856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90532338655188e-05,
+      "loss": 0.5921,
+      "step": 3857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052643638176715e-05,
+      "loss": 0.6266,
+      "step": 3858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052053236061745e-05,
+      "loss": 0.6493,
+      "step": 3859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051462659185293e-05,
+      "loss": 0.5955,
+      "step": 3860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050871907558762e-05,
+      "loss": 0.6312,
+      "step": 3861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050280981193555e-05,
+      "loss": 0.6259,
+      "step": 3862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049689880101085e-05,
+      "loss": 0.6264,
+      "step": 3863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049098604292754e-05,
+      "loss": 0.602,
+      "step": 3864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904850715377999e-05,
+      "loss": 0.6234,
+      "step": 3865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047915528574203e-05,
+      "loss": 0.5936,
+      "step": 3866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047323728686814e-05,
+      "loss": 0.6139,
+      "step": 3867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046731754129255e-05,
+      "loss": 0.6398,
+      "step": 3868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046139604912954e-05,
+      "loss": 0.631,
+      "step": 3869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045547281049338e-05,
+      "loss": 0.5865,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044954782549846e-05,
+      "loss": 0.621,
+      "step": 3871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904436210942592e-05,
+      "loss": 0.603,
+      "step": 3872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043769261688997e-05,
+      "loss": 0.6411,
+      "step": 3873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904317623935052e-05,
+      "loss": 0.6194,
+      "step": 3874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042583042421947e-05,
+      "loss": 0.6231,
+      "step": 3875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041989670914724e-05,
+      "loss": 0.5943,
+      "step": 3876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904139612484031e-05,
+      "loss": 0.5888,
+      "step": 3877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040802404210165e-05,
+      "loss": 0.5961,
+      "step": 3878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040208509035745e-05,
+      "loss": 0.5966,
+      "step": 3879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039614439328524e-05,
+      "loss": 0.6076,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039020195099965e-05,
+      "loss": 0.6524,
+      "step": 3881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038425776361545e-05,
+      "loss": 0.6207,
+      "step": 3882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903783118312474e-05,
+      "loss": 0.6336,
+      "step": 3883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9037236415401027e-05,
+      "loss": 0.645,
+      "step": 3884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036641473201887e-05,
+      "loss": 0.6285,
+      "step": 3885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903604635653881e-05,
+      "loss": 0.605,
+      "step": 3886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035451065423283e-05,
+      "loss": 0.6041,
+      "step": 3887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034855599866802e-05,
+      "loss": 0.6178,
+      "step": 3888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903425995988086e-05,
+      "loss": 0.6284,
+      "step": 3889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903366414547696e-05,
+      "loss": 0.6221,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033068156666602e-05,
+      "loss": 0.6377,
+      "step": 3891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903247199346129e-05,
+      "loss": 0.6002,
+      "step": 3892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903187565587254e-05,
+      "loss": 0.5832,
+      "step": 3893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903127914391186e-05,
+      "loss": 0.5968,
+      "step": 3894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903068245759077e-05,
+      "loss": 0.5993,
+      "step": 3895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9030085596920786e-05,
+      "loss": 0.6187,
+      "step": 3896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029488561913433e-05,
+      "loss": 0.5949,
+      "step": 3897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902889135258024e-05,
+      "loss": 0.6005,
+      "step": 3898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902829396893273e-05,
+      "loss": 0.6157,
+      "step": 3899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902769641098245e-05,
+      "loss": 0.5973,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027098678740922e-05,
+      "loss": 0.6354,
+      "step": 3901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902650077221969e-05,
+      "loss": 0.6524,
+      "step": 3902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025902691430297e-05,
+      "loss": 0.6094,
+      "step": 3903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025304436384296e-05,
+      "loss": 0.6248,
+      "step": 3904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024706007093234e-05,
+      "loss": 0.6209,
+      "step": 3905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902410740356866e-05,
+      "loss": 0.615,
+      "step": 3906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023508625822135e-05,
+      "loss": 0.5784,
+      "step": 3907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022909673865215e-05,
+      "loss": 0.6045,
+      "step": 3908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902231054770947e-05,
+      "loss": 0.6012,
+      "step": 3909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021711247366463e-05,
+      "loss": 0.6103,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021111772847765e-05,
+      "loss": 0.6508,
+      "step": 3911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902051212416495e-05,
+      "loss": 0.6271,
+      "step": 3912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019912301329593e-05,
+      "loss": 0.6192,
+      "step": 3913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019312304353272e-05,
+      "loss": 0.6207,
+      "step": 3914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018712133247577e-05,
+      "loss": 0.6261,
+      "step": 3915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90181117880241e-05,
+      "loss": 0.5647,
+      "step": 3916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901751126869441e-05,
+      "loss": 0.6096,
+      "step": 3917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016910575270125e-05,
+      "loss": 0.6828,
+      "step": 3918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901630970776283e-05,
+      "loss": 0.5924,
+      "step": 3919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901570866618413e-05,
+      "loss": 0.6014,
+      "step": 3920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901510745054562e-05,
+      "loss": 0.6022,
+      "step": 3921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014506060858916e-05,
+      "loss": 0.5725,
+      "step": 3922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9013904497135626e-05,
+      "loss": 0.6043,
+      "step": 3923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9013302759387367e-05,
+      "loss": 0.604,
+      "step": 3924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901270084762575e-05,
+      "loss": 0.6024,
+      "step": 3925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90120987618624e-05,
+      "loss": 0.5738,
+      "step": 3926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011496502108942e-05,
+      "loss": 0.6017,
+      "step": 3927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010894068377e-05,
+      "loss": 0.6186,
+      "step": 3928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010291460678206e-05,
+      "loss": 0.6505,
+      "step": 3929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.6038,
+      "step": 3930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90090857234266e-05,
+      "loss": 0.575,
+      "step": 3931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900848259389707e-05,
+      "loss": 0.6141,
+      "step": 3932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007879290447242e-05,
+      "loss": 0.5878,
+      "step": 3933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900727581308877e-05,
+      "loss": 0.6247,
+      "step": 3934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90066721618333e-05,
+      "loss": 0.6131,
+      "step": 3935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006068336692486e-05,
+      "loss": 0.6172,
+      "step": 3936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005464337677986e-05,
+      "loss": 0.5778,
+      "step": 3937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004860164801467e-05,
+      "loss": 0.6523,
+      "step": 3938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004255818074585e-05,
+      "loss": 0.6339,
+      "step": 3939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900365129750901e-05,
+      "loss": 0.6031,
+      "step": 3940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003046603116417e-05,
+      "loss": 0.5948,
+      "step": 3941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900244173490848e-05,
+      "loss": 0.589,
+      "step": 3942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001836692896868e-05,
+      "loss": 0.6186,
+      "step": 3943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001231477093274e-05,
+      "loss": 0.5835,
+      "step": 3944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000626087509376e-05,
+      "loss": 0.6025,
+      "step": 3945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900002052415686e-05,
+      "loss": 0.6766,
+      "step": 3946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899941478704742e-05,
+      "loss": 0.6329,
+      "step": 3947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899880887619275e-05,
+      "loss": 0.618,
+      "step": 3948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998202791604554e-05,
+      "loss": 0.6432,
+      "step": 3949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8997596533294524e-05,
+      "loss": 0.6442,
+      "step": 3950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996990101274367e-05,
+      "loss": 0.6397,
+      "step": 3951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996383495555792e-05,
+      "loss": 0.595,
+      "step": 3952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899577671615051e-05,
+      "loss": 0.6412,
+      "step": 3953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899516976307024e-05,
+      "loss": 0.5867,
+      "step": 3954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994562636326694e-05,
+      "loss": 0.6018,
+      "step": 3955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993955335931595e-05,
+      "loss": 0.6192,
+      "step": 3956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899334786189667e-05,
+      "loss": 0.6061,
+      "step": 3957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899274021423364e-05,
+      "loss": 0.6179,
+      "step": 3958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899213239295425e-05,
+      "loss": 0.6017,
+      "step": 3959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899152439807022e-05,
+      "loss": 0.6324,
+      "step": 3960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8990916229593293e-05,
+      "loss": 0.6178,
+      "step": 3961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8990307887535216e-05,
+      "loss": 0.6301,
+      "step": 3962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898969937190773e-05,
+      "loss": 0.5987,
+      "step": 3963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989090682722583e-05,
+      "loss": 0.6297,
+      "step": 3964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988481819991526e-05,
+      "loss": 0.5999,
+      "step": 3965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987872783726313e-05,
+      "loss": 0.6457,
+      "step": 3966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987263573938706e-05,
+      "loss": 0.6185,
+      "step": 3967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986654190640462e-05,
+      "loss": 0.6138,
+      "step": 3968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898604463384335e-05,
+      "loss": 0.6042,
+      "step": 3969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985434903559138e-05,
+      "loss": 0.6366,
+      "step": 3970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8984824999799592e-05,
+      "loss": 0.5819,
+      "step": 3971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8984214922576494e-05,
+      "loss": 0.6331,
+      "step": 3972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983604671901618e-05,
+      "loss": 0.5749,
+      "step": 3973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982994247786747e-05,
+      "loss": 0.6587,
+      "step": 3974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982383650243666e-05,
+      "loss": 0.571,
+      "step": 3975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981772879284162e-05,
+      "loss": 0.579,
+      "step": 3976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898116193492003e-05,
+      "loss": 0.6295,
+      "step": 3977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898055081716306e-05,
+      "loss": 0.6307,
+      "step": 3978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897993952602506e-05,
+      "loss": 0.6191,
+      "step": 3979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979328061517817e-05,
+      "loss": 0.6506,
+      "step": 3980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978716423653153e-05,
+      "loss": 0.6238,
+      "step": 3981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897810461244286e-05,
+      "loss": 0.693,
+      "step": 3982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977492627898765e-05,
+      "loss": 0.5838,
+      "step": 3983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976880470032673e-05,
+      "loss": 0.6127,
+      "step": 3984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976268138856404e-05,
+      "loss": 0.5883,
+      "step": 3985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897565563438178e-05,
+      "loss": 0.6282,
+      "step": 3986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975042956620632e-05,
+      "loss": 0.6077,
+      "step": 3987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897443010558478e-05,
+      "loss": 0.6188,
+      "step": 3988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897381708128606e-05,
+      "loss": 0.6393,
+      "step": 3989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897320388373631e-05,
+      "loss": 0.6329,
+      "step": 3990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972590512947362e-05,
+      "loss": 0.6409,
+      "step": 3991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971976968931066e-05,
+      "loss": 0.6055,
+      "step": 3992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897136325169926e-05,
+      "loss": 0.6059,
+      "step": 3993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970749361263792e-05,
+      "loss": 0.6723,
+      "step": 3994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897013529763652e-05,
+      "loss": 0.6347,
+      "step": 3995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8969521060829298e-05,
+      "loss": 0.5676,
+      "step": 3996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896890665085398e-05,
+      "loss": 0.5748,
+      "step": 3997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968292067722433e-05,
+      "loss": 0.644,
+      "step": 3998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967677311446522e-05,
+      "loss": 0.6084,
+      "step": 3999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967062382038113e-05,
+      "loss": 0.6438,
+      "step": 4000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896644727950908e-05,
+      "loss": 0.5994,
+      "step": 4001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965832003871294e-05,
+      "loss": 0.62,
+      "step": 4002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965216555136642e-05,
+      "loss": 0.6088,
+      "step": 4003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8964600933316997e-05,
+      "loss": 0.5998,
+      "step": 4004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896398513842425e-05,
+      "loss": 0.5809,
+      "step": 4005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896336917047029e-05,
+      "loss": 0.617,
+      "step": 4006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962753029467005e-05,
+      "loss": 0.6439,
+      "step": 4007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962136715426292e-05,
+      "loss": 0.6212,
+      "step": 4008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896152022836005e-05,
+      "loss": 0.639,
+      "step": 4009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960903568280186e-05,
+      "loss": 0.5765,
+      "step": 4010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960286735198595e-05,
+      "loss": 0.6014,
+      "step": 4011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959669729127194e-05,
+      "loss": 0.6263,
+      "step": 4012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959052550077895e-05,
+      "loss": 0.6068,
+      "step": 4013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895843519806261e-05,
+      "loss": 0.6461,
+      "step": 4014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957817673093258e-05,
+      "loss": 0.6125,
+      "step": 4015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957199975181764e-05,
+      "loss": 0.6233,
+      "step": 4016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956582104340048e-05,
+      "loss": 0.6158,
+      "step": 4017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955964060580045e-05,
+      "loss": 0.5994,
+      "step": 4018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955345843913685e-05,
+      "loss": 0.6192,
+      "step": 4019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89547274543529e-05,
+      "loss": 0.5749,
+      "step": 4020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954108891909634e-05,
+      "loss": 0.6098,
+      "step": 4021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953490156595823e-05,
+      "loss": 0.6012,
+      "step": 4022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895287124842342e-05,
+      "loss": 0.5917,
+      "step": 4023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895225216740437e-05,
+      "loss": 0.6084,
+      "step": 4024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951632913550625e-05,
+      "loss": 0.5882,
+      "step": 4025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951013486874143e-05,
+      "loss": 0.6038,
+      "step": 4026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950393887386876e-05,
+      "loss": 0.6289,
+      "step": 4027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894977411510079e-05,
+      "loss": 0.6098,
+      "step": 4028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949154170027853e-05,
+      "loss": 0.6124,
+      "step": 4029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948534052180038e-05,
+      "loss": 0.6184,
+      "step": 4030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947913761569302e-05,
+      "loss": 0.6156,
+      "step": 4031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947293298207637e-05,
+      "loss": 0.5814,
+      "step": 4032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894667266210701e-05,
+      "loss": 0.6082,
+      "step": 4033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894605185327941e-05,
+      "loss": 0.641,
+      "step": 4034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945430871736818e-05,
+      "loss": 0.5782,
+      "step": 4035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944809717491228e-05,
+      "loss": 0.6227,
+      "step": 4036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944188390554628e-05,
+      "loss": 0.6152,
+      "step": 4037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943566890939015e-05,
+      "loss": 0.6491,
+      "step": 4038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894294521865639e-05,
+      "loss": 0.676,
+      "step": 4039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894232337371875e-05,
+      "loss": 0.5969,
+      "step": 4040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89417013561381e-05,
+      "loss": 0.6137,
+      "step": 4041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894107916592646e-05,
+      "loss": 0.6012,
+      "step": 4042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894045680309583e-05,
+      "loss": 0.6236,
+      "step": 4043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939834267658228e-05,
+      "loss": 0.6133,
+      "step": 4044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939211559625676e-05,
+      "loss": 0.613,
+      "step": 4045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938588679010195e-05,
+      "loss": 0.6144,
+      "step": 4046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937965625823813e-05,
+      "loss": 0.5915,
+      "step": 4047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937342400078554e-05,
+      "loss": 0.6217,
+      "step": 4048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936719001786453e-05,
+      "loss": 0.6076,
+      "step": 4049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936095430959545e-05,
+      "loss": 0.6286,
+      "step": 4050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893547168760987e-05,
+      "loss": 0.5933,
+      "step": 4051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934847771749468e-05,
+      "loss": 0.6237,
+      "step": 4052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934223683390382e-05,
+      "loss": 0.5869,
+      "step": 4053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933599422544668e-05,
+      "loss": 0.6154,
+      "step": 4054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932974989224374e-05,
+      "loss": 0.607,
+      "step": 4055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932350383441554e-05,
+      "loss": 0.656,
+      "step": 4056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931725605208272e-05,
+      "loss": 0.5975,
+      "step": 4057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931100654536584e-05,
+      "loss": 0.6085,
+      "step": 4058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893047553143856e-05,
+      "loss": 0.612,
+      "step": 4059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929850235926265e-05,
+      "loss": 0.6138,
+      "step": 4060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929224768011774e-05,
+      "loss": 0.6019,
+      "step": 4061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928599127707162e-05,
+      "loss": 0.6057,
+      "step": 4062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927973315024508e-05,
+      "loss": 0.6153,
+      "step": 4063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927347329975894e-05,
+      "loss": 0.606,
+      "step": 4064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926721172573405e-05,
+      "loss": 0.6843,
+      "step": 4065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926094842829128e-05,
+      "loss": 0.58,
+      "step": 4066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925468340755157e-05,
+      "loss": 0.6139,
+      "step": 4067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924841666363592e-05,
+      "loss": 0.6225,
+      "step": 4068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892421481966652e-05,
+      "loss": 0.6447,
+      "step": 4069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923587800676054e-05,
+      "loss": 0.5798,
+      "step": 4070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922960609404297e-05,
+      "loss": 0.6182,
+      "step": 4071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922333245863355e-05,
+      "loss": 0.5992,
+      "step": 4072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892170571006534e-05,
+      "loss": 0.614,
+      "step": 4073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892107800202237e-05,
+      "loss": 0.6265,
+      "step": 4074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920450121746562e-05,
+      "loss": 0.6155,
+      "step": 4075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891982206925004e-05,
+      "loss": 0.6012,
+      "step": 4076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919193844544924e-05,
+      "loss": 0.5928,
+      "step": 4077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891856544764335e-05,
+      "loss": 0.6213,
+      "step": 4078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891793687855745e-05,
+      "loss": 0.6126,
+      "step": 4079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891730813729935e-05,
+      "loss": 0.6016,
+      "step": 4080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916679223881193e-05,
+      "loss": 0.6048,
+      "step": 4081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916050138315127e-05,
+      "loss": 0.6259,
+      "step": 4082
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891542088061329e-05,
+      "loss": 0.6135,
+      "step": 4083
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914791450787837e-05,
+      "loss": 0.603,
+      "step": 4084
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914161848850913e-05,
+      "loss": 0.5987,
+      "step": 4085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891353207481468e-05,
+      "loss": 0.6012,
+      "step": 4086
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891290212869129e-05,
+      "loss": 0.6066,
+      "step": 4087
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891227201049291e-05,
+      "loss": 0.5852,
+      "step": 4088
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89116417202317e-05,
+      "loss": 0.6178,
+      "step": 4089
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911011257919834e-05,
+      "loss": 0.5824,
+      "step": 4090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891038062356948e-05,
+      "loss": 0.6323,
+      "step": 4091
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890974981719282e-05,
+      "loss": 0.609,
+      "step": 4092
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909118838802024e-05,
+      "loss": 0.6061,
+      "step": 4093
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8908487688409282e-05,
+      "loss": 0.6328,
+      "step": 4094
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890785636602677e-05,
+      "loss": 0.5999,
+      "step": 4095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907224871666685e-05,
+      "loss": 0.6371,
+      "step": 4096
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890659320534121e-05,
+      "loss": 0.6207,
+      "step": 4097
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905961367062546e-05,
+      "loss": 0.5852,
+      "step": 4098
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890532935684289e-05,
+      "loss": 0.6447,
+      "step": 4099
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904697174694447e-05,
+      "loss": 0.6331,
+      "step": 4100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904064820629418e-05,
+      "loss": 0.6201,
+      "step": 4101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890343229466001e-05,
+      "loss": 0.6162,
+      "step": 4102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890279959679844e-05,
+      "loss": 0.5987,
+      "step": 4103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902166727056924e-05,
+      "loss": 0.5871,
+      "step": 4104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901533685447672e-05,
+      "loss": 0.6117,
+      "step": 4105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900900471982913e-05,
+      "loss": 0.5879,
+      "step": 4106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900267086674868e-05,
+      "loss": 0.606,
+      "step": 4107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899633529535767e-05,
+      "loss": 0.6119,
+      "step": 4108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898999800577842e-05,
+      "loss": 0.6326,
+      "step": 4109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898365899813328e-05,
+      "loss": 0.5952,
+      "step": 4110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897731827254462e-05,
+      "loss": 0.596,
+      "step": 4111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897097582913486e-05,
+      "loss": 0.5983,
+      "step": 4112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896463166802644e-05,
+      "loss": 0.6133,
+      "step": 4113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895828578934188e-05,
+      "loss": 0.5616,
+      "step": 4114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895193819320366e-05,
+      "loss": 0.574,
+      "step": 4115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894558887973433e-05,
+      "loss": 0.6198,
+      "step": 4116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893923784905647e-05,
+      "loss": 0.6419,
+      "step": 4117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893288510129273e-05,
+      "loss": 0.6221,
+      "step": 4118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892653063656573e-05,
+      "loss": 0.5963,
+      "step": 4119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892017445499812e-05,
+      "loss": 0.6216,
+      "step": 4120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889138165567127e-05,
+      "loss": 0.6074,
+      "step": 4121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889074569418321e-05,
+      "loss": 0.5797,
+      "step": 4122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889010956104792e-05,
+      "loss": 0.6018,
+      "step": 4123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888947325627768e-05,
+      "loss": 0.6048,
+      "step": 4124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888883677988477e-05,
+      "loss": 0.5847,
+      "step": 4125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888820013188148e-05,
+      "loss": 0.6397,
+      "step": 4126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887563312280103e-05,
+      "loss": 0.5834,
+      "step": 4127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886926321092933e-05,
+      "loss": 0.6226,
+      "step": 4128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886289158332266e-05,
+      "loss": 0.58,
+      "step": 4129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885651824010406e-05,
+      "loss": 0.6091,
+      "step": 4130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885014318139654e-05,
+      "loss": 0.5725,
+      "step": 4131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884376640732323e-05,
+      "loss": 0.6089,
+      "step": 4132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883738791800718e-05,
+      "loss": 0.5865,
+      "step": 4133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888310077135716e-05,
+      "loss": 0.5944,
+      "step": 4134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882462579413962e-05,
+      "loss": 0.5991,
+      "step": 4135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881824215983446e-05,
+      "loss": 0.5928,
+      "step": 4136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888118568107794e-05,
+      "loss": 0.6409,
+      "step": 4137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880546974709763e-05,
+      "loss": 0.6143,
+      "step": 4138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879908096891256e-05,
+      "loss": 0.6045,
+      "step": 4139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887926904763475e-05,
+      "loss": 0.6082,
+      "step": 4140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887862982695258e-05,
+      "loss": 0.5998,
+      "step": 4141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887799043485709e-05,
+      "loss": 0.5749,
+      "step": 4142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887735087136062e-05,
+      "loss": 0.6349,
+      "step": 4143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876711136475526e-05,
+      "loss": 0.6295,
+      "step": 4144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887607123021415e-05,
+      "loss": 0.582,
+      "step": 4145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887543115258885e-05,
+      "loss": 0.655,
+      "step": 4146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874790903611987e-05,
+      "loss": 0.6068,
+      "step": 4147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887415048329591e-05,
+      "loss": 0.5785,
+      "step": 4148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873509891653e-05,
+      "loss": 0.5972,
+      "step": 4149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887286912869561e-05,
+      "loss": 0.5552,
+      "step": 4150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887222819443612e-05,
+      "loss": 0.647,
+      "step": 4151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88715870888869e-05,
+      "loss": 0.6041,
+      "step": 4152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870945812060325e-05,
+      "loss": 0.6027,
+      "step": 4153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870304363968782e-05,
+      "loss": 0.6396,
+      "step": 4154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886966274462465e-05,
+      "loss": 0.6108,
+      "step": 4155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886902095404032e-05,
+      "loss": 0.6197,
+      "step": 4156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8868378992228178e-05,
+      "loss": 0.6128,
+      "step": 4157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886773685920062e-05,
+      "loss": 0.6037,
+      "step": 4158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8867094554970046e-05,
+      "loss": 0.6742,
+      "step": 4159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886645207954885e-05,
+      "loss": 0.6267,
+      "step": 4160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865809432949445e-05,
+      "loss": 0.6255,
+      "step": 4161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886516661518423e-05,
+      "loss": 0.6138,
+      "step": 4162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886452362626562e-05,
+      "loss": 0.5884,
+      "step": 4163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863880466206022e-05,
+      "loss": 0.5945,
+      "step": 4164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863237135017865e-05,
+      "loss": 0.5901,
+      "step": 4165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886259363271356e-05,
+      "loss": 0.6451,
+      "step": 4166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861949959305532e-05,
+      "loss": 0.5924,
+      "step": 4167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886130611480621e-05,
+      "loss": 0.278,
+      "step": 4168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860662099228026e-05,
+      "loss": 0.6029,
+      "step": 4169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860017912583406e-05,
+      "loss": 0.5944,
+      "step": 4170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859373554884796e-05,
+      "loss": 0.6452,
+      "step": 4171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858729026144626e-05,
+      "loss": 0.5981,
+      "step": 4172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858084326375348e-05,
+      "loss": 0.5811,
+      "step": 4173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857439455589407e-05,
+      "loss": 0.6547,
+      "step": 4174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856794413799253e-05,
+      "loss": 0.6082,
+      "step": 4175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885614920101734e-05,
+      "loss": 0.6298,
+      "step": 4176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855503817256118e-05,
+      "loss": 0.626,
+      "step": 4177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854858262528057e-05,
+      "loss": 0.6147,
+      "step": 4178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854212536845612e-05,
+      "loss": 0.5936,
+      "step": 4179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853566640221253e-05,
+      "loss": 0.6302,
+      "step": 4180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885292057266745e-05,
+      "loss": 0.6086,
+      "step": 4181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852274334196678e-05,
+      "loss": 0.624,
+      "step": 4182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885162792482141e-05,
+      "loss": 0.6237,
+      "step": 4183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850981344554125e-05,
+      "loss": 0.6089,
+      "step": 4184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885033459340731e-05,
+      "loss": 0.6226,
+      "step": 4185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884968767139345e-05,
+      "loss": 0.6067,
+      "step": 4186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8849040578525036e-05,
+      "loss": 0.6278,
+      "step": 4187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884839331481456e-05,
+      "loss": 0.6121,
+      "step": 4188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884774588027451e-05,
+      "loss": 0.6417,
+      "step": 4189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847098274917397e-05,
+      "loss": 0.5981,
+      "step": 4190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846450498755723e-05,
+      "loss": 0.6206,
+      "step": 4191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845802551801986e-05,
+      "loss": 0.618,
+      "step": 4192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845154434068703e-05,
+      "loss": 0.6401,
+      "step": 4193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844506145568387e-05,
+      "loss": 0.6024,
+      "step": 4194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843857686313548e-05,
+      "loss": 0.5889,
+      "step": 4195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884320905631671e-05,
+      "loss": 0.5736,
+      "step": 4196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884256025559039e-05,
+      "loss": 0.6115,
+      "step": 4197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841911284147123e-05,
+      "loss": 0.5999,
+      "step": 4198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884126214199943e-05,
+      "loss": 0.614,
+      "step": 4199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884061282915985e-05,
+      "loss": 0.612,
+      "step": 4200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8839963345640916e-05,
+      "loss": 0.5939,
+      "step": 4201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8839313691455163e-05,
+      "loss": 0.6257,
+      "step": 4202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838663866615138e-05,
+      "loss": 0.6038,
+      "step": 4203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883801387113339e-05,
+      "loss": 0.5524,
+      "step": 4204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883736370502246e-05,
+      "loss": 0.6188,
+      "step": 4205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836713368294904e-05,
+      "loss": 0.614,
+      "step": 4206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883606286096328e-05,
+      "loss": 0.6084,
+      "step": 4207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8835412183040143e-05,
+      "loss": 0.6182,
+      "step": 4208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883476133453806e-05,
+      "loss": 0.6175,
+      "step": 4209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883411031546959e-05,
+      "loss": 0.6358,
+      "step": 4210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833459125847303e-05,
+      "loss": 0.6042,
+      "step": 4211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832807765683777e-05,
+      "loss": 0.5683,
+      "step": 4212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883215623499158e-05,
+      "loss": 0.5702,
+      "step": 4213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88315045337833e-05,
+      "loss": 0.6044,
+      "step": 4214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830852662071507e-05,
+      "loss": 0.6153,
+      "step": 4215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830200619868795e-05,
+      "loss": 0.6051,
+      "step": 4216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882954840718775e-05,
+      "loss": 0.6264,
+      "step": 4217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882889602404096e-05,
+      "loss": 0.6091,
+      "step": 4218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8828243470441026e-05,
+      "loss": 0.6093,
+      "step": 4219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882759074640055e-05,
+      "loss": 0.6055,
+      "step": 4220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882693785193212e-05,
+      "loss": 0.5995,
+      "step": 4221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882628478704835e-05,
+      "loss": 0.6189,
+      "step": 4222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825631551761846e-05,
+      "loss": 0.6262,
+      "step": 4223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824978146085225e-05,
+      "loss": 0.6134,
+      "step": 4224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824324570031094e-05,
+      "loss": 0.2859,
+      "step": 4225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882367082361207e-05,
+      "loss": 0.6424,
+      "step": 4226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823016906840788e-05,
+      "loss": 0.5696,
+      "step": 4227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822362819729857e-05,
+      "loss": 0.6368,
+      "step": 4228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8821708562291917e-05,
+      "loss": 0.62,
+      "step": 4229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882105413453959e-05,
+      "loss": 0.6162,
+      "step": 4230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820399536485513e-05,
+      "loss": 0.6176,
+      "step": 4231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819744768142327e-05,
+      "loss": 0.6242,
+      "step": 4232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881908982952267e-05,
+      "loss": 0.2735,
+      "step": 4233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8818434720639188e-05,
+      "loss": 0.6383,
+      "step": 4234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881777944150453e-05,
+      "loss": 0.6235,
+      "step": 4235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8817123992131344e-05,
+      "loss": 0.6519,
+      "step": 4236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816468372532288e-05,
+      "loss": 0.6109,
+      "step": 4237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8815812582720016e-05,
+      "loss": 0.6275,
+      "step": 4238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881515662270719e-05,
+      "loss": 0.6271,
+      "step": 4239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8814500492506475e-05,
+      "loss": 0.6156,
+      "step": 4240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881384419213054e-05,
+      "loss": 0.6136,
+      "step": 4241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881318772159205e-05,
+      "loss": 0.6052,
+      "step": 4242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881253108090368e-05,
+      "loss": 0.6239,
+      "step": 4243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811874270078113e-05,
+      "loss": 0.5589,
+      "step": 4244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881121728912803e-05,
+      "loss": 0.597,
+      "step": 4245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810560138066106e-05,
+      "loss": 0.6213,
+      "step": 4246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809902816905035e-05,
+      "loss": 0.5952,
+      "step": 4247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809245325657506e-05,
+      "loss": 0.5812,
+      "step": 4248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880858766433621e-05,
+      "loss": 0.6332,
+      "step": 4249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880792983295385e-05,
+      "loss": 0.6431,
+      "step": 4250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807271831523123e-05,
+      "loss": 0.5922,
+      "step": 4251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880661366005673e-05,
+      "loss": 0.5692,
+      "step": 4252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880595531856738e-05,
+      "loss": 0.6343,
+      "step": 4253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805296807067788e-05,
+      "loss": 0.5939,
+      "step": 4254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880463812557066e-05,
+      "loss": 0.608,
+      "step": 4255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803979274088712e-05,
+      "loss": 0.6469,
+      "step": 4256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803320252634673e-05,
+      "loss": 0.5792,
+      "step": 4257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802661061221253e-05,
+      "loss": 0.6315,
+      "step": 4258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802001699861193e-05,
+      "loss": 0.61,
+      "step": 4259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801342168567214e-05,
+      "loss": 0.6291,
+      "step": 4260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880068246735205e-05,
+      "loss": 0.602,
+      "step": 4261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880002259622844e-05,
+      "loss": 0.6309,
+      "step": 4262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799362555209122e-05,
+      "loss": 0.6342,
+      "step": 4263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798702344306837e-05,
+      "loss": 0.6123,
+      "step": 4264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798041963534337e-05,
+      "loss": 0.6005,
+      "step": 4265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797381412904366e-05,
+      "loss": 0.5959,
+      "step": 4266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879672069242968e-05,
+      "loss": 0.6042,
+      "step": 4267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879605980212303e-05,
+      "loss": 0.6501,
+      "step": 4268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879539874199718e-05,
+      "loss": 0.6145,
+      "step": 4269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 0.6125,
+      "step": 4270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794076112338932e-05,
+      "loss": 0.5954,
+      "step": 4271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879341454283207e-05,
+      "loss": 0.6153,
+      "step": 4272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792752803557077e-05,
+      "loss": 0.5994,
+      "step": 4273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792090894526723e-05,
+      "loss": 0.6066,
+      "step": 4274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791428815753797e-05,
+      "loss": 0.5988,
+      "step": 4275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790766567251077e-05,
+      "loss": 0.6057,
+      "step": 4276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879010414903135e-05,
+      "loss": 0.6386,
+      "step": 4277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789441561107403e-05,
+      "loss": 0.6193,
+      "step": 4278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788778803492027e-05,
+      "loss": 0.6507,
+      "step": 4279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788115876198018e-05,
+      "loss": 0.6541,
+      "step": 4280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787452779238176e-05,
+      "loss": 0.5776,
+      "step": 4281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786789512625306e-05,
+      "loss": 0.6447,
+      "step": 4282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786126076372208e-05,
+      "loss": 0.5927,
+      "step": 4283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785462470491692e-05,
+      "loss": 0.5923,
+      "step": 4284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878479869499657e-05,
+      "loss": 0.5827,
+      "step": 4285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784134749899656e-05,
+      "loss": 0.5899,
+      "step": 4286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878347063521377e-05,
+      "loss": 0.6104,
+      "step": 4287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782806350951732e-05,
+      "loss": 0.6117,
+      "step": 4288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878214189712637e-05,
+      "loss": 0.6312,
+      "step": 4289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781477273750508e-05,
+      "loss": 0.6288,
+      "step": 4290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878081248083698e-05,
+      "loss": 0.5774,
+      "step": 4291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780147518398623e-05,
+      "loss": 0.2747,
+      "step": 4292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779482386448268e-05,
+      "loss": 0.6199,
+      "step": 4293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877881708499876e-05,
+      "loss": 0.6362,
+      "step": 4294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8778151614062944e-05,
+      "loss": 0.6116,
+      "step": 4295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877748597365367e-05,
+      "loss": 0.6269,
+      "step": 4296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776820163783786e-05,
+      "loss": 0.6384,
+      "step": 4297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776154184466143e-05,
+      "loss": 0.6258,
+      "step": 4298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775488035713607e-05,
+      "loss": 0.6026,
+      "step": 4299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877482171753903e-05,
+      "loss": 0.5979,
+      "step": 4300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8774155229955284e-05,
+      "loss": 0.6489,
+      "step": 4301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877348857297523e-05,
+      "loss": 0.6103,
+      "step": 4302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877282174661174e-05,
+      "loss": 0.6348,
+      "step": 4303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8772154750877696e-05,
+      "loss": 0.6152,
+      "step": 4304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771487585785966e-05,
+      "loss": 0.6088,
+      "step": 4305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770820251349432e-05,
+      "loss": 0.5868,
+      "step": 4306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877015274758098e-05,
+      "loss": 0.5989,
+      "step": 4307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876948507449349e-05,
+      "loss": 0.6272,
+      "step": 4308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768817232099864e-05,
+      "loss": 0.6303,
+      "step": 4309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876814922041299e-05,
+      "loss": 0.582,
+      "step": 4310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8767481039445763e-05,
+      "loss": 0.6066,
+      "step": 4311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8766812689211083e-05,
+      "loss": 0.6105,
+      "step": 4312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766144169721855e-05,
+      "loss": 0.6344,
+      "step": 4313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765475480990986e-05,
+      "loss": 0.6175,
+      "step": 4314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764806623031388e-05,
+      "loss": 0.5865,
+      "step": 4315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876413759585597e-05,
+      "loss": 0.6003,
+      "step": 4316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763468399477648e-05,
+      "loss": 0.5799,
+      "step": 4317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762799033909344e-05,
+      "loss": 0.6326,
+      "step": 4318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762129499163984e-05,
+      "loss": 0.5907,
+      "step": 4319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876145979525449e-05,
+      "loss": 0.6273,
+      "step": 4320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876078992219379e-05,
+      "loss": 0.6231,
+      "step": 4321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876011987999482e-05,
+      "loss": 0.6311,
+      "step": 4322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759449668670515e-05,
+      "loss": 0.668,
+      "step": 4323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758779288233816e-05,
+      "loss": 0.6387,
+      "step": 4324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875810873869766e-05,
+      "loss": 0.6454,
+      "step": 4325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757438020075004e-05,
+      "loss": 0.636,
+      "step": 4326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756767132378788e-05,
+      "loss": 0.6295,
+      "step": 4327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875609607562196e-05,
+      "loss": 0.5947,
+      "step": 4328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755424849817486e-05,
+      "loss": 0.6196,
+      "step": 4329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875475345497832e-05,
+      "loss": 0.6291,
+      "step": 4330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754081891117426e-05,
+      "loss": 0.6098,
+      "step": 4331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875341015824777e-05,
+      "loss": 0.5987,
+      "step": 4332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8752738256382316e-05,
+      "loss": 0.6102,
+      "step": 4333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875206618553404e-05,
+      "loss": 0.2872,
+      "step": 4334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751393945715913e-05,
+      "loss": 0.6256,
+      "step": 4335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875072153694092e-05,
+      "loss": 0.6189,
+      "step": 4336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875004895922204e-05,
+      "loss": 0.6017,
+      "step": 4337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749376212572254e-05,
+      "loss": 0.611,
+      "step": 4338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8748703297004555e-05,
+      "loss": 0.606,
+      "step": 4339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8748030212531936e-05,
+      "loss": 0.6084,
+      "step": 4340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747356959167383e-05,
+      "loss": 0.6168,
+      "step": 4341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746683536923904e-05,
+      "loss": 0.63,
+      "step": 4342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746009945814493e-05,
+      "loss": 0.6109,
+      "step": 4343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874533618585216e-05,
+      "loss": 0.5904,
+      "step": 4344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8744662257049908e-05,
+      "loss": 0.6232,
+      "step": 4345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874398815942075e-05,
+      "loss": 0.6432,
+      "step": 4346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743313892977704e-05,
+      "loss": 0.6001,
+      "step": 4347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8742639457733785e-05,
+      "loss": 0.6347,
+      "step": 4348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874196485370201e-05,
+      "loss": 0.6278,
+      "step": 4349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741290080895406e-05,
+      "loss": 0.6097,
+      "step": 4350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740615139327002e-05,
+      "loss": 0.5901,
+      "step": 4351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739940029009827e-05,
+      "loss": 0.6241,
+      "step": 4352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739264749956914e-05,
+      "loss": 0.586,
+      "step": 4353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87385893021813e-05,
+      "loss": 0.6095,
+      "step": 4354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873791368569603e-05,
+      "loss": 0.6379,
+      "step": 4355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737237900514142e-05,
+      "loss": 0.6087,
+      "step": 4356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736561946648687e-05,
+      "loss": 0.6306,
+      "step": 4357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873588582411271e-05,
+      "loss": 0.6327,
+      "step": 4358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873520953291927e-05,
+      "loss": 0.5824,
+      "step": 4359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873453307308142e-05,
+      "loss": 0.6079,
+      "step": 4360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733856444612222e-05,
+      "loss": 0.6165,
+      "step": 4361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873317964752474e-05,
+      "loss": 0.6231,
+      "step": 4362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732502681832035e-05,
+      "loss": 0.5878,
+      "step": 4363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731825547547184e-05,
+      "loss": 0.6011,
+      "step": 4364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731148244683253e-05,
+      "loss": 0.6349,
+      "step": 4365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730470773253326e-05,
+      "loss": 0.6187,
+      "step": 4366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872979313327048e-05,
+      "loss": 0.637,
+      "step": 4367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8729115324747788e-05,
+      "loss": 0.6094,
+      "step": 4368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872843734769835e-05,
+      "loss": 0.5958,
+      "step": 4369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727759202135246e-05,
+      "loss": 0.587,
+      "step": 4370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727080888071573e-05,
+      "loss": 0.6346,
+      "step": 4371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726402405520425e-05,
+      "loss": 0.6147,
+      "step": 4372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87257237544949e-05,
+      "loss": 0.64,
+      "step": 4373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725044935008105e-05,
+      "loss": 0.5917,
+      "step": 4374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724365947073136e-05,
+      "loss": 0.5818,
+      "step": 4375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872368679070311e-05,
+      "loss": 0.5865,
+      "step": 4376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872300746591114e-05,
+      "loss": 0.6276,
+      "step": 4377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872232797271034e-05,
+      "loss": 0.5862,
+      "step": 4378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721648311113818e-05,
+      "loss": 0.6111,
+      "step": 4379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720968481134706e-05,
+      "loss": 0.5963,
+      "step": 4380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720288482786132e-05,
+      "loss": 0.6275,
+      "step": 4381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871960831608121e-05,
+      "loss": 0.587,
+      "step": 4382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871892798103309e-05,
+      "loss": 0.6211,
+      "step": 4383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8718247477654894e-05,
+      "loss": 0.5686,
+      "step": 4384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717566805959766e-05,
+      "loss": 0.6181,
+      "step": 4385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871688596596084e-05,
+      "loss": 0.5949,
+      "step": 4386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716204957671272e-05,
+      "loss": 0.5972,
+      "step": 4387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8715523781104196e-05,
+      "loss": 0.6571,
+      "step": 4388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714842436272774e-05,
+      "loss": 0.5915,
+      "step": 4389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8714160923190156e-05,
+      "loss": 0.6185,
+      "step": 4390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8713479241869498e-05,
+      "loss": 0.6238,
+      "step": 4391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712797392323963e-05,
+      "loss": 0.6134,
+      "step": 4392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712115374566715e-05,
+      "loss": 0.6205,
+      "step": 4393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711433188610922e-05,
+      "loss": 0.6308,
+      "step": 4394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871075083446975e-05,
+      "loss": 0.5983,
+      "step": 4395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710068312156376e-05,
+      "loss": 0.6062,
+      "step": 4396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870938562168398e-05,
+      "loss": 0.6276,
+      "step": 4397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870870276306574e-05,
+      "loss": 0.586,
+      "step": 4398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708019736314834e-05,
+      "loss": 0.6075,
+      "step": 4399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707336541444454e-05,
+      "loss": 0.6007,
+      "step": 4400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706653178467792e-05,
+      "loss": 0.5981,
+      "step": 4401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705969647398034e-05,
+      "loss": 0.606,
+      "step": 4402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870528594824838e-05,
+      "loss": 0.2897,
+      "step": 4403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704602081032036e-05,
+      "loss": 0.6505,
+      "step": 4404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703918045762197e-05,
+      "loss": 0.6184,
+      "step": 4405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703233842452072e-05,
+      "loss": 0.639,
+      "step": 4406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702549471114867e-05,
+      "loss": 0.6366,
+      "step": 4407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701864931763804e-05,
+      "loss": 0.6451,
+      "step": 4408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701180224412087e-05,
+      "loss": 0.5889,
+      "step": 4409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870049534907294e-05,
+      "loss": 0.6045,
+      "step": 4410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869981030575959e-05,
+      "loss": 0.5932,
+      "step": 4411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869912509448526e-05,
+      "loss": 0.6227,
+      "step": 4412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698439715263173e-05,
+      "loss": 0.6122,
+      "step": 4413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697754168106567e-05,
+      "loss": 0.615,
+      "step": 4414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697068453028677e-05,
+      "loss": 0.6175,
+      "step": 4415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869638257004274e-05,
+      "loss": 0.6255,
+      "step": 4416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695696519162e-05,
+      "loss": 0.6433,
+      "step": 4417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86950103003997e-05,
+      "loss": 0.5952,
+      "step": 4418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869432391376909e-05,
+      "loss": 0.6026,
+      "step": 4419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869363735928342e-05,
+      "loss": 0.6377,
+      "step": 4420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869295063695594e-05,
+      "loss": 0.5955,
+      "step": 4421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869226374679992e-05,
+      "loss": 0.6011,
+      "step": 4422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691576688828613e-05,
+      "loss": 0.6275,
+      "step": 4423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690889463055285e-05,
+      "loss": 0.6058,
+      "step": 4424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690202069493204e-05,
+      "loss": 0.5946,
+      "step": 4425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868951450815564e-05,
+      "loss": 0.6196,
+      "step": 4426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868882677905587e-05,
+      "loss": 0.6203,
+      "step": 4427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8688138882207168e-05,
+      "loss": 0.564,
+      "step": 4428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868745081762282e-05,
+      "loss": 0.6045,
+      "step": 4429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686762585316102e-05,
+      "loss": 0.5871,
+      "step": 4430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868607418530031e-05,
+      "loss": 0.5479,
+      "step": 4431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685385617588728e-05,
+      "loss": 0.6394,
+      "step": 4432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684696882194654e-05,
+      "loss": 0.6107,
+      "step": 4433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684007979131376e-05,
+      "loss": 0.6415,
+      "step": 4434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683318908412208e-05,
+      "loss": 0.5917,
+      "step": 4435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8682629670050443e-05,
+      "loss": 0.6336,
+      "step": 4436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868194026405939e-05,
+      "loss": 0.5601,
+      "step": 4437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681250690452365e-05,
+      "loss": 0.6016,
+      "step": 4438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680560949242674e-05,
+      "loss": 0.6374,
+      "step": 4439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679871040443632e-05,
+      "loss": 0.5699,
+      "step": 4440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679180964068565e-05,
+      "loss": 0.5848,
+      "step": 4441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867849072013079e-05,
+      "loss": 0.5805,
+      "step": 4442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677800308643637e-05,
+      "loss": 0.5854,
+      "step": 4443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677109729620434e-05,
+      "loss": 0.6392,
+      "step": 4444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676418983074515e-05,
+      "loss": 0.6027,
+      "step": 4445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867572806901921e-05,
+      "loss": 0.6281,
+      "step": 4446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675036987467864e-05,
+      "loss": 0.6364,
+      "step": 4447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674345738433818e-05,
+      "loss": 0.6168,
+      "step": 4448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673654321930418e-05,
+      "loss": 0.2756,
+      "step": 4449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867296273797101e-05,
+      "loss": 0.6614,
+      "step": 4450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867227098656895e-05,
+      "loss": 0.6431,
+      "step": 4451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671579067737586e-05,
+      "loss": 0.6001,
+      "step": 4452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670886981490283e-05,
+      "loss": 0.5915,
+      "step": 4453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670194727840402e-05,
+      "loss": 0.6233,
+      "step": 4454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8669502306801304e-05,
+      "loss": 0.5945,
+      "step": 4455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866880971838636e-05,
+      "loss": 0.6265,
+      "step": 4456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866811696260894e-05,
+      "loss": 0.5581,
+      "step": 4457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667424039482424e-05,
+      "loss": 0.5885,
+      "step": 4458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866673094902018e-05,
+      "loss": 0.6102,
+      "step": 4459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666037691235597e-05,
+      "loss": 0.6244,
+      "step": 4460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665344266142054e-05,
+      "loss": 0.6169,
+      "step": 4461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664650673752945e-05,
+      "loss": 0.5953,
+      "step": 4462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663956914081652e-05,
+      "loss": 0.5713,
+      "step": 4463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663262987141576e-05,
+      "loss": 0.2788,
+      "step": 4464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866256889294611e-05,
+      "loss": 0.6468,
+      "step": 4465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866187463150866e-05,
+      "loss": 0.6125,
+      "step": 4466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866118020284262e-05,
+      "loss": 0.6083,
+      "step": 4467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8660485606961405e-05,
+      "loss": 0.6278,
+      "step": 4468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865979084387842e-05,
+      "loss": 0.6362,
+      "step": 4469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659095913607083e-05,
+      "loss": 0.6719,
+      "step": 4470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865840081616081e-05,
+      "loss": 0.588,
+      "step": 4471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657705551553015e-05,
+      "loss": 0.5884,
+      "step": 4472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657010119797125e-05,
+      "loss": 0.5859,
+      "step": 4473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865631452090657e-05,
+      "loss": 0.6425,
+      "step": 4474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655618754894774e-05,
+      "loss": 0.6227,
+      "step": 4475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865492282177517e-05,
+      "loss": 0.609,
+      "step": 4476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8654226721561194e-05,
+      "loss": 0.6391,
+      "step": 4477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865353045426629e-05,
+      "loss": 0.5816,
+      "step": 4478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652834019903894e-05,
+      "loss": 0.5909,
+      "step": 4479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652137418487454e-05,
+      "loss": 0.5915,
+      "step": 4480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651440650030423e-05,
+      "loss": 0.5654,
+      "step": 4481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650743714546246e-05,
+      "loss": 0.5988,
+      "step": 4482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650046612048378e-05,
+      "loss": 0.5824,
+      "step": 4483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649349342550283e-05,
+      "loss": 0.6267,
+      "step": 4484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648651906065424e-05,
+      "loss": 0.601,
+      "step": 4485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647954302607256e-05,
+      "loss": 0.582,
+      "step": 4486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864725653218926e-05,
+      "loss": 0.6008,
+      "step": 4487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646558594824893e-05,
+      "loss": 0.6237,
+      "step": 4488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864586049052764e-05,
+      "loss": 0.6282,
+      "step": 4489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645162219310977e-05,
+      "loss": 0.6163,
+      "step": 4490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644463781188387e-05,
+      "loss": 0.6101,
+      "step": 4491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643765176173346e-05,
+      "loss": 0.6234,
+      "step": 4492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643066404279347e-05,
+      "loss": 0.6405,
+      "step": 4493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642367465519885e-05,
+      "loss": 0.6275,
+      "step": 4494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641668359908447e-05,
+      "loss": 0.59,
+      "step": 4495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640969087458532e-05,
+      "loss": 0.6393,
+      "step": 4496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640269648183642e-05,
+      "loss": 0.6234,
+      "step": 4497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8639570042097278e-05,
+      "loss": 0.57,
+      "step": 4498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863887026921295e-05,
+      "loss": 0.6221,
+      "step": 4499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638170329544164e-05,
+      "loss": 0.5652,
+      "step": 4500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637470223104437e-05,
+      "loss": 0.6061,
+      "step": 4501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636769949907283e-05,
+      "loss": 0.6325,
+      "step": 4502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636069509966223e-05,
+      "loss": 0.5937,
+      "step": 4503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863536890329478e-05,
+      "loss": 0.6472,
+      "step": 4504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634668129906477e-05,
+      "loss": 0.6359,
+      "step": 4505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863396718981485e-05,
+      "loss": 0.616,
+      "step": 4506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633266083033424e-05,
+      "loss": 0.6189,
+      "step": 4507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863256480957574e-05,
+      "loss": 0.6299,
+      "step": 4508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631863369455337e-05,
+      "loss": 0.6242,
+      "step": 4509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863116176268575e-05,
+      "loss": 0.5862,
+      "step": 4510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630459989280535e-05,
+      "loss": 0.5925,
+      "step": 4511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629758049253232e-05,
+      "loss": 0.5993,
+      "step": 4512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629055942617398e-05,
+      "loss": 0.5861,
+      "step": 4513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628353669386585e-05,
+      "loss": 0.5943,
+      "step": 4514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627651229574356e-05,
+      "loss": 0.5591,
+      "step": 4515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626948623194266e-05,
+      "loss": 0.5842,
+      "step": 4516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862624585025988e-05,
+      "loss": 0.5944,
+      "step": 4517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862554291078477e-05,
+      "loss": 0.6158,
+      "step": 4518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624839804782512e-05,
+      "loss": 0.5842,
+      "step": 4519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624136532266673e-05,
+      "loss": 0.5916,
+      "step": 4520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623433093250825e-05,
+      "loss": 0.6337,
+      "step": 4521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622729487748564e-05,
+      "loss": 0.6146,
+      "step": 4522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862202571577346e-05,
+      "loss": 0.5957,
+      "step": 4523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862132177733911e-05,
+      "loss": 0.5989,
+      "step": 4524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620617672459097e-05,
+      "loss": 0.613,
+      "step": 4525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861991340114702e-05,
+      "loss": 0.5749,
+      "step": 4526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861920896341647e-05,
+      "loss": 0.5921,
+      "step": 4527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618504359281055e-05,
+      "loss": 0.5963,
+      "step": 4528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617799588754374e-05,
+      "loss": 0.6151,
+      "step": 4529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617094651850032e-05,
+      "loss": 0.6271,
+      "step": 4530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861638954858164e-05,
+      "loss": 0.6399,
+      "step": 4531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615684278962815e-05,
+      "loss": 0.62,
+      "step": 4532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8614978843007163e-05,
+      "loss": 0.5923,
+      "step": 4533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8614273240728317e-05,
+      "loss": 0.6113,
+      "step": 4534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861356747213989e-05,
+      "loss": 0.6025,
+      "step": 4535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612861537255508e-05,
+      "loss": 0.5881,
+      "step": 4536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612155436088802e-05,
+      "loss": 0.589,
+      "step": 4537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611449168653404e-05,
+      "loss": 0.5829,
+      "step": 4538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610742734962953e-05,
+      "loss": 0.6102,
+      "step": 4539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610036135031084e-05,
+      "loss": 0.609,
+      "step": 4540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609329368871435e-05,
+      "loss": 0.2758,
+      "step": 4541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608622436497657e-05,
+      "loss": 0.5862,
+      "step": 4542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607915337923397e-05,
+      "loss": 0.6015,
+      "step": 4543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607208073162307e-05,
+      "loss": 0.6105,
+      "step": 4544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8606500642228037e-05,
+      "loss": 0.6234,
+      "step": 4545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605793045134254e-05,
+      "loss": 0.6231,
+      "step": 4546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860508528189461e-05,
+      "loss": 0.6219,
+      "step": 4547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860437735252277e-05,
+      "loss": 0.6187,
+      "step": 4548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603669257032407e-05,
+      "loss": 0.6414,
+      "step": 4549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8602960995437187e-05,
+      "loss": 0.5928,
+      "step": 4550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860225256775079e-05,
+      "loss": 0.5953,
+      "step": 4551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601543973986885e-05,
+      "loss": 0.6103,
+      "step": 4552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8600835214159157e-05,
+      "loss": 0.5793,
+      "step": 4553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860012628828129e-05,
+      "loss": 0.587,
+      "step": 4554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599417196366967e-05,
+      "loss": 0.5985,
+      "step": 4555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598707938429884e-05,
+      "loss": 0.6066,
+      "step": 4556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597998514483724e-05,
+      "loss": 0.6484,
+      "step": 4557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597288924542198e-05,
+      "loss": 0.625,
+      "step": 4558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859657916861899e-05,
+      "loss": 0.5898,
+      "step": 4559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595869246727814e-05,
+      "loss": 0.6305,
+      "step": 4560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859515915888237e-05,
+      "loss": 0.6367,
+      "step": 4561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594448905096372e-05,
+      "loss": 0.6104,
+      "step": 4562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593738485383528e-05,
+      "loss": 0.5941,
+      "step": 4563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593027899757557e-05,
+      "loss": 0.626,
+      "step": 4564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859231714823217e-05,
+      "loss": 0.5941,
+      "step": 4565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591606230821103e-05,
+      "loss": 0.6526,
+      "step": 4566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590895147538067e-05,
+      "loss": 0.5961,
+      "step": 4567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590183898396795e-05,
+      "loss": 0.6129,
+      "step": 4568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589472483411026e-05,
+      "loss": 0.6383,
+      "step": 4569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588760902594485e-05,
+      "loss": 0.6392,
+      "step": 4570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588049155960917e-05,
+      "loss": 0.5808,
+      "step": 4571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587337243524058e-05,
+      "loss": 0.6136,
+      "step": 4572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858662516529765e-05,
+      "loss": 0.61,
+      "step": 4573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858591292129545e-05,
+      "loss": 0.6115,
+      "step": 4574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585200511531203e-05,
+      "loss": 0.579,
+      "step": 4575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584487936018663e-05,
+      "loss": 0.6001,
+      "step": 4576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858377519477159e-05,
+      "loss": 0.6055,
+      "step": 4577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583062287803738e-05,
+      "loss": 0.5941,
+      "step": 4578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582349215128873e-05,
+      "loss": 0.6011,
+      "step": 4579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858163597676077e-05,
+      "loss": 0.6281,
+      "step": 4580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580922572713193e-05,
+      "loss": 0.6201,
+      "step": 4581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858020900299991e-05,
+      "loss": 0.6195,
+      "step": 4582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8579495267634703e-05,
+      "loss": 0.5909,
+      "step": 4583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857878136663135e-05,
+      "loss": 0.5984,
+      "step": 4584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8578067300003634e-05,
+      "loss": 0.6282,
+      "step": 4585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577353067765344e-05,
+      "loss": 0.5938,
+      "step": 4586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8576638669930266e-05,
+      "loss": 0.6083,
+      "step": 4587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575924106512193e-05,
+      "loss": 0.5703,
+      "step": 4588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857520937752492e-05,
+      "loss": 0.6325,
+      "step": 4589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857449448298225e-05,
+      "loss": 0.5887,
+      "step": 4590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573779422897974e-05,
+      "loss": 0.6237,
+      "step": 4591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573064197285908e-05,
+      "loss": 0.6582,
+      "step": 4592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572348806159857e-05,
+      "loss": 0.6048,
+      "step": 4593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571633249533633e-05,
+      "loss": 0.6125,
+      "step": 4594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857091752742105e-05,
+      "loss": 0.6138,
+      "step": 4595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570201639835922e-05,
+      "loss": 0.6366,
+      "step": 4596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856948558679208e-05,
+      "loss": 0.5846,
+      "step": 4597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568769368303335e-05,
+      "loss": 0.6129,
+      "step": 4598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856805298438353e-05,
+      "loss": 0.6003,
+      "step": 4599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567336435046482e-05,
+      "loss": 0.6111,
+      "step": 4600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566619720306032e-05,
+      "loss": 0.6075,
+      "step": 4601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856590284017602e-05,
+      "loss": 0.626,
+      "step": 4602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565185794670274e-05,
+      "loss": 0.6202,
+      "step": 4603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856446858380265e-05,
+      "loss": 0.6005,
+      "step": 4604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856375120758699e-05,
+      "loss": 0.6159,
+      "step": 4605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563033666037143e-05,
+      "loss": 0.5945,
+      "step": 4606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562315959166965e-05,
+      "loss": 0.5842,
+      "step": 4607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8561598086990304e-05,
+      "loss": 0.5893,
+      "step": 4608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856088004952103e-05,
+      "loss": 0.621,
+      "step": 4609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8560161846773002e-05,
+      "loss": 0.5537,
+      "step": 4610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8559443478760086e-05,
+      "loss": 0.6352,
+      "step": 4611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558724945496148e-05,
+      "loss": 0.6391,
+      "step": 4612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558006246995066e-05,
+      "loss": 0.6563,
+      "step": 4613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557287383270703e-05,
+      "loss": 0.593,
+      "step": 4614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556568354336955e-05,
+      "loss": 0.5612,
+      "step": 4615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855584916020769e-05,
+      "loss": 0.6368,
+      "step": 4616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85551298008968e-05,
+      "loss": 0.6195,
+      "step": 4617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554410276418174e-05,
+      "loss": 0.6232,
+      "step": 4618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553690586785697e-05,
+      "loss": 0.601,
+      "step": 4619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552970732013267e-05,
+      "loss": 0.6243,
+      "step": 4620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552250712114783e-05,
+      "loss": 0.6356,
+      "step": 4621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8551530527104143e-05,
+      "loss": 0.6081,
+      "step": 4622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8550810176995255e-05,
+      "loss": 0.6078,
+      "step": 4623
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550089661802024e-05,
+      "loss": 0.628,
+      "step": 4624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854936898153836e-05,
+      "loss": 0.6304,
+      "step": 4625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548648136218177e-05,
+      "loss": 0.6237,
+      "step": 4626
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854792712585539e-05,
+      "loss": 0.648,
+      "step": 4627
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547205950463925e-05,
+      "loss": 0.5979,
+      "step": 4628
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546484610057698e-05,
+      "loss": 0.5778,
+      "step": 4629
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545763104650643e-05,
+      "loss": 0.5816,
+      "step": 4630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854504143425668e-05,
+      "loss": 0.5867,
+      "step": 4631
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854431959888975e-05,
+      "loss": 0.6372,
+      "step": 4632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543597598563786e-05,
+      "loss": 0.5919,
+      "step": 4633
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542875433292728e-05,
+      "loss": 0.5848,
+      "step": 4634
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542153103090515e-05,
+      "loss": 0.642,
+      "step": 4635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8541430607971095e-05,
+      "loss": 0.6462,
+      "step": 4636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540707947948418e-05,
+      "loss": 0.615,
+      "step": 4637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539985123036436e-05,
+      "loss": 0.6271,
+      "step": 4638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539262133249097e-05,
+      "loss": 0.6218,
+      "step": 4639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538538978600372e-05,
+      "loss": 0.595,
+      "step": 4640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853781565910421e-05,
+      "loss": 0.6211,
+      "step": 4641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537092174774583e-05,
+      "loss": 0.6104,
+      "step": 4642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853636852562546e-05,
+      "loss": 0.5875,
+      "step": 4643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8535644711670804e-05,
+      "loss": 0.6252,
+      "step": 4644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534920732924594e-05,
+      "loss": 0.602,
+      "step": 4645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853419658940081e-05,
+      "loss": 0.5692,
+      "step": 4646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533472281113428e-05,
+      "loss": 0.5987,
+      "step": 4647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532747808076433e-05,
+      "loss": 0.6193,
+      "step": 4648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532023170303812e-05,
+      "loss": 0.5964,
+      "step": 4649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8531298367809557e-05,
+      "loss": 0.6197,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530573400607656e-05,
+      "loss": 0.5772,
+      "step": 4651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529848268712114e-05,
+      "loss": 0.5789,
+      "step": 4652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852912297213692e-05,
+      "loss": 0.5876,
+      "step": 4653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528397510896086e-05,
+      "loss": 0.6061,
+      "step": 4654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527671885003614e-05,
+      "loss": 0.5967,
+      "step": 4655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526946094473514e-05,
+      "loss": 0.5999,
+      "step": 4656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526220139319797e-05,
+      "loss": 0.577,
+      "step": 4657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525494019556477e-05,
+      "loss": 0.5827,
+      "step": 4658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852476773519758e-05,
+      "loss": 0.6199,
+      "step": 4659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852404128625712e-05,
+      "loss": 0.5933,
+      "step": 4660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8523314672749123e-05,
+      "loss": 0.5813,
+      "step": 4661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852258789468762e-05,
+      "loss": 0.5984,
+      "step": 4662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521860952086642e-05,
+      "loss": 0.6185,
+      "step": 4663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521133844960225e-05,
+      "loss": 0.602,
+      "step": 4664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85204065733224e-05,
+      "loss": 0.5926,
+      "step": 4665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519679137187215e-05,
+      "loss": 0.612,
+      "step": 4666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8518951536568714e-05,
+      "loss": 0.6443,
+      "step": 4667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851822377148094e-05,
+      "loss": 0.5903,
+      "step": 4668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517495841937946e-05,
+      "loss": 0.6187,
+      "step": 4669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516767747953784e-05,
+      "loss": 0.6107,
+      "step": 4670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516039489542513e-05,
+      "loss": 0.6133,
+      "step": 4671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851531106671819e-05,
+      "loss": 0.5813,
+      "step": 4672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851458247949488e-05,
+      "loss": 0.6025,
+      "step": 4673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851385372788665e-05,
+      "loss": 0.5914,
+      "step": 4674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513124811907568e-05,
+      "loss": 0.5756,
+      "step": 4675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851239573157171e-05,
+      "loss": 0.2607,
+      "step": 4676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511666486893144e-05,
+      "loss": 0.5776,
+      "step": 4677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851093707788596e-05,
+      "loss": 0.5829,
+      "step": 4678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510207504564232e-05,
+      "loss": 0.6206,
+      "step": 4679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509477766942045e-05,
+      "loss": 0.5915,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850874786503349e-05,
+      "loss": 0.5783,
+      "step": 4681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508017798852666e-05,
+      "loss": 0.6446,
+      "step": 4682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507287568413656e-05,
+      "loss": 0.5905,
+      "step": 4683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850655717373056e-05,
+      "loss": 0.5943,
+      "step": 4684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505826614817485e-05,
+      "loss": 0.5795,
+      "step": 4685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850509589168853e-05,
+      "loss": 0.6313,
+      "step": 4686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504365004357808e-05,
+      "loss": 0.6178,
+      "step": 4687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503633952839428e-05,
+      "loss": 0.5865,
+      "step": 4688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85029027371475e-05,
+      "loss": 0.5874,
+      "step": 4689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 0.582,
+      "step": 4690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501439813299476e-05,
+      "loss": 0.6236,
+      "step": 4691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500708105171628e-05,
+      "loss": 0.5411,
+      "step": 4692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499976232926722e-05,
+      "loss": 0.5998,
+      "step": 4693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499244196578886e-05,
+      "loss": 0.6016,
+      "step": 4694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498511996142255e-05,
+      "loss": 0.5891,
+      "step": 4695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8497779631630965e-05,
+      "loss": 0.5917,
+      "step": 4696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849704710305915e-05,
+      "loss": 0.6275,
+      "step": 4697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849631441044096e-05,
+      "loss": 0.6251,
+      "step": 4698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849558155379054e-05,
+      "loss": 0.6011,
+      "step": 4699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8494848533122035e-05,
+      "loss": 0.6422,
+      "step": 4700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84941153484496e-05,
+      "loss": 0.2882,
+      "step": 4701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493381999787387e-05,
+      "loss": 0.5688,
+      "step": 4702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849264848714955e-05,
+      "loss": 0.6145,
+      "step": 4703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491914810550266e-05,
+      "loss": 0.6044,
+      "step": 4704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849118097000368e-05,
+      "loss": 0.6218,
+      "step": 4705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8490446965523977e-05,
+      "loss": 0.5669,
+      "step": 4706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489712797125315e-05,
+      "loss": 0.5923,
+      "step": 4707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488978464821875e-05,
+      "loss": 0.6076,
+      "step": 4708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848824396862783e-05,
+      "loss": 0.5989,
+      "step": 4709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848750930855737e-05,
+      "loss": 0.6494,
+      "step": 4710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848677448462467e-05,
+      "loss": 0.5896,
+      "step": 4711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848603949684391e-05,
+      "loss": 0.592,
+      "step": 4712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485304345229296e-05,
+      "loss": 0.6364,
+      "step": 4713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848456902979501e-05,
+      "loss": 0.5958,
+      "step": 4714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483833550555252e-05,
+      "loss": 0.5901,
+      "step": 4715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848309790752422e-05,
+      "loss": 0.5897,
+      "step": 4716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482362100716118e-05,
+      "loss": 0.6184,
+      "step": 4717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481626130145146e-05,
+      "loss": 0.5946,
+      "step": 4718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480889995825527e-05,
+      "loss": 0.6108,
+      "step": 4719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480153697771455e-05,
+      "loss": 0.5588,
+      "step": 4720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847941723599716e-05,
+      "loss": 0.6011,
+      "step": 4721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847868061051685e-05,
+      "loss": 0.6024,
+      "step": 4722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847794382134475e-05,
+      "loss": 0.5877,
+      "step": 4723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847720686849509e-05,
+      "loss": 0.5928,
+      "step": 4724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476469751982095e-05,
+      "loss": 0.6176,
+      "step": 4725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847573247181999e-05,
+      "loss": 0.5965,
+      "step": 4726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8474995028023013e-05,
+      "loss": 0.591,
+      "step": 4727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84742574206054e-05,
+      "loss": 0.5787,
+      "step": 4728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8473519649581396e-05,
+      "loss": 0.6326,
+      "step": 4729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472781714965242e-05,
+      "loss": 0.6036,
+      "step": 4730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472043616771184e-05,
+      "loss": 0.5392,
+      "step": 4731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471305355013474e-05,
+      "loss": 0.6279,
+      "step": 4732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847056692970636e-05,
+      "loss": 0.6183,
+      "step": 4733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469828340864104e-05,
+      "loss": 0.6009,
+      "step": 4734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469089588500964e-05,
+      "loss": 0.6491,
+      "step": 4735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84683506726312e-05,
+      "loss": 0.5976,
+      "step": 4736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467611593269075e-05,
+      "loss": 0.599,
+      "step": 4737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846687235042887e-05,
+      "loss": 0.6028,
+      "step": 4738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466132944124848e-05,
+      "loss": 0.6057,
+      "step": 4739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8465393374371282e-05,
+      "loss": 0.6518,
+      "step": 4740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464653641182458e-05,
+      "loss": 0.6098,
+      "step": 4741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846391374457265e-05,
+      "loss": 0.6296,
+      "step": 4742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8463173684556147e-05,
+      "loss": 0.5848,
+      "step": 4743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462433461147235e-05,
+      "loss": 0.6288,
+      "step": 4744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461693074360207e-05,
+      "loss": 0.5793,
+      "step": 4745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8460952524209355e-05,
+      "loss": 0.6311,
+      "step": 4746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846021181070898e-05,
+      "loss": 0.5888,
+      "step": 4747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459470933873377e-05,
+      "loss": 0.6461,
+      "step": 4748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458729893716852e-05,
+      "loss": 0.6317,
+      "step": 4749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845798869025371e-05,
+      "loss": 0.5811,
+      "step": 4750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845724732349827e-05,
+      "loss": 0.6088,
+      "step": 4751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845650579346483e-05,
+      "loss": 0.6108,
+      "step": 4752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455764100167716e-05,
+      "loss": 0.6149,
+      "step": 4753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455022243621247e-05,
+      "loss": 0.613,
+      "step": 4754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454280223839745e-05,
+      "loss": 0.6011,
+      "step": 4755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453538040837535e-05,
+      "loss": 0.619,
+      "step": 4756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845279569462894e-05,
+      "loss": 0.589,
+      "step": 4757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84520531852283e-05,
+      "loss": 0.5862,
+      "step": 4758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845131051264995e-05,
+      "loss": 0.6056,
+      "step": 4759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450567676908222e-05,
+      "loss": 0.5883,
+      "step": 4760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844982467801746e-05,
+      "loss": 0.6174,
+      "step": 4761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844908151599201e-05,
+      "loss": 0.6316,
+      "step": 4762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844833819084622e-05,
+      "loss": 0.6295,
+      "step": 4763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447594702594442e-05,
+      "loss": 0.6239,
+      "step": 4764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446851051251027e-05,
+      "loss": 0.584,
+      "step": 4765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446107236830336e-05,
+      "loss": 0.5883,
+      "step": 4766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844536325934672e-05,
+      "loss": 0.589,
+      "step": 4767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8444619118814553e-05,
+      "loss": 0.5869,
+      "step": 4768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84438748152482e-05,
+      "loss": 0.6604,
+      "step": 4769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443130348662022e-05,
+      "loss": 0.592,
+      "step": 4770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84423857190704e-05,
+      "loss": 0.6107,
+      "step": 4771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844164092648771e-05,
+      "loss": 0.5826,
+      "step": 4772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844089597092833e-05,
+      "loss": 0.5738,
+      "step": 4773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8440150852406636e-05,
+      "loss": 0.5962,
+      "step": 4774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439405570937027e-05,
+      "loss": 0.5861,
+      "step": 4775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438660126533875e-05,
+      "loss": 0.6181,
+      "step": 4776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.843791451921159e-05,
+      "loss": 0.5662,
+      "step": 4777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.843716874898455e-05,
+      "loss": 0.613,
+      "step": 4778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8436422815867165e-05,
+      "loss": 0.5904,
+      "step": 4779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435676719873828e-05,
+      "loss": 0.585,
+      "step": 4780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843493046101895e-05,
+      "loss": 0.6052,
+      "step": 4781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434184039316935e-05,
+      "loss": 0.5858,
+      "step": 4782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843343745478219e-05,
+      "loss": 0.592,
+      "step": 4783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432690707429137e-05,
+      "loss": 0.5931,
+      "step": 4784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8431943797272187e-05,
+      "loss": 0.6325,
+      "step": 4785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843119672432576e-05,
+      "loss": 0.5895,
+      "step": 4786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430449488604284e-05,
+      "loss": 0.599,
+      "step": 4787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429702090122184e-05,
+      "loss": 0.613,
+      "step": 4788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8428954528893884e-05,
+      "loss": 0.6,
+      "step": 4789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8428206804933825e-05,
+      "loss": 0.634,
+      "step": 4790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427458918256432e-05,
+      "loss": 0.6433,
+      "step": 4791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426710868876157e-05,
+      "loss": 0.6079,
+      "step": 4792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842596265680743e-05,
+      "loss": 0.6194,
+      "step": 4793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84252142820647e-05,
+      "loss": 0.6755,
+      "step": 4794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842446574466242e-05,
+      "loss": 0.5996,
+      "step": 4795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842371704461504e-05,
+      "loss": 0.6233,
+      "step": 4796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842296818193701e-05,
+      "loss": 0.6143,
+      "step": 4797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422219156642792e-05,
+      "loss": 0.6125,
+      "step": 4798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8421469968746843e-05,
+      "loss": 0.6394,
+      "step": 4799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420720618263632e-05,
+      "loss": 0.6293,
+      "step": 4800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419971105207625e-05,
+      "loss": 0.6152,
+      "step": 4801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419221429593287e-05,
+      "loss": 0.595,
+      "step": 4802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418471591435098e-05,
+      "loss": 0.5739,
+      "step": 4803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841772159074753e-05,
+      "loss": 0.5452,
+      "step": 4804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416971427545065e-05,
+      "loss": 0.5634,
+      "step": 4805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416221101842186e-05,
+      "loss": 0.592,
+      "step": 4806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841547061365338e-05,
+      "loss": 0.6268,
+      "step": 4807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8414719962993134e-05,
+      "loss": 0.605,
+      "step": 4808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841396914987594e-05,
+      "loss": 0.6342,
+      "step": 4809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413218174316295e-05,
+      "loss": 0.6005,
+      "step": 4810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412467036328698e-05,
+      "loss": 0.6054,
+      "step": 4811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411715735927645e-05,
+      "loss": 0.6106,
+      "step": 4812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410964273127653e-05,
+      "loss": 0.6121,
+      "step": 4813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410212647943215e-05,
+      "loss": 0.5847,
+      "step": 4814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409460860388852e-05,
+      "loss": 0.5966,
+      "step": 4815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408708910479077e-05,
+      "loss": 0.6135,
+      "step": 4816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8407956798228407e-05,
+      "loss": 0.6074,
+      "step": 4817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840720452365136e-05,
+      "loss": 0.2611,
+      "step": 4818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406452086762462e-05,
+      "loss": 0.6226,
+      "step": 4819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405699487576237e-05,
+      "loss": 0.591,
+      "step": 4820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404946726107217e-05,
+      "loss": 0.5913,
+      "step": 4821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404193802369937e-05,
+      "loss": 0.597,
+      "step": 4822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840344071637893e-05,
+      "loss": 0.2475,
+      "step": 4823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402687468148734e-05,
+      "loss": 0.5817,
+      "step": 4824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401934057693895e-05,
+      "loss": 0.6017,
+      "step": 4825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840118048502896e-05,
+      "loss": 0.6122,
+      "step": 4826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400426750168474e-05,
+      "loss": 0.5987,
+      "step": 4827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399672853126986e-05,
+      "loss": 0.5775,
+      "step": 4828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839891879391906e-05,
+      "loss": 0.6082,
+      "step": 4829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398164572559243e-05,
+      "loss": 0.6207,
+      "step": 4830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397410189062106e-05,
+      "loss": 0.5822,
+      "step": 4831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839665564344221e-05,
+      "loss": 0.6066,
+      "step": 4832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839590093571412e-05,
+      "loss": 0.5887,
+      "step": 4833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8395146065892406e-05,
+      "loss": 0.6316,
+      "step": 4834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839439103399165e-05,
+      "loss": 0.2635,
+      "step": 4835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393635840026418e-05,
+      "loss": 0.5963,
+      "step": 4836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392880484011296e-05,
+      "loss": 0.5779,
+      "step": 4837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392124965960868e-05,
+      "loss": 0.6217,
+      "step": 4838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391369285889717e-05,
+      "loss": 0.5976,
+      "step": 4839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390613443812432e-05,
+      "loss": 0.5851,
+      "step": 4840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838985743974361e-05,
+      "loss": 0.6505,
+      "step": 4841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389101273697846e-05,
+      "loss": 0.617,
+      "step": 4842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388344945689733e-05,
+      "loss": 0.632,
+      "step": 4843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387588455733877e-05,
+      "loss": 0.5716,
+      "step": 4844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838683180384488e-05,
+      "loss": 0.5674,
+      "step": 4845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8386074990037356e-05,
+      "loss": 0.5808,
+      "step": 4846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385318014325913e-05,
+      "loss": 0.621,
+      "step": 4847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8384560876725163e-05,
+      "loss": 0.6273,
+      "step": 4848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383803577249728e-05,
+      "loss": 0.6228,
+      "step": 4849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383046115914225e-05,
+      "loss": 0.6401,
+      "step": 4850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838228849273328e-05,
+      "loss": 0.6076,
+      "step": 4851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838153070772152e-05,
+      "loss": 0.5924,
+      "step": 4852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838077276089357e-05,
+      "loss": 0.6024,
+      "step": 4853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380014652264072e-05,
+      "loss": 0.5727,
+      "step": 4854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8379256381847655e-05,
+      "loss": 0.5748,
+      "step": 4855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837849794965896e-05,
+      "loss": 0.5845,
+      "step": 4856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837773935571263e-05,
+      "loss": 0.6183,
+      "step": 4857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376980600023312e-05,
+      "loss": 0.6378,
+      "step": 4858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376221682605657e-05,
+      "loss": 0.6165,
+      "step": 4859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837546260347431e-05,
+      "loss": 0.6453,
+      "step": 4860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837470336264393e-05,
+      "loss": 0.6132,
+      "step": 4861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373943960129174e-05,
+      "loss": 0.5563,
+      "step": 4862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373184395944706e-05,
+      "loss": 0.5789,
+      "step": 4863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837242467010519e-05,
+      "loss": 0.6012,
+      "step": 4864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371664782625287e-05,
+      "loss": 0.5874,
+      "step": 4865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8370904733519675e-05,
+      "loss": 0.6026,
+      "step": 4866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8370144522803025e-05,
+      "loss": 0.586,
+      "step": 4867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369384150490013e-05,
+      "loss": 0.5548,
+      "step": 4868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8368623616595322e-05,
+      "loss": 0.5705,
+      "step": 4869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836786292113363e-05,
+      "loss": 0.5951,
+      "step": 4870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836710206411963e-05,
+      "loss": 0.642,
+      "step": 4871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366341045568004e-05,
+      "loss": 0.6283,
+      "step": 4872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836557986549345e-05,
+      "loss": 0.5755,
+      "step": 4873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836481852391066e-05,
+      "loss": 0.6567,
+      "step": 4874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364057020834336e-05,
+      "loss": 0.5802,
+      "step": 4875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363295356279175e-05,
+      "loss": 0.6212,
+      "step": 4876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362533530259884e-05,
+      "loss": 0.5538,
+      "step": 4877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361771542791176e-05,
+      "loss": 0.5747,
+      "step": 4878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361009393887756e-05,
+      "loss": 0.6131,
+      "step": 4879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360247083564343e-05,
+      "loss": 0.5885,
+      "step": 4880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359484611835647e-05,
+      "loss": 0.584,
+      "step": 4881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358721978716398e-05,
+      "loss": 0.6034,
+      "step": 4882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835795918422131e-05,
+      "loss": 0.6521,
+      "step": 4883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835719622836512e-05,
+      "loss": 0.6199,
+      "step": 4884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356433111162547e-05,
+      "loss": 0.5879,
+      "step": 4885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355669832628332e-05,
+      "loss": 0.6204,
+      "step": 4886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354906392777207e-05,
+      "loss": 0.6213,
+      "step": 4887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354142791623914e-05,
+      "loss": 0.6238,
+      "step": 4888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353379029183194e-05,
+      "loss": 0.5893,
+      "step": 4889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835261510546979e-05,
+      "loss": 0.2611,
+      "step": 4890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351851020498457e-05,
+      "loss": 0.6219,
+      "step": 4891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835108677428394e-05,
+      "loss": 0.5614,
+      "step": 4892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350322366840997e-05,
+      "loss": 0.6318,
+      "step": 4893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8349557798184385e-05,
+      "loss": 0.5991,
+      "step": 4894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348793068328864e-05,
+      "loss": 0.6144,
+      "step": 4895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348028177289198e-05,
+      "loss": 0.6063,
+      "step": 4896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347263125080154e-05,
+      "loss": 0.6232,
+      "step": 4897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346497911716506e-05,
+      "loss": 0.6237,
+      "step": 4898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834573253721303e-05,
+      "loss": 0.616,
+      "step": 4899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834496700158449e-05,
+      "loss": 0.6251,
+      "step": 4900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834420130484568e-05,
+      "loss": 0.6048,
+      "step": 4901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834343544701137e-05,
+      "loss": 0.6011,
+      "step": 4902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342669428096352e-05,
+      "loss": 0.6,
+      "step": 4903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341903248115418e-05,
+      "loss": 0.6343,
+      "step": 4904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341136907083356e-05,
+      "loss": 0.6066,
+      "step": 4905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340370405014966e-05,
+      "loss": 0.6084,
+      "step": 4906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339603741925037e-05,
+      "loss": 0.6188,
+      "step": 4907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338836917828374e-05,
+      "loss": 0.617,
+      "step": 4908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338069932739788e-05,
+      "loss": 0.6344,
+      "step": 4909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833730278667408e-05,
+      "loss": 0.5794,
+      "step": 4910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336535479646063e-05,
+      "loss": 0.5976,
+      "step": 4911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8335768011670553e-05,
+      "loss": 0.5839,
+      "step": 4912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833500038276236e-05,
+      "loss": 0.569,
+      "step": 4913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8334232592936313e-05,
+      "loss": 0.605,
+      "step": 4914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833346464220723e-05,
+      "loss": 0.6237,
+      "step": 4915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332696530589936e-05,
+      "loss": 0.6121,
+      "step": 4916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331928258099263e-05,
+      "loss": 0.5786,
+      "step": 4917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833115982475004e-05,
+      "loss": 0.5968,
+      "step": 4918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833039123055711e-05,
+      "loss": 0.6108,
+      "step": 4919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329622475535306e-05,
+      "loss": 0.2714,
+      "step": 4920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832885355969947e-05,
+      "loss": 0.5873,
+      "step": 4921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832808448306445e-05,
+      "loss": 0.595,
+      "step": 4922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832731524564509e-05,
+      "loss": 0.6021,
+      "step": 4923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832654584745624e-05,
+      "loss": 0.6258,
+      "step": 4924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325776288512765e-05,
+      "loss": 0.5684,
+      "step": 4925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832500656882951e-05,
+      "loss": 0.6025,
+      "step": 4926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832423668842134e-05,
+      "loss": 0.5743,
+      "step": 4927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832346664730312e-05,
+      "loss": 0.6111,
+      "step": 4928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8322696445489713e-05,
+      "loss": 0.6037,
+      "step": 4929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832192608299599e-05,
+      "loss": 0.6317,
+      "step": 4930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321155559836828e-05,
+      "loss": 0.5828,
+      "step": 4931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8320384876027096e-05,
+      "loss": 0.6152,
+      "step": 4932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831961403158168e-05,
+      "loss": 0.6054,
+      "step": 4933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318843026515458e-05,
+      "loss": 0.6481,
+      "step": 4934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318071860843315e-05,
+      "loss": 0.6133,
+      "step": 4935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831730053458014e-05,
+      "loss": 0.6117,
+      "step": 4936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8316529047740826e-05,
+      "loss": 0.6387,
+      "step": 4937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315757400340267e-05,
+      "loss": 0.605,
+      "step": 4938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314985592393358e-05,
+      "loss": 0.5815,
+      "step": 4939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314213623915003e-05,
+      "loss": 0.5955,
+      "step": 4940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313441494920106e-05,
+      "loss": 0.548,
+      "step": 4941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831266920542357e-05,
+      "loss": 0.6103,
+      "step": 4942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831189675544031e-05,
+      "loss": 0.6035,
+      "step": 4943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311124144985234e-05,
+      "loss": 0.5985,
+      "step": 4944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8310351374073262e-05,
+      "loss": 0.5915,
+      "step": 4945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309578442719315e-05,
+      "loss": 0.5627,
+      "step": 4946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308805350938307e-05,
+      "loss": 0.5949,
+      "step": 4947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308032098745172e-05,
+      "loss": 0.6043,
+      "step": 4948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830725868615484e-05,
+      "loss": 0.6312,
+      "step": 4949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830648511318223e-05,
+      "loss": 0.5913,
+      "step": 4950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830571137984229e-05,
+      "loss": 0.5962,
+      "step": 4951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304937486149955e-05,
+      "loss": 0.5817,
+      "step": 4952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304163432120162e-05,
+      "loss": 0.6001,
+      "step": 4953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830338921776786e-05,
+      "loss": 0.6002,
+      "step": 4954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830261484310799e-05,
+      "loss": 0.6013,
+      "step": 4955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301840308155507e-05,
+      "loss": 0.6076,
+      "step": 4956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301065612925364e-05,
+      "loss": 0.6466,
+      "step": 4957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830029075743252e-05,
+      "loss": 0.581,
+      "step": 4958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299515741691926e-05,
+      "loss": 0.6122,
+      "step": 4959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829874056571855e-05,
+      "loss": 0.6253,
+      "step": 4960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297965229527363e-05,
+      "loss": 0.6064,
+      "step": 4961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297189733133326e-05,
+      "loss": 0.6218,
+      "step": 4962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829641407655141e-05,
+      "loss": 0.5928,
+      "step": 4963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295638259796598e-05,
+      "loss": 0.6118,
+      "step": 4964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294862282883864e-05,
+      "loss": 0.5851,
+      "step": 4965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294086145828187e-05,
+      "loss": 0.6075,
+      "step": 4966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8293309848644554e-05,
+      "loss": 0.609,
+      "step": 4967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829253339134795e-05,
+      "loss": 0.5825,
+      "step": 4968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829175677395337e-05,
+      "loss": 0.5998,
+      "step": 4969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290979996475804e-05,
+      "loss": 0.6009,
+      "step": 4970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290203058930252e-05,
+      "loss": 0.6027,
+      "step": 4971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289425961331705e-05,
+      "loss": 0.5988,
+      "step": 4972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8288648703695177e-05,
+      "loss": 0.5741,
+      "step": 4973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287871286035666e-05,
+      "loss": 0.61,
+      "step": 4974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828709370836819e-05,
+      "loss": 0.5809,
+      "step": 4975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828631597070775e-05,
+      "loss": 0.5896,
+      "step": 4976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828553807306937e-05,
+      "loss": 0.6328,
+      "step": 4977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828476001546806e-05,
+      "loss": 0.5743,
+      "step": 4978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828398179791885e-05,
+      "loss": 0.5903,
+      "step": 4979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828320342043676e-05,
+      "loss": 0.5888,
+      "step": 4980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8282424883036817e-05,
+      "loss": 0.621,
+      "step": 4981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828164618573405e-05,
+      "loss": 0.6453,
+      "step": 4982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280867328543504e-05,
+      "loss": 0.5798,
+      "step": 4983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8280088311480203e-05,
+      "loss": 0.6088,
+      "step": 4984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279309134559187e-05,
+      "loss": 0.5947,
+      "step": 4985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827852979779551e-05,
+      "loss": 0.6353,
+      "step": 4986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827775030120421e-05,
+      "loss": 0.5532,
+      "step": 4987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276970644800337e-05,
+      "loss": 0.6002,
+      "step": 4988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276190828598944e-05,
+      "loss": 0.6162,
+      "step": 4989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275410852615086e-05,
+      "loss": 0.5718,
+      "step": 4990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274630716863826e-05,
+      "loss": 0.6018,
+      "step": 4991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8273850421360217e-05,
+      "loss": 0.647,
+      "step": 4992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8273069966119327e-05,
+      "loss": 0.5958,
+      "step": 4993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827228935115623e-05,
+      "loss": 0.5388,
+      "step": 4994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271508576485987e-05,
+      "loss": 0.6176,
+      "step": 4995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270727642123675e-05,
+      "loss": 0.6027,
+      "step": 4996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269946548084373e-05,
+      "loss": 0.6011,
+      "step": 4997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269165294383162e-05,
+      "loss": 0.5672,
+      "step": 4998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268383881035123e-05,
+      "loss": 0.6163,
+      "step": 4999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267602308055342e-05,
+      "loss": 0.5959,
+      "step": 5000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266820575458908e-05,
+      "loss": 0.5841,
+      "step": 5001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266038683260917e-05,
+      "loss": 0.6174,
+      "step": 5002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826525663147646e-05,
+      "loss": 0.5929,
+      "step": 5003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264474420120634e-05,
+      "loss": 0.5824,
+      "step": 5004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826369204920855e-05,
+      "loss": 0.5629,
+      "step": 5005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82629095187553e-05,
+      "loss": 0.5654,
+      "step": 5006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262126828776e-05,
+      "loss": 0.604,
+      "step": 5007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261343979285756e-05,
+      "loss": 0.5956,
+      "step": 5008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826056097029969e-05,
+      "loss": 0.6019,
+      "step": 5009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8259777801832916e-05,
+      "loss": 0.615,
+      "step": 5010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258994473900544e-05,
+      "loss": 0.6074,
+      "step": 5011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825821098651771e-05,
+      "loss": 0.6248,
+      "step": 5012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257427339699534e-05,
+      "loss": 0.5868,
+      "step": 5013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8256643533461145e-05,
+      "loss": 0.597,
+      "step": 5014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825585956781768e-05,
+      "loss": 0.5887,
+      "step": 5015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825507544278427e-05,
+      "loss": 0.6074,
+      "step": 5016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254291158376053e-05,
+      "loss": 0.6117,
+      "step": 5017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253506714608176e-05,
+      "loss": 0.5857,
+      "step": 5018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8252722111495776e-05,
+      "loss": 0.5932,
+      "step": 5019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825193734905401e-05,
+      "loss": 0.62,
+      "step": 5020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825115242729802e-05,
+      "loss": 0.6297,
+      "step": 5021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8250367346242966e-05,
+      "loss": 0.5789,
+      "step": 5022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249582105904e-05,
+      "loss": 0.5613,
+      "step": 5023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248796706296288e-05,
+      "loss": 0.6129,
+      "step": 5024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248011147434987e-05,
+      "loss": 0.5946,
+      "step": 5025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824722542933527e-05,
+      "loss": 0.5833,
+      "step": 5026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246439552012302e-05,
+      "loss": 0.6059,
+      "step": 5027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824565351548125e-05,
+      "loss": 0.6047,
+      "step": 5028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82448673197573e-05,
+      "loss": 0.5915,
+      "step": 5029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824408096485563e-05,
+      "loss": 0.6082,
+      "step": 5030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8243294450791408e-05,
+      "loss": 0.578,
+      "step": 5031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8242507777579833e-05,
+      "loss": 0.5958,
+      "step": 5032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824172094523609e-05,
+      "loss": 0.5905,
+      "step": 5033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240933953775365e-05,
+      "loss": 0.6075,
+      "step": 5034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240146803212854e-05,
+      "loss": 0.5992,
+      "step": 5035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239359493563755e-05,
+      "loss": 0.6182,
+      "step": 5036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238572024843265e-05,
+      "loss": 0.5618,
+      "step": 5037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237784397066596e-05,
+      "loss": 0.6,
+      "step": 5038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8236996610248943e-05,
+      "loss": 0.6017,
+      "step": 5039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823620866440552e-05,
+      "loss": 0.5508,
+      "step": 5040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823542055955154e-05,
+      "loss": 0.5831,
+      "step": 5041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234632295702218e-05,
+      "loss": 0.6226,
+      "step": 5042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823384387287277e-05,
+      "loss": 0.5936,
+      "step": 5043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8233055291078424e-05,
+      "loss": 0.6216,
+      "step": 5044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232266550334398e-05,
+      "loss": 0.5973,
+      "step": 5045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823147765065592e-05,
+      "loss": 0.5761,
+      "step": 5046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230688592058224e-05,
+      "loss": 0.5972,
+      "step": 5047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822989937455654e-05,
+      "loss": 0.2576,
+      "step": 5048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822910999816611e-05,
+      "loss": 0.5895,
+      "step": 5049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228320462902172e-05,
+      "loss": 0.6513,
+      "step": 5050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227530768779968e-05,
+      "loss": 0.6366,
+      "step": 5051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822674091581474e-05,
+      "loss": 0.5944,
+      "step": 5052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225950904021745e-05,
+      "loss": 0.6115,
+      "step": 5053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822516073341623e-05,
+      "loss": 0.5804,
+      "step": 5054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822437040401345e-05,
+      "loss": 0.6178,
+      "step": 5055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223579915828667e-05,
+      "loss": 0.6204,
+      "step": 5056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222789268877142e-05,
+      "loss": 0.5851,
+      "step": 5057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221998463174137e-05,
+      "loss": 0.5814,
+      "step": 5058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221207498734917e-05,
+      "loss": 0.6234,
+      "step": 5059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220416375574757e-05,
+      "loss": 0.5896,
+      "step": 5060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821962509370893e-05,
+      "loss": 0.6013,
+      "step": 5061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8218833653152716e-05,
+      "loss": 0.5928,
+      "step": 5062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821804205392139e-05,
+      "loss": 0.6128,
+      "step": 5063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217250296030235e-05,
+      "loss": 0.6086,
+      "step": 5064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216458379494538e-05,
+      "loss": 0.6056,
+      "step": 5065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215666304329586e-05,
+      "loss": 0.601,
+      "step": 5066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214874070550675e-05,
+      "loss": 0.603,
+      "step": 5067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214081678173098e-05,
+      "loss": 0.6253,
+      "step": 5068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8213289127212152e-05,
+      "loss": 0.6294,
+      "step": 5069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212496417683135e-05,
+      "loss": 0.5907,
+      "step": 5070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211703549601362e-05,
+      "loss": 0.6093,
+      "step": 5071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821091052298213e-05,
+      "loss": 0.5999,
+      "step": 5072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210117337840755e-05,
+      "loss": 0.6207,
+      "step": 5073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820932399419255e-05,
+      "loss": 0.6142,
+      "step": 5074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820853049205283e-05,
+      "loss": 0.6211,
+      "step": 5075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207736831436913e-05,
+      "loss": 0.6125,
+      "step": 5076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206943012360123e-05,
+      "loss": 0.5677,
+      "step": 5077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206149034837787e-05,
+      "loss": 0.609,
+      "step": 5078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8205354898885236e-05,
+      "loss": 0.6053,
+      "step": 5079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204560604517795e-05,
+      "loss": 0.6242,
+      "step": 5080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203766151750805e-05,
+      "loss": 0.5528,
+      "step": 5081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202971540599597e-05,
+      "loss": 0.5828,
+      "step": 5082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820217677107952e-05,
+      "loss": 0.5925,
+      "step": 5083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8201381843205913e-05,
+      "loss": 0.5811,
+      "step": 5084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8200586756994126e-05,
+      "loss": 0.6403,
+      "step": 5085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199791512459507e-05,
+      "loss": 0.5611,
+      "step": 5086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198996109617412e-05,
+      "loss": 0.596,
+      "step": 5087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198200548483193e-05,
+      "loss": 0.6111,
+      "step": 5088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197404829072214e-05,
+      "loss": 0.5714,
+      "step": 5089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8196608951399834e-05,
+      "loss": 0.6061,
+      "step": 5090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195812915481418e-05,
+      "loss": 0.5919,
+      "step": 5091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195016721332334e-05,
+      "loss": 0.557,
+      "step": 5092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819422036896796e-05,
+      "loss": 0.5829,
+      "step": 5093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193423858403664e-05,
+      "loss": 0.617,
+      "step": 5094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192627189654827e-05,
+      "loss": 0.6332,
+      "step": 5095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191830362736828e-05,
+      "loss": 0.6127,
+      "step": 5096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191033377665048e-05,
+      "loss": 0.6187,
+      "step": 5097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819023623445488e-05,
+      "loss": 0.5913,
+      "step": 5098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818943893312171e-05,
+      "loss": 0.5832,
+      "step": 5099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8188641473680936e-05,
+      "loss": 0.6189,
+      "step": 5100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8187843856147943e-05,
+      "loss": 0.5846,
+      "step": 5101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818704608053814e-05,
+      "loss": 0.608,
+      "step": 5102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186248146866928e-05,
+      "loss": 0.5999,
+      "step": 5103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8185450055149706e-05,
+      "loss": 0.6019,
+      "step": 5104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818465180540189e-05,
+      "loss": 0.6461,
+      "step": 5105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8183853397638884e-05,
+      "loss": 0.5827,
+      "step": 5106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8183054831876105e-05,
+      "loss": 0.6082,
+      "step": 5107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182256108128973e-05,
+      "loss": 0.6295,
+      "step": 5108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181457226412907e-05,
+      "loss": 0.6354,
+      "step": 5109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8180658186743333e-05,
+      "loss": 0.5733,
+      "step": 5110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817985898913567e-05,
+      "loss": 0.601,
+      "step": 5111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179059633605352e-05,
+      "loss": 0.6346,
+      "step": 5112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817826012016781e-05,
+      "loss": 0.605,
+      "step": 5113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177460448838484e-05,
+      "loss": 0.619,
+      "step": 5114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817666061963281e-05,
+      "loss": 0.5919,
+      "step": 5115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817586063256623e-05,
+      "loss": 0.5977,
+      "step": 5116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175060487654187e-05,
+      "loss": 0.5756,
+      "step": 5117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817426018491213e-05,
+      "loss": 0.5537,
+      "step": 5118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817345972435551e-05,
+      "loss": 0.6105,
+      "step": 5119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817265910599978e-05,
+      "loss": 0.6176,
+      "step": 5120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171858329860403e-05,
+      "loss": 0.5868,
+      "step": 5121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171057395952827e-05,
+      "loss": 0.5712,
+      "step": 5122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170256304292525e-05,
+      "loss": 0.5702,
+      "step": 5123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816945505489496e-05,
+      "loss": 0.5924,
+      "step": 5124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81686536477756e-05,
+      "loss": 0.6287,
+      "step": 5125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816785208294992e-05,
+      "loss": 0.6044,
+      "step": 5126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167050360433394e-05,
+      "loss": 0.5548,
+      "step": 5127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8166248480241498e-05,
+      "loss": 0.6014,
+      "step": 5128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165446442389716e-05,
+      "loss": 0.6219,
+      "step": 5129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164644246893532e-05,
+      "loss": 0.6343,
+      "step": 5130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8163841893768433e-05,
+      "loss": 0.5757,
+      "step": 5131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816303938302991e-05,
+      "loss": 0.618,
+      "step": 5132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162236714693457e-05,
+      "loss": 0.5834,
+      "step": 5133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161433888774564e-05,
+      "loss": 0.6324,
+      "step": 5134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816063090528874e-05,
+      "loss": 0.5859,
+      "step": 5135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159827764251485e-05,
+      "loss": 0.5836,
+      "step": 5136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81590244656783e-05,
+      "loss": 0.6055,
+      "step": 5137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81582210095847e-05,
+      "loss": 0.5901,
+      "step": 5138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815741739598619e-05,
+      "loss": 0.5942,
+      "step": 5139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156613624898293e-05,
+      "loss": 0.5906,
+      "step": 5140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815580969633652e-05,
+      "loss": 0.5958,
+      "step": 5141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155005610316395e-05,
+      "loss": 0.5797,
+      "step": 5142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815420136685344e-05,
+      "loss": 0.6261,
+      "step": 5143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153396965963187e-05,
+      "loss": 0.2708,
+      "step": 5144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815259240766116e-05,
+      "loss": 0.6025,
+      "step": 5145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815178769196289e-05,
+      "loss": 0.604,
+      "step": 5146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150982818883922e-05,
+      "loss": 0.5703,
+      "step": 5147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815017778843979e-05,
+      "loss": 0.6249,
+      "step": 5148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149372600646035e-05,
+      "loss": 0.5877,
+      "step": 5149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8148567255518203e-05,
+      "loss": 0.5787,
+      "step": 5150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147761753071846e-05,
+      "loss": 0.5803,
+      "step": 5151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814695609332251e-05,
+      "loss": 0.6019,
+      "step": 5152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814615027628575e-05,
+      "loss": 0.5774,
+      "step": 5153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145344301977126e-05,
+      "loss": 0.5752,
+      "step": 5154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81445381704122e-05,
+      "loss": 0.6192,
+      "step": 5155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8143731881606534e-05,
+      "loss": 0.5695,
+      "step": 5156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8142925435575686e-05,
+      "loss": 0.5785,
+      "step": 5157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8142118832335235e-05,
+      "loss": 0.5648,
+      "step": 5158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141312071900752e-05,
+      "loss": 0.5614,
+      "step": 5159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814050515428781e-05,
+      "loss": 0.6014,
+      "step": 5160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139698079511992e-05,
+      "loss": 0.6767,
+      "step": 5161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8138890847588878e-05,
+      "loss": 0.6139,
+      "step": 5162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8138083458534047e-05,
+      "loss": 0.6106,
+      "step": 5163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137275912363092e-05,
+      "loss": 0.594,
+      "step": 5164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81364682090916e-05,
+      "loss": 0.5924,
+      "step": 5165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8135660348735174e-05,
+      "loss": 0.5873,
+      "step": 5166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81348523313094e-05,
+      "loss": 0.5569,
+      "step": 5167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8134044156829885e-05,
+      "loss": 0.5845,
+      "step": 5168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133235825312224e-05,
+      "loss": 0.6034,
+      "step": 5169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132427336772033e-05,
+      "loss": 0.5796,
+      "step": 5170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8131618691224916e-05,
+      "loss": 0.6101,
+      "step": 5171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813080988868648e-05,
+      "loss": 0.5967,
+      "step": 5172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8130000929172345e-05,
+      "loss": 0.5873,
+      "step": 5173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812919181269813e-05,
+      "loss": 0.5983,
+      "step": 5174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8128382539279455e-05,
+      "loss": 0.5891,
+      "step": 5175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127573108931946e-05,
+      "loss": 0.622,
+      "step": 5176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8126763521671224e-05,
+      "loss": 0.6198,
+      "step": 5177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125953777512925e-05,
+      "loss": 0.6123,
+      "step": 5178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812514387647268e-05,
+      "loss": 0.5922,
+      "step": 5179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124333818566124e-05,
+      "loss": 0.5901,
+      "step": 5180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81235236038089e-05,
+      "loss": 0.6013,
+      "step": 5181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8122713232216645e-05,
+      "loss": 0.5893,
+      "step": 5182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121902703805008e-05,
+      "loss": 0.6322,
+      "step": 5183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121092018589635e-05,
+      "loss": 0.5923,
+      "step": 5184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8120281176586178e-05,
+      "loss": 0.6475,
+      "step": 5185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119470177810298e-05,
+      "loss": 0.6063,
+      "step": 5186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811865902227764e-05,
+      "loss": 0.5606,
+      "step": 5187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811784771000387e-05,
+      "loss": 0.5941,
+      "step": 5188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117036241004656e-05,
+      "loss": 0.6166,
+      "step": 5189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811622461529566e-05,
+      "loss": 0.5709,
+      "step": 5190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811541283289255e-05,
+      "loss": 0.6169,
+      "step": 5191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114600893811e-05,
+      "loss": 0.565,
+      "step": 5192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811378879806669e-05,
+      "loss": 0.6085,
+      "step": 5193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811297654567529e-05,
+      "loss": 0.5741,
+      "step": 5194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112164136652487e-05,
+      "loss": 0.6308,
+      "step": 5195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8111351571013966e-05,
+      "loss": 0.5984,
+      "step": 5196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110538848775415e-05,
+      "loss": 0.5698,
+      "step": 5197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810972596995252e-05,
+      "loss": 0.2743,
+      "step": 5198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810891293456098e-05,
+      "loss": 0.6159,
+      "step": 5199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108099742616486e-05,
+      "loss": 0.6212,
+      "step": 5200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107286394134744e-05,
+      "loss": 0.6198,
+      "step": 5201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8106472889131454e-05,
+      "loss": 0.6164,
+      "step": 5202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810565922762232e-05,
+      "loss": 0.5989,
+      "step": 5203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104845409623056e-05,
+      "loss": 0.5935,
+      "step": 5204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104031435149366e-05,
+      "loss": 0.6215,
+      "step": 5205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810321730421697e-05,
+      "loss": 0.5563,
+      "step": 5206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102403016841586e-05,
+      "loss": 0.5944,
+      "step": 5207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101588573038934e-05,
+      "loss": 0.6417,
+      "step": 5208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100773972824734e-05,
+      "loss": 0.5956,
+      "step": 5209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809995921621472e-05,
+      "loss": 0.5809,
+      "step": 5210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099144303224613e-05,
+      "loss": 0.5939,
+      "step": 5211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809832923387016e-05,
+      "loss": 0.6039,
+      "step": 5212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809751400816708e-05,
+      "loss": 0.5874,
+      "step": 5213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8096698626131126e-05,
+      "loss": 0.6079,
+      "step": 5214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809588308777803e-05,
+      "loss": 0.6055,
+      "step": 5215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095067393123544e-05,
+      "loss": 0.6503,
+      "step": 5216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809425154218341e-05,
+      "loss": 0.6039,
+      "step": 5217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8093435534973386e-05,
+      "loss": 0.6112,
+      "step": 5218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8092619371509222e-05,
+      "loss": 0.599,
+      "step": 5219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8091803051806676e-05,
+      "loss": 0.2549,
+      "step": 5220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090986575881504e-05,
+      "loss": 0.5847,
+      "step": 5221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.5878,
+      "step": 5222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8089353155426353e-05,
+      "loss": 0.5903,
+      "step": 5223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088536210927907e-05,
+      "loss": 0.6059,
+      "step": 5224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808771911026991e-05,
+      "loss": 0.5869,
+      "step": 5225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086901853468135e-05,
+      "loss": 0.5808,
+      "step": 5226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086084440538362e-05,
+      "loss": 0.5835,
+      "step": 5227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8085266871496368e-05,
+      "loss": 0.6133,
+      "step": 5228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084449146357945e-05,
+      "loss": 0.6171,
+      "step": 5229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083631265138875e-05,
+      "loss": 0.643,
+      "step": 5230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808281322785495e-05,
+      "loss": 0.6225,
+      "step": 5231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081995034521963e-05,
+      "loss": 0.6303,
+      "step": 5232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081176685155708e-05,
+      "loss": 0.6193,
+      "step": 5233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8080358179771988e-05,
+      "loss": 0.2795,
+      "step": 5234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079539518386605e-05,
+      "loss": 0.5984,
+      "step": 5235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8078720701015358e-05,
+      "loss": 0.5892,
+      "step": 5236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077901727674063e-05,
+      "loss": 0.5949,
+      "step": 5237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807708259837853e-05,
+      "loss": 0.5868,
+      "step": 5238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076263313144568e-05,
+      "loss": 0.6218,
+      "step": 5239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075443871988e-05,
+      "loss": 0.5914,
+      "step": 5240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8074624274924645e-05,
+      "loss": 0.564,
+      "step": 5241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073804521970326e-05,
+      "loss": 0.6475,
+      "step": 5242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072984613140866e-05,
+      "loss": 0.5776,
+      "step": 5243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80721645484521e-05,
+      "loss": 0.6011,
+      "step": 5244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.807134432791986e-05,
+      "loss": 0.602,
+      "step": 5245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070523951559976e-05,
+      "loss": 0.6014,
+      "step": 5246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806970341938829e-05,
+      "loss": 0.634,
+      "step": 5247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068882731420648e-05,
+      "loss": 0.5891,
+      "step": 5248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068061887672884e-05,
+      "loss": 0.6451,
+      "step": 5249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067240888160854e-05,
+      "loss": 0.6186,
+      "step": 5250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066419732900407e-05,
+      "loss": 0.6266,
+      "step": 5251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065598421907392e-05,
+      "loss": 0.5865,
+      "step": 5252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806477695519767e-05,
+      "loss": 0.5662,
+      "step": 5253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80639553327871e-05,
+      "loss": 0.5827,
+      "step": 5254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806313355469154e-05,
+      "loss": 0.5455,
+      "step": 5255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806231162092686e-05,
+      "loss": 0.2511,
+      "step": 5256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061489531508932e-05,
+      "loss": 0.6006,
+      "step": 5257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8060667286453618e-05,
+      "loss": 0.5862,
+      "step": 5258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80598448857768e-05,
+      "loss": 0.5995,
+      "step": 5259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805902232949435e-05,
+      "loss": 0.6307,
+      "step": 5260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058199617622154e-05,
+      "loss": 0.6148,
+      "step": 5261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057376750176094e-05,
+      "loss": 0.6038,
+      "step": 5262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805655372717205e-05,
+      "loss": 0.6009,
+      "step": 5263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805573054862592e-05,
+      "loss": 0.6198,
+      "step": 5264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054907214553593e-05,
+      "loss": 0.602,
+      "step": 5265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8054083724970967e-05,
+      "loss": 0.6023,
+      "step": 5266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053260079893936e-05,
+      "loss": 0.5879,
+      "step": 5267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052436279338407e-05,
+      "loss": 0.592,
+      "step": 5268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805161232332028e-05,
+      "loss": 0.5838,
+      "step": 5269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050788211855463e-05,
+      "loss": 0.5889,
+      "step": 5270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049963944959867e-05,
+      "loss": 0.624,
+      "step": 5271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049139522649408e-05,
+      "loss": 0.6094,
+      "step": 5272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804831494494e-05,
+      "loss": 0.6316,
+      "step": 5273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8047490211847564e-05,
+      "loss": 0.6063,
+      "step": 5274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046665323388023e-05,
+      "loss": 0.5937,
+      "step": 5275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8045840279577297e-05,
+      "loss": 0.6255,
+      "step": 5276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804501508043132e-05,
+      "loss": 0.5873,
+      "step": 5277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044189725966022e-05,
+      "loss": 0.6192,
+      "step": 5278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043364216197342e-05,
+      "loss": 0.6103,
+      "step": 5279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042538551141207e-05,
+      "loss": 0.5827,
+      "step": 5280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804171273081357e-05,
+      "loss": 0.59,
+      "step": 5281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040886755230365e-05,
+      "loss": 0.5918,
+      "step": 5282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040060624407543e-05,
+      "loss": 0.6078,
+      "step": 5283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803923433836105e-05,
+      "loss": 0.6165,
+      "step": 5284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8038407897106844e-05,
+      "loss": 0.6495,
+      "step": 5285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037581300660877e-05,
+      "loss": 0.5996,
+      "step": 5286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036754549039107e-05,
+      "loss": 0.602,
+      "step": 5287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035927642257496e-05,
+      "loss": 0.5796,
+      "step": 5288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803510058033201e-05,
+      "loss": 0.6159,
+      "step": 5289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8034273363278615e-05,
+      "loss": 0.6049,
+      "step": 5290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803344599111328e-05,
+      "loss": 0.5938,
+      "step": 5291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8032618463851983e-05,
+      "loss": 0.6182,
+      "step": 5292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031790781510695e-05,
+      "loss": 0.6416,
+      "step": 5293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80309629441054e-05,
+      "loss": 0.5606,
+      "step": 5294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030134951652082e-05,
+      "loss": 0.5957,
+      "step": 5295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802930680416672e-05,
+      "loss": 0.5827,
+      "step": 5296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028478501665304e-05,
+      "loss": 0.5959,
+      "step": 5297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802765004416383e-05,
+      "loss": 0.5646,
+      "step": 5298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802682143167829e-05,
+      "loss": 0.5875,
+      "step": 5299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025992664224675e-05,
+      "loss": 0.6184,
+      "step": 5300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025163741818997e-05,
+      "loss": 0.5747,
+      "step": 5301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802433466447725e-05,
+      "loss": 0.6391,
+      "step": 5302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802350543221545e-05,
+      "loss": 0.5627,
+      "step": 5303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8022676045049597e-05,
+      "loss": 0.5793,
+      "step": 5304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802184650299571e-05,
+      "loss": 0.6215,
+      "step": 5305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8021016806069796e-05,
+      "loss": 0.5511,
+      "step": 5306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020186954287883e-05,
+      "loss": 0.5871,
+      "step": 5307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8019356947665986e-05,
+      "loss": 0.6194,
+      "step": 5308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801852678622013e-05,
+      "loss": 0.6021,
+      "step": 5309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801769646996635e-05,
+      "loss": 0.5946,
+      "step": 5310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016865998920664e-05,
+      "loss": 0.6311,
+      "step": 5311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801603537309911e-05,
+      "loss": 0.5982,
+      "step": 5312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801520459251773e-05,
+      "loss": 0.5819,
+      "step": 5313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801437365719256e-05,
+      "loss": 0.5954,
+      "step": 5314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013542567139635e-05,
+      "loss": 0.5531,
+      "step": 5315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8012711322375015e-05,
+      "loss": 0.5709,
+      "step": 5316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801187992291473e-05,
+      "loss": 0.2606,
+      "step": 5317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011048368774847e-05,
+      "loss": 0.6157,
+      "step": 5318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8010216659971416e-05,
+      "loss": 0.6077,
+      "step": 5319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009384796520487e-05,
+      "loss": 0.6303,
+      "step": 5320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800855277843813e-05,
+      "loss": 0.6112,
+      "step": 5321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80077206057404e-05,
+      "loss": 0.5837,
+      "step": 5322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8006888278443368e-05,
+      "loss": 0.6149,
+      "step": 5323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8006055796563103e-05,
+      "loss": 0.5555,
+      "step": 5324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005223160115676e-05,
+      "loss": 0.6391,
+      "step": 5325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8004390369117164e-05,
+      "loss": 0.6183,
+      "step": 5326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800355742358364e-05,
+      "loss": 0.5994,
+      "step": 5327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800272432353119e-05,
+      "loss": 0.6599,
+      "step": 5328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001891068975895e-05,
+      "loss": 0.5789,
+      "step": 5329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001057659933844e-05,
+      "loss": 0.5893,
+      "step": 5330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800022409642113e-05,
+      "loss": 0.6203,
+      "step": 5331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799939037845384e-05,
+      "loss": 0.5768,
+      "step": 5332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799855650604807e-05,
+      "loss": 0.6055,
+      "step": 5333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7997722479219927e-05,
+      "loss": 0.6097,
+      "step": 5334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996888297985504e-05,
+      "loss": 0.6159,
+      "step": 5335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996053962360908e-05,
+      "loss": 0.5835,
+      "step": 5336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7995219472362254e-05,
+      "loss": 0.6342,
+      "step": 5337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799438482800564e-05,
+      "loss": 0.595,
+      "step": 5338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799355002930719e-05,
+      "loss": 0.6229,
+      "step": 5339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992715076283024e-05,
+      "loss": 0.6003,
+      "step": 5340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7991879968949248e-05,
+      "loss": 0.6268,
+      "step": 5341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7991044707322e-05,
+      "loss": 0.5741,
+      "step": 5342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799020929141739e-05,
+      "loss": 0.5736,
+      "step": 5343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798937372125156e-05,
+      "loss": 0.5781,
+      "step": 5344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988537996840634e-05,
+      "loss": 0.5699,
+      "step": 5345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987702118200754e-05,
+      "loss": 0.6137,
+      "step": 5346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7986866085348052e-05,
+      "loss": 0.6195,
+      "step": 5347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798602989829867e-05,
+      "loss": 0.6187,
+      "step": 5348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798519355706875e-05,
+      "loss": 0.5845,
+      "step": 5349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984357061674443e-05,
+      "loss": 0.5516,
+      "step": 5350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983520412131893e-05,
+      "loss": 0.5702,
+      "step": 5351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982683608457258e-05,
+      "loss": 0.5938,
+      "step": 5352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798184665066669e-05,
+      "loss": 0.5719,
+      "step": 5353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798100953877635e-05,
+      "loss": 0.6189,
+      "step": 5354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980172272802398e-05,
+      "loss": 0.6016,
+      "step": 5355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979334852760996e-05,
+      "loss": 0.254,
+      "step": 5356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978497278668316e-05,
+      "loss": 0.5893,
+      "step": 5357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797765955054053e-05,
+      "loss": 0.6103,
+      "step": 5358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79768216683938e-05,
+      "loss": 0.6176,
+      "step": 5359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975983632244317e-05,
+      "loss": 0.6322,
+      "step": 5360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975145442108254e-05,
+      "loss": 0.6091,
+      "step": 5361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797430709800179e-05,
+      "loss": 0.6035,
+      "step": 5362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973468599941116e-05,
+      "loss": 0.5778,
+      "step": 5363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972629947942416e-05,
+      "loss": 0.5935,
+      "step": 5364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971791142021887e-05,
+      "loss": 0.6201,
+      "step": 5365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970952182195713e-05,
+      "loss": 0.5817,
+      "step": 5366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970113068480098e-05,
+      "loss": 0.6197,
+      "step": 5367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969273800891245e-05,
+      "loss": 0.6221,
+      "step": 5368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7968434379445355e-05,
+      "loss": 0.6138,
+      "step": 5369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796759480415863e-05,
+      "loss": 0.589,
+      "step": 5370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796675507504728e-05,
+      "loss": 0.607,
+      "step": 5371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7965915192127524e-05,
+      "loss": 0.5668,
+      "step": 5372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7965075155415567e-05,
+      "loss": 0.5839,
+      "step": 5373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964234964927634e-05,
+      "loss": 0.608,
+      "step": 5374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7963394620679945e-05,
+      "loss": 0.5809,
+      "step": 5375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796255412268872e-05,
+      "loss": 0.5547,
+      "step": 5376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796171347097019e-05,
+      "loss": 0.6057,
+      "step": 5377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960872665540586e-05,
+      "loss": 0.5919,
+      "step": 5378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960031706416136e-05,
+      "loss": 0.5803,
+      "step": 5379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7959190593613075e-05,
+      "loss": 0.6126,
+      "step": 5380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795834932714765e-05,
+      "loss": 0.5946,
+      "step": 5381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7957507907036095e-05,
+      "loss": 0.589,
+      "step": 5382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795666633329466e-05,
+      "loss": 0.6368,
+      "step": 5383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795582460593959e-05,
+      "loss": 0.589,
+      "step": 5384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954982724987132e-05,
+      "loss": 0.5962,
+      "step": 5385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954140690453544e-05,
+      "loss": 0.59,
+      "step": 5386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7953298502355083e-05,
+      "loss": 0.6177,
+      "step": 5387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952456160708006e-05,
+      "loss": 0.5797,
+      "step": 5388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951613665528577e-05,
+      "loss": 0.6046,
+      "step": 5389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950771016833064e-05,
+      "loss": 0.6166,
+      "step": 5390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949928214637734e-05,
+      "loss": 0.56,
+      "step": 5391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949085258958853e-05,
+      "loss": 0.5825,
+      "step": 5392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79482421498127e-05,
+      "loss": 0.6071,
+      "step": 5393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794739888721555e-05,
+      "loss": 0.6417,
+      "step": 5394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7946555471183685e-05,
+      "loss": 0.647,
+      "step": 5395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945711901733393e-05,
+      "loss": 0.6345,
+      "step": 5396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7944868178880952e-05,
+      "loss": 0.6182,
+      "step": 5397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7944024302642655e-05,
+      "loss": 0.5948,
+      "step": 5398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943180273034792e-05,
+      "loss": 0.5842,
+      "step": 5399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794233609007366e-05,
+      "loss": 0.5987,
+      "step": 5400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941491753775558e-05,
+      "loss": 0.5995,
+      "step": 5401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940647264156787e-05,
+      "loss": 0.6123,
+      "step": 5402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939802621233648e-05,
+      "loss": 0.5768,
+      "step": 5403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793895782502245e-05,
+      "loss": 0.6131,
+      "step": 5404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79381128755395e-05,
+      "loss": 0.5877,
+      "step": 5405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7937267772801116e-05,
+      "loss": 0.5887,
+      "step": 5406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793642251682361e-05,
+      "loss": 0.6197,
+      "step": 5407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935577107623304e-05,
+      "loss": 0.5835,
+      "step": 5408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934731545216515e-05,
+      "loss": 0.5924,
+      "step": 5409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793388582961957e-05,
+      "loss": 0.6134,
+      "step": 5410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79330399608488e-05,
+      "loss": 0.6356,
+      "step": 5411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932193938920526e-05,
+      "loss": 0.5779,
+      "step": 5412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931347763851094e-05,
+      "loss": 0.5989,
+      "step": 5413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930501435656834e-05,
+      "loss": 0.592,
+      "step": 5414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929654954354085e-05,
+      "loss": 0.6131,
+      "step": 5415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7928808319959186e-05,
+      "loss": 0.6258,
+      "step": 5416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792796153248849e-05,
+      "loss": 0.5946,
+      "step": 5417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927114591958345e-05,
+      "loss": 0.6104,
+      "step": 5418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7926267498385094e-05,
+      "loss": 0.5795,
+      "step": 5419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79254202517851e-05,
+      "loss": 0.613,
+      "step": 5420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924572852174716e-05,
+      "loss": 0.6199,
+      "step": 5421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7923725299570303e-05,
+      "loss": 0.6222,
+      "step": 5422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922877593988222e-05,
+      "loss": 0.5797,
+      "step": 5423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922029735444845e-05,
+      "loss": 0.6549,
+      "step": 5424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792118172395653e-05,
+      "loss": 0.546,
+      "step": 5425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792033355953966e-05,
+      "loss": 0.6197,
+      "step": 5426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919485242210605e-05,
+      "loss": 0.6073,
+      "step": 5427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7918636771985743e-05,
+      "loss": 0.5999,
+      "step": 5428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791778814888146e-05,
+      "loss": 0.5628,
+      "step": 5429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791693937291413e-05,
+      "loss": 0.5898,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7916090444100145e-05,
+      "loss": 0.6018,
+      "step": 5431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7915241362455898e-05,
+      "loss": 0.59,
+      "step": 5432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914392127997775e-05,
+      "loss": 0.6334,
+      "step": 5433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913542740742176e-05,
+      "loss": 0.5736,
+      "step": 5434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79126932007055e-05,
+      "loss": 0.6111,
+      "step": 5435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7911843507904145e-05,
+      "loss": 0.5917,
+      "step": 5436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910993662354516e-05,
+      "loss": 0.621,
+      "step": 5437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910143664073022e-05,
+      "loss": 0.6146,
+      "step": 5438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790929351307607e-05,
+      "loss": 0.5613,
+      "step": 5439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790844320938008e-05,
+      "loss": 0.5801,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907592753001464e-05,
+      "loss": 0.5983,
+      "step": 5441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790674214395664e-05,
+      "loss": 0.5894,
+      "step": 5442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790589138226203e-05,
+      "loss": 0.6106,
+      "step": 5443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905040467934066e-05,
+      "loss": 0.6445,
+      "step": 5444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7904189400989165e-05,
+      "loss": 0.6223,
+      "step": 5445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903338181443764e-05,
+      "loss": 0.6009,
+      "step": 5446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902486809314298e-05,
+      "loss": 0.5999,
+      "step": 5447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79016352846172e-05,
+      "loss": 0.6278,
+      "step": 5448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900783607368915e-05,
+      "loss": 0.5821,
+      "step": 5449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789993177758588e-05,
+      "loss": 0.5465,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899079795284545e-05,
+      "loss": 0.5687,
+      "step": 5451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789822766048136e-05,
+      "loss": 0.6105,
+      "step": 5452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897375373192763e-05,
+      "loss": 0.5951,
+      "step": 5453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7896522933435228e-05,
+      "loss": 0.6158,
+      "step": 5454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78956703412252e-05,
+      "loss": 0.5744,
+      "step": 5455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894817596579145e-05,
+      "loss": 0.6093,
+      "step": 5456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893964699513523e-05,
+      "loss": 0.5764,
+      "step": 5457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78931116500448e-05,
+      "loss": 0.5989,
+      "step": 5458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789225844818945e-05,
+      "loss": 0.5936,
+      "step": 5459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.6175,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890551587384742e-05,
+      "loss": 0.5973,
+      "step": 5461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889697928468342e-05,
+      "loss": 0.5828,
+      "step": 5462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888844117231224e-05,
+      "loss": 0.621,
+      "step": 5463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887990153689857e-05,
+      "loss": 0.5907,
+      "step": 5464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887136037860742e-05,
+      "loss": 0.5889,
+      "step": 5465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788628176976036e-05,
+      "loss": 0.5975,
+      "step": 5466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788542734940521e-05,
+      "loss": 0.6125,
+      "step": 5467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7884572776811783e-05,
+      "loss": 0.6061,
+      "step": 5468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788371805199658e-05,
+      "loss": 0.6373,
+      "step": 5469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882863174976105e-05,
+      "loss": 0.5551,
+      "step": 5470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882008145766853e-05,
+      "loss": 0.5907,
+      "step": 5471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7881152964385343e-05,
+      "loss": 0.61,
+      "step": 5472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788029763084808e-05,
+      "loss": 0.6086,
+      "step": 5473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879442145171576e-05,
+      "loss": 0.5497,
+      "step": 5474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787858650737235e-05,
+      "loss": 0.6282,
+      "step": 5475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787773071746692e-05,
+      "loss": 0.5892,
+      "step": 5476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7876874775471806e-05,
+      "loss": 0.5728,
+      "step": 5477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7876018681403534e-05,
+      "loss": 0.5926,
+      "step": 5478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7875162435278638e-05,
+      "loss": 0.6325,
+      "step": 5479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874306037113642e-05,
+      "loss": 0.626,
+      "step": 5480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873449486925084e-05,
+      "loss": 0.5695,
+      "step": 5481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7872592784729497e-05,
+      "loss": 0.629,
+      "step": 5482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7871735930543422e-05,
+      "loss": 0.6059,
+      "step": 5483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870878924383405e-05,
+      "loss": 0.5745,
+      "step": 5484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870021766265985e-05,
+      "loss": 0.5704,
+      "step": 5485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7869164456207713e-05,
+      "loss": 0.5931,
+      "step": 5486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786830699422515e-05,
+      "loss": 0.6067,
+      "step": 5487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7867449380334834e-05,
+      "loss": 0.6141,
+      "step": 5488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866591614553335e-05,
+      "loss": 0.5956,
+      "step": 5489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7865733696897205e-05,
+      "loss": 0.6104,
+      "step": 5490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864875627383017e-05,
+      "loss": 0.5951,
+      "step": 5491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864017406027324e-05,
+      "loss": 0.5941,
+      "step": 5492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7863159032846708e-05,
+      "loss": 0.5395,
+      "step": 5493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862300507857733e-05,
+      "loss": 0.6131,
+      "step": 5494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861441831076975e-05,
+      "loss": 0.6231,
+      "step": 5495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860583002521014e-05,
+      "loss": 0.6314,
+      "step": 5496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785972402220643e-05,
+      "loss": 0.6186,
+      "step": 5497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785886489014981e-05,
+      "loss": 0.5944,
+      "step": 5498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785800560636773e-05,
+      "loss": 0.5965,
+      "step": 5499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785714617087679e-05,
+      "loss": 0.5784,
+      "step": 5500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7856286583693577e-05,
+      "loss": 0.5873,
+      "step": 5501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7855426844834696e-05,
+      "loss": 0.5922,
+      "step": 5502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785456695431673e-05,
+      "loss": 0.6104,
+      "step": 5503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785370691215629e-05,
+      "loss": 0.6153,
+      "step": 5504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785284671836998e-05,
+      "loss": 0.6089,
+      "step": 5505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7851986372974407e-05,
+      "loss": 0.6086,
+      "step": 5506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785112587598618e-05,
+      "loss": 0.5693,
+      "step": 5507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850265227421907e-05,
+      "loss": 0.5777,
+      "step": 5508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7849404427298215e-05,
+      "loss": 0.6049,
+      "step": 5509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784854347563171e-05,
+      "loss": 0.6122,
+      "step": 5510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847682372439024e-05,
+      "loss": 0.5726,
+      "step": 5511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846821117736777e-05,
+      "loss": 0.6044,
+      "step": 5512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845959711541597e-05,
+      "loss": 0.6148,
+      "step": 5513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784509815387012e-05,
+      "loss": 0.6085,
+      "step": 5514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784423644473897e-05,
+      "loss": 0.607,
+      "step": 5515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7843374584164788e-05,
+      "loss": 0.5948,
+      "step": 5516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7842512572164214e-05,
+      "loss": 0.5507,
+      "step": 5517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841650408753888e-05,
+      "loss": 0.625,
+      "step": 5518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784078809395046e-05,
+      "loss": 0.6167,
+      "step": 5519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783992562777057e-05,
+      "loss": 0.2631,
+      "step": 5520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783906301023088e-05,
+      "loss": 0.5997,
+      "step": 5521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7838200241348033e-05,
+      "loss": 0.616,
+      "step": 5522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7837337321138695e-05,
+      "loss": 0.5829,
+      "step": 5523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783647424961952e-05,
+      "loss": 0.6198,
+      "step": 5524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783561102680717e-05,
+      "loss": 0.604,
+      "step": 5525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834747652718313e-05,
+      "loss": 0.6014,
+      "step": 5526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833884127369617e-05,
+      "loss": 0.5767,
+      "step": 5527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833020450777756e-05,
+      "loss": 0.6358,
+      "step": 5528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7832156622959402e-05,
+      "loss": 0.5945,
+      "step": 5529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783129264393123e-05,
+      "loss": 0.6001,
+      "step": 5530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7830428513709925e-05,
+      "loss": 0.5969,
+      "step": 5531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7829564232312163e-05,
+      "loss": 0.616,
+      "step": 5532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7828699799754643e-05,
+      "loss": 0.6062,
+      "step": 5533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782783521605404e-05,
+      "loss": 0.6031,
+      "step": 5534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826970481227053e-05,
+      "loss": 0.5827,
+      "step": 5535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826105595290373e-05,
+      "loss": 0.5821,
+      "step": 5536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825240558260706e-05,
+      "loss": 0.5871,
+      "step": 5537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824375370154743e-05,
+      "loss": 0.5645,
+      "step": 5538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823510030989192e-05,
+      "loss": 0.5937,
+      "step": 5539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7822644540780763e-05,
+      "loss": 0.6042,
+      "step": 5540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7821778899546157e-05,
+      "loss": 0.5866,
+      "step": 5541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7820913107302093e-05,
+      "loss": 0.6157,
+      "step": 5542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7820047164065282e-05,
+      "loss": 0.5853,
+      "step": 5543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819181069852444e-05,
+      "loss": 0.5819,
+      "step": 5544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.6157,
+      "step": 5545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7817448428565574e-05,
+      "loss": 0.5901,
+      "step": 5546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7816581881524996e-05,
+      "loss": 0.5894,
+      "step": 5547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781571518357529e-05,
+      "loss": 0.5629,
+      "step": 5548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7814848334733193e-05,
+      "loss": 0.604,
+      "step": 5549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813981335015436e-05,
+      "loss": 0.6021,
+      "step": 5550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813114184438763e-05,
+      "loss": 0.6339,
+      "step": 5551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812246883019912e-05,
+      "loss": 0.5875,
+      "step": 5552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781137943077563e-05,
+      "loss": 0.5602,
+      "step": 5553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7810511827722665e-05,
+      "loss": 0.6115,
+      "step": 5554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.780964407387776e-05,
+      "loss": 0.2583,
+      "step": 5555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7808776169257673e-05,
+      "loss": 0.6165,
+      "step": 5556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807908113879163e-05,
+      "loss": 0.6308,
+      "step": 5557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807039907758986e-05,
+      "loss": 0.5832,
+      "step": 5558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.78061715509139e-05,
+      "loss": 0.5955,
+      "step": 5559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780530304336068e-05,
+      "loss": 0.6038,
+      "step": 5560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7804434385116084e-05,
+      "loss": 0.5582,
+      "step": 5561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7803565576196884e-05,
+      "loss": 0.589,
+      "step": 5562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802696616619857e-05,
+      "loss": 0.5839,
+      "step": 5563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801827506401775e-05,
+      "loss": 0.5846,
+      "step": 5564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800958245559425e-05,
+      "loss": 0.6135,
+      "step": 5565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780008883410958e-05,
+      "loss": 0.5708,
+      "step": 5566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7799219272069032e-05,
+      "loss": 0.5839,
+      "step": 5567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798349559454565e-05,
+      "loss": 0.6128,
+      "step": 5568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779747969628297e-05,
+      "loss": 0.5852,
+      "step": 5569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779660968257104e-05,
+      "loss": 0.5526,
+      "step": 5570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7795739518335578e-05,
+      "loss": 0.6167,
+      "step": 5571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794869203593378e-05,
+      "loss": 0.606,
+      "step": 5572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793998738361243e-05,
+      "loss": 0.5764,
+      "step": 5573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779312812265598e-05,
+      "loss": 0.5933,
+      "step": 5574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792257356494397e-05,
+      "loss": 0.6178,
+      "step": 5575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7791386439893302e-05,
+      "loss": 0.6122,
+      "step": 5576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790515372869514e-05,
+      "loss": 0.6026,
+      "step": 5577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778964415543985e-05,
+      "loss": 0.5885,
+      "step": 5578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788772787621126e-05,
+      "loss": 0.6143,
+      "step": 5579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7787901269430168e-05,
+      "loss": 0.6374,
+      "step": 5580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77870296008838e-05,
+      "loss": 0.5998,
+      "step": 5581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778615778199885e-05,
+      "loss": 0.5873,
+      "step": 5582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778528581279215e-05,
+      "loss": 0.5903,
+      "step": 5583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778441369328054e-05,
+      "loss": 0.5826,
+      "step": 5584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7783541423480853e-05,
+      "loss": 0.6256,
+      "step": 5585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7782669003409924e-05,
+      "loss": 0.2656,
+      "step": 5586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7781796433084607e-05,
+      "loss": 0.6225,
+      "step": 5587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778092371252174e-05,
+      "loss": 0.5763,
+      "step": 5588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778005084173817e-05,
+      "loss": 0.609,
+      "step": 5589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779177820750763e-05,
+      "loss": 0.5808,
+      "step": 5590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7778304649576357e-05,
+      "loss": 0.6099,
+      "step": 5591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777431328231818e-05,
+      "loss": 0.552,
+      "step": 5592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776557856734006e-05,
+      "loss": 0.6006,
+      "step": 5593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7775684235099784e-05,
+      "loss": 0.6308,
+      "step": 5594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774810463346015e-05,
+      "loss": 0.5927,
+      "step": 5595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773936541489577e-05,
+      "loss": 0.548,
+      "step": 5596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777306246954733e-05,
+      "loss": 0.5795,
+      "step": 5597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7772188247536163e-05,
+      "loss": 0.6166,
+      "step": 5598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771313875472942e-05,
+      "loss": 0.622,
+      "step": 5599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7770439353374555e-05,
+      "loss": 0.5835,
+      "step": 5600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776956468125788e-05,
+      "loss": 0.5706,
+      "step": 5601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776868985913981e-05,
+      "loss": 0.6178,
+      "step": 5602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776781488703723e-05,
+      "loss": 0.596,
+      "step": 5603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7766939764967036e-05,
+      "loss": 0.5831,
+      "step": 5604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7766064492946122e-05,
+      "loss": 0.6286,
+      "step": 5605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765189070991384e-05,
+      "loss": 0.564,
+      "step": 5606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7764313499119725e-05,
+      "loss": 0.5933,
+      "step": 5607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776343777734805e-05,
+      "loss": 0.6105,
+      "step": 5608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7762561905693264e-05,
+      "loss": 0.5873,
+      "step": 5609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761685884172274e-05,
+      "loss": 0.6084,
+      "step": 5610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760809712802002e-05,
+      "loss": 0.6097,
+      "step": 5611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7759933391599357e-05,
+      "loss": 0.5539,
+      "step": 5612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7759056920581256e-05,
+      "loss": 0.6197,
+      "step": 5613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7758180299764622e-05,
+      "loss": 0.6191,
+      "step": 5614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775730352916638e-05,
+      "loss": 0.2653,
+      "step": 5615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775642660880346e-05,
+      "loss": 0.6079,
+      "step": 5616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775554953869279e-05,
+      "loss": 0.5891,
+      "step": 5617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77546723188513e-05,
+      "loss": 0.5737,
+      "step": 5618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775379494929593e-05,
+      "loss": 0.5988,
+      "step": 5619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752917430043615e-05,
+      "loss": 0.5975,
+      "step": 5620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77520397611113e-05,
+      "loss": 0.5969,
+      "step": 5621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775116194251593e-05,
+      "loss": 0.6589,
+      "step": 5622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7750283974274445e-05,
+      "loss": 0.5838,
+      "step": 5623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77494058564038e-05,
+      "loss": 0.5953,
+      "step": 5624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7748527588920956e-05,
+      "loss": 0.5844,
+      "step": 5625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774764917184286e-05,
+      "loss": 0.6184,
+      "step": 5626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746770605186473e-05,
+      "loss": 0.5796,
+      "step": 5627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7745891888968753e-05,
+      "loss": 0.5636,
+      "step": 5628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7745013023206672e-05,
+      "loss": 0.579,
+      "step": 5629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744134007917195e-05,
+      "loss": 0.5735,
+      "step": 5630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774325484311729e-05,
+      "loss": 0.592,
+      "step": 5631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7742375528823933e-05,
+      "loss": 0.622,
+      "step": 5632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77414960650541e-05,
+      "loss": 0.6023,
+      "step": 5633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7740616451824767e-05,
+      "loss": 0.6156,
+      "step": 5634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773973668915292e-05,
+      "loss": 0.599,
+      "step": 5635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7738856777055544e-05,
+      "loss": 0.6151,
+      "step": 5636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737976715549625e-05,
+      "loss": 0.6313,
+      "step": 5637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737096504652153e-05,
+      "loss": 0.624,
+      "step": 5638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736216144380125e-05,
+      "loss": 0.6013,
+      "step": 5639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773533563475053e-05,
+      "loss": 0.5868,
+      "step": 5640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7734454975780378e-05,
+      "loss": 0.627,
+      "step": 5641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733574167486665e-05,
+      "loss": 0.5932,
+      "step": 5642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732693209886397e-05,
+      "loss": 0.5685,
+      "step": 5643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7731812102996582e-05,
+      "loss": 0.2521,
+      "step": 5644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773093084683423e-05,
+      "loss": 0.5484,
+      "step": 5645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7730049441416355e-05,
+      "loss": 0.5942,
+      "step": 5646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729167886759974e-05,
+      "loss": 0.5671,
+      "step": 5647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772828618288211e-05,
+      "loss": 0.5843,
+      "step": 5648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772740432979978e-05,
+      "loss": 0.5952,
+      "step": 5649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7726522327530013e-05,
+      "loss": 0.5831,
+      "step": 5650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725640176089837e-05,
+      "loss": 0.6153,
+      "step": 5651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772475787549628e-05,
+      "loss": 0.5955,
+      "step": 5652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7723875425766376e-05,
+      "loss": 0.5879,
+      "step": 5653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722992826917166e-05,
+      "loss": 0.6031,
+      "step": 5654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722110078965685e-05,
+      "loss": 0.5859,
+      "step": 5655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721227181928982e-05,
+      "loss": 0.5889,
+      "step": 5656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7720344135824095e-05,
+      "loss": 0.578,
+      "step": 5657
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719460940668075e-05,
+      "loss": 0.565,
+      "step": 5658
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718577596477976e-05,
+      "loss": 0.6074,
+      "step": 5659
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771769410327085e-05,
+      "loss": 0.5883,
+      "step": 5660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716810461063754e-05,
+      "loss": 0.5518,
+      "step": 5661
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771592666987375e-05,
+      "loss": 0.6411,
+      "step": 5662
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715042729717895e-05,
+      "loss": 0.6004,
+      "step": 5663
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771415864061326e-05,
+      "loss": 0.5659,
+      "step": 5664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771327440257691e-05,
+      "loss": 0.6179,
+      "step": 5665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771239001562592e-05,
+      "loss": 0.6017,
+      "step": 5666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771150547977736e-05,
+      "loss": 0.6316,
+      "step": 5667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771062079504831e-05,
+      "loss": 0.5955,
+      "step": 5668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7709735961455853e-05,
+      "loss": 0.5836,
+      "step": 5669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770885097901706e-05,
+      "loss": 0.6207,
+      "step": 5670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770796584774903e-05,
+      "loss": 0.6428,
+      "step": 5671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707080567668847e-05,
+      "loss": 0.6054,
+      "step": 5672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7706195138793602e-05,
+      "loss": 0.5595,
+      "step": 5673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770530956114039e-05,
+      "loss": 0.6244,
+      "step": 5674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704423834726303e-05,
+      "loss": 0.5964,
+      "step": 5675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703537959568452e-05,
+      "loss": 0.6033,
+      "step": 5676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770265193568393e-05,
+      "loss": 0.602,
+      "step": 5677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701765763089846e-05,
+      "loss": 0.554,
+      "step": 5678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7700879441803306e-05,
+      "loss": 0.6046,
+      "step": 5679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699992971841427e-05,
+      "loss": 0.6534,
+      "step": 5680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699106353221322e-05,
+      "loss": 0.5985,
+      "step": 5681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698219585960105e-05,
+      "loss": 0.5779,
+      "step": 5682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76973326700749e-05,
+      "loss": 0.254,
+      "step": 5683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769644560558283e-05,
+      "loss": 0.601,
+      "step": 5684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769555839250102e-05,
+      "loss": 0.5939,
+      "step": 5685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7694671030846598e-05,
+      "loss": 0.6337,
+      "step": 5686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7693783520636692e-05,
+      "loss": 0.5956,
+      "step": 5687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7692895861888442e-05,
+      "loss": 0.5982,
+      "step": 5688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7692008054618986e-05,
+      "loss": 0.5683,
+      "step": 5689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769112009884546e-05,
+      "loss": 0.6599,
+      "step": 5690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690231994585013e-05,
+      "loss": 0.6179,
+      "step": 5691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768934374185478e-05,
+      "loss": 0.5894,
+      "step": 5692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7688455340671927e-05,
+      "loss": 0.5937,
+      "step": 5693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687566791053587e-05,
+      "loss": 0.5971,
+      "step": 5694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686678093016928e-05,
+      "loss": 0.6067,
+      "step": 5695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76857892465791e-05,
+      "loss": 0.5772,
+      "step": 5696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684900251757268e-05,
+      "loss": 0.632,
+      "step": 5697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684011108568593e-05,
+      "loss": 0.5784,
+      "step": 5698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683121817030238e-05,
+      "loss": 0.615,
+      "step": 5699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768223237715938e-05,
+      "loss": 0.5983,
+      "step": 5700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681342788973183e-05,
+      "loss": 0.5854,
+      "step": 5701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7680453052488825e-05,
+      "loss": 0.6161,
+      "step": 5702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679563167723486e-05,
+      "loss": 0.5961,
+      "step": 5703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678673134694343e-05,
+      "loss": 0.6137,
+      "step": 5704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677782953418577e-05,
+      "loss": 0.5391,
+      "step": 5705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767689262391338e-05,
+      "loss": 0.5617,
+      "step": 5706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7676002146195935e-05,
+      "loss": 0.5832,
+      "step": 5707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7675111520283438e-05,
+      "loss": 0.5771,
+      "step": 5708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767422074619308e-05,
+      "loss": 0.5706,
+      "step": 5709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7673329823942065e-05,
+      "loss": 0.531,
+      "step": 5710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767243875354759e-05,
+      "loss": 0.6255,
+      "step": 5711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767154753502685e-05,
+      "loss": 0.5804,
+      "step": 5712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670656168397065e-05,
+      "loss": 0.6122,
+      "step": 5713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7669764653675434e-05,
+      "loss": 0.5827,
+      "step": 5714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668872990879175e-05,
+      "loss": 0.5904,
+      "step": 5715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7667981180025498e-05,
+      "loss": 0.6257,
+      "step": 5716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7667089221131624e-05,
+      "loss": 0.6448,
+      "step": 5717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666197114214775e-05,
+      "loss": 0.6044,
+      "step": 5718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7665304859292166e-05,
+      "loss": 0.6056,
+      "step": 5719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664412456381033e-05,
+      "loss": 0.6156,
+      "step": 5720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76635199054986e-05,
+      "loss": 0.6035,
+      "step": 5721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7662627206662097e-05,
+      "loss": 0.5744,
+      "step": 5722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766173435988876e-05,
+      "loss": 0.6007,
+      "step": 5723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766084136519583e-05,
+      "loss": 0.6299,
+      "step": 5724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765994822260054e-05,
+      "loss": 0.6181,
+      "step": 5725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659054932120147e-05,
+      "loss": 0.5731,
+      "step": 5726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7658161493771883e-05,
+      "loss": 0.6115,
+      "step": 5727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657267907573002e-05,
+      "loss": 0.5627,
+      "step": 5728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656374173540755e-05,
+      "loss": 0.6025,
+      "step": 5729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76554802916924e-05,
+      "loss": 0.583,
+      "step": 5730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765458626204519e-05,
+      "loss": 0.5533,
+      "step": 5731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765369208461639e-05,
+      "loss": 0.5768,
+      "step": 5732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652797759423262e-05,
+      "loss": 0.5998,
+      "step": 5733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765190328648307e-05,
+      "loss": 0.6053,
+      "step": 5734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651008665813083e-05,
+      "loss": 0.6039,
+      "step": 5735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650113897430575e-05,
+      "loss": 0.615,
+      "step": 5736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764921898135282e-05,
+      "loss": 0.6015,
+      "step": 5737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764832391759709e-05,
+      "loss": 0.6027,
+      "step": 5738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647428706180675e-05,
+      "loss": 0.5897,
+      "step": 5739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7646533347120852e-05,
+      "loss": 0.5984,
+      "step": 5740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764563784043491e-05,
+      "loss": 0.5884,
+      "step": 5741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644742186140134e-05,
+      "loss": 0.5861,
+      "step": 5742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764384638425382e-05,
+      "loss": 0.5799,
+      "step": 5743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764295043479326e-05,
+      "loss": 0.2567,
+      "step": 5744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642054337775756e-05,
+      "loss": 0.5607,
+      "step": 5745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641158093218603e-05,
+      "loss": 0.5747,
+      "step": 5746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640261701139102e-05,
+      "loss": 0.5842,
+      "step": 5747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639365161554564e-05,
+      "loss": 0.5661,
+      "step": 5748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7638468474482297e-05,
+      "loss": 0.5682,
+      "step": 5749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763757163993961e-05,
+      "loss": 0.6009,
+      "step": 5750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636674657943825e-05,
+      "loss": 0.5782,
+      "step": 5751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7635777528512254e-05,
+      "loss": 0.5889,
+      "step": 5752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763488025166221e-05,
+      "loss": 0.5902,
+      "step": 5753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763398282741103e-05,
+      "loss": 0.5472,
+      "step": 5754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633085255776033e-05,
+      "loss": 0.572,
+      "step": 5755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632187536774545e-05,
+      "loss": 0.5918,
+      "step": 5756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631289670423903e-05,
+      "loss": 0.5867,
+      "step": 5757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630391656741438e-05,
+      "loss": 0.5796,
+      "step": 5758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7629493495744493e-05,
+      "loss": 0.599,
+      "step": 5759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76285951874504e-05,
+      "loss": 0.6072,
+      "step": 5760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627696731876504e-05,
+      "loss": 0.5927,
+      "step": 5761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626798129040153e-05,
+      "loss": 0.5917,
+      "step": 5762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76258993789587e-05,
+      "loss": 0.6073,
+      "step": 5763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625000481649488e-05,
+      "loss": 0.2523,
+      "step": 5764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7624101437129874e-05,
+      "loss": 0.5854,
+      "step": 5765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762320224541722e-05,
+      "loss": 0.2411,
+      "step": 5766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622302906528878e-05,
+      "loss": 0.6041,
+      "step": 5767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621403420482217e-05,
+      "loss": 0.5668,
+      "step": 5768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620503787294604e-05,
+      "loss": 0.6007,
+      "step": 5769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76196040069834e-05,
+      "loss": 0.632,
+      "step": 5770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761870407956598e-05,
+      "loss": 0.5604,
+      "step": 5771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7617804005059724e-05,
+      "loss": 0.5863,
+      "step": 5772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616903783482002e-05,
+      "loss": 0.5937,
+      "step": 5773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616003414850195e-05,
+      "loss": 0.598,
+      "step": 5774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761510289918169e-05,
+      "loss": 0.5779,
+      "step": 5775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761420223649387e-05,
+      "loss": 0.6075,
+      "step": 5776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761330142680412e-05,
+      "loss": 0.6312,
+      "step": 5777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612400470129836e-05,
+      "loss": 0.5754,
+      "step": 5778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7611499366488408e-05,
+      "loss": 0.5966,
+      "step": 5779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761059811589724e-05,
+      "loss": 0.5545,
+      "step": 5780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609696718373722e-05,
+      "loss": 0.5887,
+      "step": 5781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7608795173935264e-05,
+      "loss": 0.5752,
+      "step": 5782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760789348259927e-05,
+      "loss": 0.6203,
+      "step": 5783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760699164438315e-05,
+      "loss": 0.5795,
+      "step": 5784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7606089659304307e-05,
+      "loss": 0.6336,
+      "step": 5785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7605187527380162e-05,
+      "loss": 0.6167,
+      "step": 5786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7604285248628133e-05,
+      "loss": 0.2603,
+      "step": 5787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7603382823065637e-05,
+      "loss": 0.64,
+      "step": 5788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7602480250710095e-05,
+      "loss": 0.6051,
+      "step": 5789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7601577531578936e-05,
+      "loss": 0.6084,
+      "step": 5790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600674665689585e-05,
+      "loss": 0.5962,
+      "step": 5791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7599771653059473e-05,
+      "loss": 0.579,
+      "step": 5792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759886849370604e-05,
+      "loss": 0.5561,
+      "step": 5793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597965187646713e-05,
+      "loss": 0.6228,
+      "step": 5794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597061734898936e-05,
+      "loss": 0.6019,
+      "step": 5795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7596158135480156e-05,
+      "loss": 0.5477,
+      "step": 5796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7595254389407807e-05,
+      "loss": 0.6047,
+      "step": 5797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594350496699346e-05,
+      "loss": 0.5698,
+      "step": 5798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7593446457372224e-05,
+      "loss": 0.5676,
+      "step": 5799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7592542271443888e-05,
+      "loss": 0.5846,
+      "step": 5800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75916379389318e-05,
+      "loss": 0.5682,
+      "step": 5801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759073345985342e-05,
+      "loss": 0.6366,
+      "step": 5802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589828834226204e-05,
+      "loss": 0.2856,
+      "step": 5803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7588924062067623e-05,
+      "loss": 0.5899,
+      "step": 5804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758801914339514e-05,
+      "loss": 0.5945,
+      "step": 5805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758711407822623e-05,
+      "loss": 0.6072,
+      "step": 5806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7586208866578368e-05,
+      "loss": 0.621,
+      "step": 5807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7585303508469024e-05,
+      "loss": 0.5686,
+      "step": 5808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758439800391568e-05,
+      "loss": 0.5967,
+      "step": 5809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583492352935817e-05,
+      "loss": 0.5791,
+      "step": 5810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758258655554692e-05,
+      "loss": 0.6093,
+      "step": 5811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758168061176648e-05,
+      "loss": 0.613,
+      "step": 5812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758077452161198e-05,
+      "loss": 0.6224,
+      "step": 5813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579868285100922e-05,
+      "loss": 0.5792,
+      "step": 5814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75789619022508e-05,
+      "loss": 0.6072,
+      "step": 5815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7578055373079105e-05,
+      "loss": 0.5858,
+      "step": 5816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757714869760335e-05,
+      "loss": 0.6344,
+      "step": 5817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757624187584103e-05,
+      "loss": 0.6156,
+      "step": 5818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757533490780966e-05,
+      "loss": 0.6163,
+      "step": 5819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7574427793526743e-05,
+      "loss": 0.5696,
+      "step": 5820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75735205330098e-05,
+      "loss": 0.5995,
+      "step": 5821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572613126276337e-05,
+      "loss": 0.6141,
+      "step": 5822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757170557334388e-05,
+      "loss": 0.5951,
+      "step": 5823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7570797874229952e-05,
+      "loss": 0.6045,
+      "step": 5824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756989002895207e-05,
+      "loss": 0.5496,
+      "step": 5825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568982037527766e-05,
+      "loss": 0.5864,
+      "step": 5826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568073899974568e-05,
+      "loss": 0.6193,
+      "step": 5827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756716561631001e-05,
+      "loss": 0.577,
+      "step": 5828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566257186551628e-05,
+      "loss": 0.6055,
+      "step": 5829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7565348610716963e-05,
+      "loss": 0.613,
+      "step": 5830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564439888823548e-05,
+      "loss": 0.6195,
+      "step": 5831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7563531020888934e-05,
+      "loss": 0.5725,
+      "step": 5832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562622006930665e-05,
+      "loss": 0.5961,
+      "step": 5833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756171284696629e-05,
+      "loss": 0.6169,
+      "step": 5834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560803541013365e-05,
+      "loss": 0.6063,
+      "step": 5835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7559894089089443e-05,
+      "loss": 0.5473,
+      "step": 5836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755898449121208e-05,
+      "loss": 0.5813,
+      "step": 5837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7558074747398843e-05,
+      "loss": 0.5948,
+      "step": 5838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755716485766729e-05,
+      "loss": 0.5901,
+      "step": 5839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755625482203499e-05,
+      "loss": 0.6389,
+      "step": 5840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555344640519513e-05,
+      "loss": 0.5766,
+      "step": 5841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755443431313843e-05,
+      "loss": 0.5983,
+      "step": 5842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7553523839909316e-05,
+      "loss": 0.5955,
+      "step": 5843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7552613220849747e-05,
+      "loss": 0.5967,
+      "step": 5844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755170245597731e-05,
+      "loss": 0.5756,
+      "step": 5845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7550791545309583e-05,
+      "loss": 0.5833,
+      "step": 5846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549880488864156e-05,
+      "loss": 0.5908,
+      "step": 5847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548969286658612e-05,
+      "loss": 0.5837,
+      "step": 5848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548057938710547e-05,
+      "loss": 0.6003,
+      "step": 5849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7547146445037554e-05,
+      "loss": 0.6208,
+      "step": 5850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546234805657235e-05,
+      "loss": 0.6121,
+      "step": 5851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754532302058719e-05,
+      "loss": 0.6088,
+      "step": 5852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544411089845013e-05,
+      "loss": 0.6341,
+      "step": 5853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7543499013448318e-05,
+      "loss": 0.594,
+      "step": 5854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542586791414716e-05,
+      "loss": 0.6198,
+      "step": 5855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541674423761812e-05,
+      "loss": 0.617,
+      "step": 5856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7540761910507224e-05,
+      "loss": 0.5543,
+      "step": 5857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753984925166857e-05,
+      "loss": 0.5993,
+      "step": 5858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538936447263467e-05,
+      "loss": 0.5819,
+      "step": 5859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753802349730954e-05,
+      "loss": 0.6033,
+      "step": 5860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7537110401824413e-05,
+      "loss": 0.6391,
+      "step": 5861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7536197160825717e-05,
+      "loss": 0.588,
+      "step": 5862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753528377433108e-05,
+      "loss": 0.5971,
+      "step": 5863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7534370242358143e-05,
+      "loss": 0.6016,
+      "step": 5864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7533456564924535e-05,
+      "loss": 0.6018,
+      "step": 5865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75325427420479e-05,
+      "loss": 0.6174,
+      "step": 5866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7531628773745874e-05,
+      "loss": 0.586,
+      "step": 5867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530714660036112e-05,
+      "loss": 0.6197,
+      "step": 5868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529800400936255e-05,
+      "loss": 0.5738,
+      "step": 5869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752888599646396e-05,
+      "loss": 0.5767,
+      "step": 5870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527971446636874e-05,
+      "loss": 0.5716,
+      "step": 5871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752705675147266e-05,
+      "loss": 0.5585,
+      "step": 5872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526141910988973e-05,
+      "loss": 0.598,
+      "step": 5873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7525226925203474e-05,
+      "loss": 0.2525,
+      "step": 5874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7524311794133833e-05,
+      "loss": 0.5902,
+      "step": 5875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523396517797713e-05,
+      "loss": 0.6024,
+      "step": 5876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752248109621279e-05,
+      "loss": 0.6161,
+      "step": 5877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521565529396733e-05,
+      "loss": 0.609,
+      "step": 5878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7520649817367218e-05,
+      "loss": 0.5548,
+      "step": 5879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519733960141927e-05,
+      "loss": 0.6177,
+      "step": 5880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751881795773854e-05,
+      "loss": 0.6385,
+      "step": 5881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517901810174742e-05,
+      "loss": 0.6269,
+      "step": 5882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751698551746822e-05,
+      "loss": 0.6184,
+      "step": 5883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751606907963666e-05,
+      "loss": 0.6052,
+      "step": 5884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515152496697765e-05,
+      "loss": 0.5934,
+      "step": 5885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7514235768669225e-05,
+      "loss": 0.5824,
+      "step": 5886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513318895568734e-05,
+      "loss": 0.5824,
+      "step": 5887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7512401877414e-05,
+      "loss": 0.5596,
+      "step": 5888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511484714222728e-05,
+      "loss": 0.5673,
+      "step": 5889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751056740601262e-05,
+      "loss": 0.5798,
+      "step": 5890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7509649952801392e-05,
+      "loss": 0.6011,
+      "step": 5891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508732354606747e-05,
+      "loss": 0.5734,
+      "step": 5892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7507814611446407e-05,
+      "loss": 0.612,
+      "step": 5893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7506896723338092e-05,
+      "loss": 0.5733,
+      "step": 5894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750597869029952e-05,
+      "loss": 0.551,
+      "step": 5895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7505060512348414e-05,
+      "loss": 0.5643,
+      "step": 5896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504142189502506e-05,
+      "loss": 0.6331,
+      "step": 5897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503223721779517e-05,
+      "loss": 0.5813,
+      "step": 5898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7502305109197184e-05,
+      "loss": 0.5588,
+      "step": 5899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750138635177324e-05,
+      "loss": 0.5952,
+      "step": 5900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7500467449525426e-05,
+      "loss": 0.5975,
+      "step": 5901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749954840247148e-05,
+      "loss": 0.5895,
+      "step": 5902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7498629210629147e-05,
+      "loss": 0.6206,
+      "step": 5903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497709874016173e-05,
+      "loss": 0.5714,
+      "step": 5904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7496790392650306e-05,
+      "loss": 0.5991,
+      "step": 5905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74958707665493e-05,
+      "loss": 0.6135,
+      "step": 5906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494950995730903e-05,
+      "loss": 0.6212,
+      "step": 5907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494031080212882e-05,
+      "loss": 0.5909,
+      "step": 5908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7493111020012987e-05,
+      "loss": 0.5937,
+      "step": 5909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749219081514899e-05,
+      "loss": 0.6049,
+      "step": 5910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749127046563865e-05,
+      "loss": 0.5918,
+      "step": 5911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749034997149974e-05,
+      "loss": 0.5988,
+      "step": 5912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748942933275003e-05,
+      "loss": 0.6225,
+      "step": 5913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7488508549407293e-05,
+      "loss": 0.5559,
+      "step": 5914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748758762148931e-05,
+      "loss": 0.584,
+      "step": 5915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748666654901385e-05,
+      "loss": 0.6051,
+      "step": 5916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485745331998708e-05,
+      "loss": 0.5894,
+      "step": 5917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7484823970461665e-05,
+      "loss": 0.5807,
+      "step": 5918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7483902464420507e-05,
+      "loss": 0.6335,
+      "step": 5919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748298081389302e-05,
+      "loss": 0.6142,
+      "step": 5920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748205901889701e-05,
+      "loss": 0.5561,
+      "step": 5921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481137079450266e-05,
+      "loss": 0.572,
+      "step": 5922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748021499557059e-05,
+      "loss": 0.601,
+      "step": 5923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747929276727578e-05,
+      "loss": 0.5693,
+      "step": 5924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7478370394583647e-05,
+      "loss": 0.5922,
+      "step": 5925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747744787751199e-05,
+      "loss": 0.606,
+      "step": 5926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747652521607863e-05,
+      "loss": 0.5758,
+      "step": 5927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475602410301365e-05,
+      "loss": 0.6002,
+      "step": 5928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474679460198028e-05,
+      "loss": 0.5593,
+      "step": 5929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473756365786426e-05,
+      "loss": 0.6077,
+      "step": 5930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747283312708439e-05,
+      "loss": 0.5588,
+      "step": 5931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471909744109732e-05,
+      "loss": 0.6,
+      "step": 5932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470986216880287e-05,
+      "loss": 0.563,
+      "step": 5933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470062545413883e-05,
+      "loss": 0.5755,
+      "step": 5934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7469138729728356e-05,
+      "loss": 0.6333,
+      "step": 5935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468214769841542e-05,
+      "loss": 0.6059,
+      "step": 5936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7467290665771272e-05,
+      "loss": 0.5772,
+      "step": 5937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746636641753539e-05,
+      "loss": 0.5753,
+      "step": 5938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746544202515174e-05,
+      "loss": 0.6178,
+      "step": 5939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464517488638174e-05,
+      "loss": 0.593,
+      "step": 5940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7463592808012536e-05,
+      "loss": 0.5792,
+      "step": 5941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7462667983292675e-05,
+      "loss": 0.5569,
+      "step": 5942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461743014496454e-05,
+      "loss": 0.2483,
+      "step": 5943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7460817901641728e-05,
+      "loss": 0.6033,
+      "step": 5944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745989264474635e-05,
+      "loss": 0.6156,
+      "step": 5945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458967243828195e-05,
+      "loss": 0.5727,
+      "step": 5946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745804169890512e-05,
+      "loss": 0.555,
+      "step": 5947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457116009995e-05,
+      "loss": 0.6026,
+      "step": 5948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74561901771157e-05,
+      "loss": 0.5944,
+      "step": 5949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74552642002851e-05,
+      "loss": 0.5738,
+      "step": 5950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454338079521077e-05,
+      "loss": 0.6015,
+      "step": 5951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745341181484151e-05,
+      "loss": 0.6026,
+      "step": 5952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7452485406264278e-05,
+      "loss": 0.6158,
+      "step": 5953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451558853807272e-05,
+      "loss": 0.5699,
+      "step": 5954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450632157488378e-05,
+      "loss": 0.5665,
+      "step": 5955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744970531732548e-05,
+      "loss": 0.6216,
+      "step": 5956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7448778333336485e-05,
+      "loss": 0.6307,
+      "step": 5957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447851205539286e-05,
+      "loss": 0.6261,
+      "step": 5958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446923933951775e-05,
+      "loss": 0.6213,
+      "step": 5959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744599651859186e-05,
+      "loss": 0.5542,
+      "step": 5960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7445068959477444e-05,
+      "loss": 0.629,
+      "step": 5961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744414125662644e-05,
+      "loss": 0.6037,
+      "step": 5962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443213410056752e-05,
+      "loss": 0.5988,
+      "step": 5963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7442285419786293e-05,
+      "loss": 0.6303,
+      "step": 5964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441357285832984e-05,
+      "loss": 0.5878,
+      "step": 5965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440429008214738e-05,
+      "loss": 0.5878,
+      "step": 5966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439500586949483e-05,
+      "loss": 0.5811,
+      "step": 5967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438572022055138e-05,
+      "loss": 0.5788,
+      "step": 5968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7437643313549633e-05,
+      "loss": 0.6016,
+      "step": 5969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74367144614509e-05,
+      "loss": 0.5829,
+      "step": 5970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435785465776867e-05,
+      "loss": 0.6047,
+      "step": 5971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743485632654547e-05,
+      "loss": 0.5571,
+      "step": 5972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743392704377465e-05,
+      "loss": 0.5939,
+      "step": 5973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7432997617482342e-05,
+      "loss": 0.5519,
+      "step": 5974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74320680476865e-05,
+      "loss": 0.5983,
+      "step": 5975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743113833440506e-05,
+      "loss": 0.6213,
+      "step": 5976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743020847765598e-05,
+      "loss": 0.5845,
+      "step": 5977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429278477457208e-05,
+      "loss": 0.5998,
+      "step": 5978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7428348333826698e-05,
+      "loss": 0.5843,
+      "step": 5979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742741804678241e-05,
+      "loss": 0.5873,
+      "step": 5980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426487616342302e-05,
+      "loss": 0.5898,
+      "step": 5981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742555704252434e-05,
+      "loss": 0.5912,
+      "step": 5982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7424626325346487e-05,
+      "loss": 0.6144,
+      "step": 5983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742369546482671e-05,
+      "loss": 0.6466,
+      "step": 5984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422764460982987e-05,
+      "loss": 0.5747,
+      "step": 5985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7421833313833286e-05,
+      "loss": 0.5892,
+      "step": 5986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742090202339559e-05,
+      "loss": 0.6058,
+      "step": 5987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419970589687872e-05,
+      "loss": 0.5307,
+      "step": 5988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419039012728117e-05,
+      "loss": 0.5992,
+      "step": 5989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418107292534315e-05,
+      "loss": 0.6163,
+      "step": 5990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741717542912445e-05,
+      "loss": 0.6112,
+      "step": 5991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741624342251651e-05,
+      "loss": 0.5674,
+      "step": 5992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741531127272849e-05,
+      "loss": 0.5873,
+      "step": 5993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741437897977839e-05,
+      "loss": 0.5915,
+      "step": 5994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741344654368421e-05,
+      "loss": 0.642,
+      "step": 5995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741251396446394e-05,
+      "loss": 0.5751,
+      "step": 5996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411581242135602e-05,
+      "loss": 0.6287,
+      "step": 5997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410648376717195e-05,
+      "loss": 0.62,
+      "step": 5998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740971536822672e-05,
+      "loss": 0.2421,
+      "step": 5999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408782216682208e-05,
+      "loss": 0.6304,
+      "step": 6000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7407848922101663e-05,
+      "loss": 0.5429,
+      "step": 6001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406915484503105e-05,
+      "loss": 0.5837,
+      "step": 6002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740598190390456e-05,
+      "loss": 0.576,
+      "step": 6003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7405048180324046e-05,
+      "loss": 0.6044,
+      "step": 6004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740411431377959e-05,
+      "loss": 0.6091,
+      "step": 6005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403180304289226e-05,
+      "loss": 0.6183,
+      "step": 6006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740224615187098e-05,
+      "loss": 0.6172,
+      "step": 6007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401311856542897e-05,
+      "loss": 0.6284,
+      "step": 6008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7400377418323004e-05,
+      "loss": 0.5942,
+      "step": 6009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7399442837229347e-05,
+      "loss": 0.5997,
+      "step": 6010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739850811327997e-05,
+      "loss": 0.5425,
+      "step": 6011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397573246492917e-05,
+      "loss": 0.5707,
+      "step": 6012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7396638236886234e-05,
+      "loss": 0.2446,
+      "step": 6013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739570308447798e-05,
+      "loss": 0.5707,
+      "step": 6014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.73947677892862e-05,
+      "loss": 0.6278,
+      "step": 6015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739383235132896e-05,
+      "loss": 0.5636,
+      "step": 6016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7392896770624314e-05,
+      "loss": 0.6158,
+      "step": 6017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391961047190325e-05,
+      "loss": 0.5878,
+      "step": 6018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391025181045064e-05,
+      "loss": 0.5797,
+      "step": 6019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 0.6115,
+      "step": 6020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389153020692985e-05,
+      "loss": 0.6065,
+      "step": 6021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7388216726522315e-05,
+      "loss": 0.5878,
+      "step": 6022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387280289712655e-05,
+      "loss": 0.5689,
+      "step": 6023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7386343710282084e-05,
+      "loss": 0.6204,
+      "step": 6024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385406988248696e-05,
+      "loss": 0.5884,
+      "step": 6025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7384470123630558e-05,
+      "loss": 0.6374,
+      "step": 6026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738353311644577e-05,
+      "loss": 0.6205,
+      "step": 6027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7382595966712416e-05,
+      "loss": 0.5495,
+      "step": 6028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738165867444859e-05,
+      "loss": 0.5848,
+      "step": 6029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380721239672392e-05,
+      "loss": 0.6074,
+      "step": 6030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7379783662401912e-05,
+      "loss": 0.5703,
+      "step": 6031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737884594265526e-05,
+      "loss": 0.5828,
+      "step": 6032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377908080450532e-05,
+      "loss": 0.6067,
+      "step": 6033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7376970075805845e-05,
+      "loss": 0.5574,
+      "step": 6034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7376031928739296e-05,
+      "loss": 0.6033,
+      "step": 6035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375093639269006e-05,
+      "loss": 0.6039,
+      "step": 6036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7374155207413084e-05,
+      "loss": 0.5742,
+      "step": 6037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373216633189653e-05,
+      "loss": 0.6138,
+      "step": 6038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737227791661683e-05,
+      "loss": 0.6001,
+      "step": 6039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371339057712736e-05,
+      "loss": 0.6237,
+      "step": 6040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73704000564955e-05,
+      "loss": 0.5846,
+      "step": 6041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7369460912983255e-05,
+      "loss": 0.585,
+      "step": 6042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368521627194123e-05,
+      "loss": 0.596,
+      "step": 6043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7367582199146245e-05,
+      "loss": 0.5991,
+      "step": 6044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366642628857757e-05,
+      "loss": 0.5822,
+      "step": 6045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365702916346796e-05,
+      "loss": 0.5637,
+      "step": 6046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364763061631505e-05,
+      "loss": 0.5771,
+      "step": 6047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736382306473003e-05,
+      "loss": 0.5883,
+      "step": 6048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736288292566052e-05,
+      "loss": 0.5979,
+      "step": 6049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361942644441124e-05,
+      "loss": 0.611,
+      "step": 6050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361002221089996e-05,
+      "loss": 0.593,
+      "step": 6051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736006165562529e-05,
+      "loss": 0.5973,
+      "step": 6052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359120948065166e-05,
+      "loss": 0.5751,
+      "step": 6053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7358180098427788e-05,
+      "loss": 0.5996,
+      "step": 6054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735723910673132e-05,
+      "loss": 0.6179,
+      "step": 6055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356297972993923e-05,
+      "loss": 0.6006,
+      "step": 6056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7355356697233774e-05,
+      "loss": 0.6157,
+      "step": 6057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735441527946904e-05,
+      "loss": 0.6444,
+      "step": 6058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353473719717904e-05,
+      "loss": 0.582,
+      "step": 6059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7352532017998537e-05,
+      "loss": 0.5864,
+      "step": 6060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735159017432912e-05,
+      "loss": 0.6109,
+      "step": 6061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735064818872784e-05,
+      "loss": 0.6179,
+      "step": 6062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349706061212876e-05,
+      "loss": 0.5706,
+      "step": 6063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734876379180243e-05,
+      "loss": 0.6134,
+      "step": 6064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347821380514682e-05,
+      "loss": 0.5746,
+      "step": 6065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734687882736783e-05,
+      "loss": 0.6077,
+      "step": 6066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345936132380073e-05,
+      "loss": 0.5948,
+      "step": 6067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344993295569614e-05,
+      "loss": 0.5625,
+      "step": 6068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734405031695465e-05,
+      "loss": 0.6111,
+      "step": 6069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343107196553385e-05,
+      "loss": 0.5897,
+      "step": 6070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734216393438403e-05,
+      "loss": 0.5843,
+      "step": 6071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7341220530464796e-05,
+      "loss": 0.5594,
+      "step": 6072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340276984813896e-05,
+      "loss": 0.606,
+      "step": 6073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733933329744955e-05,
+      "loss": 0.6082,
+      "step": 6074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733838946838997e-05,
+      "loss": 0.5851,
+      "step": 6075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337445497653386e-05,
+      "loss": 0.5907,
+      "step": 6076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336501385258017e-05,
+      "loss": 0.5967,
+      "step": 6077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733555713122209e-05,
+      "loss": 0.58,
+      "step": 6078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733461273556384e-05,
+      "loss": 0.5552,
+      "step": 6079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333668198301494e-05,
+      "loss": 0.638,
+      "step": 6080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733272351945329e-05,
+      "loss": 0.5764,
+      "step": 6081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331778699037466e-05,
+      "loss": 0.5837,
+      "step": 6082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330833737072262e-05,
+      "loss": 0.5871,
+      "step": 6083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7329888633575927e-05,
+      "loss": 0.6228,
+      "step": 6084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73289433885667e-05,
+      "loss": 0.6214,
+      "step": 6085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327998002062836e-05,
+      "loss": 0.5921,
+      "step": 6086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327052474082583e-05,
+      "loss": 0.6065,
+      "step": 6087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326106804644193e-05,
+      "loss": 0.5911,
+      "step": 6088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325160993765934e-05,
+      "loss": 0.6059,
+      "step": 6089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324215041466054e-05,
+      "loss": 0.5766,
+      "step": 6090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323268947762827e-05,
+      "loss": 0.5925,
+      "step": 6091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732232271267451e-05,
+      "loss": 0.591,
+      "step": 6092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321376336219372e-05,
+      "loss": 0.5952,
+      "step": 6093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732042981841569e-05,
+      "loss": 0.6052,
+      "step": 6094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731948315928173e-05,
+      "loss": 0.6152,
+      "step": 6095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318536358835773e-05,
+      "loss": 0.6062,
+      "step": 6096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7317589417096098e-05,
+      "loss": 0.6064,
+      "step": 6097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731664233408099e-05,
+      "loss": 0.5654,
+      "step": 6098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731569510980873e-05,
+      "loss": 0.6055,
+      "step": 6099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.73147477442976e-05,
+      "loss": 0.5917,
+      "step": 6100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7313800237565898e-05,
+      "loss": 0.6071,
+      "step": 6101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7312852589631913e-05,
+      "loss": 0.5879,
+      "step": 6102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311904800513942e-05,
+      "loss": 0.6229,
+      "step": 6103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310956870230287e-05,
+      "loss": 0.5573,
+      "step": 6104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310008798799243e-05,
+      "loss": 0.6274,
+      "step": 6105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7309060586239117e-05,
+      "loss": 0.5831,
+      "step": 6106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7308112232568214e-05,
+      "loss": 0.6234,
+      "step": 6107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730716373780484e-05,
+      "loss": 0.5766,
+      "step": 6108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306215101967315e-05,
+      "loss": 0.5859,
+      "step": 6109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7305266325073948e-05,
+      "loss": 0.6225,
+      "step": 6110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730431740714306e-05,
+      "loss": 0.2349,
+      "step": 6111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7303368348192964e-05,
+      "loss": 0.5647,
+      "step": 6112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302419148241988e-05,
+      "loss": 0.5673,
+      "step": 6113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730146980730846e-05,
+      "loss": 0.5894,
+      "step": 6114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.73005203254107e-05,
+      "loss": 0.5663,
+      "step": 6115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729957070256705e-05,
+      "loss": 0.5993,
+      "step": 6116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298620938795837e-05,
+      "loss": 0.6001,
+      "step": 6117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297671034115397e-05,
+      "loss": 0.629,
+      "step": 6118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729672098854407e-05,
+      "loss": 0.5873,
+      "step": 6119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295770802100197e-05,
+      "loss": 0.5839,
+      "step": 6120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7294820474802124e-05,
+      "loss": 0.5797,
+      "step": 6121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72938700066682e-05,
+      "loss": 0.5817,
+      "step": 6122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7292919397716772e-05,
+      "loss": 0.555,
+      "step": 6123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291968647966194e-05,
+      "loss": 0.6144,
+      "step": 6124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729101775743482e-05,
+      "loss": 0.5546,
+      "step": 6125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729006672614101e-05,
+      "loss": 0.6012,
+      "step": 6126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7289115554103128e-05,
+      "loss": 0.5584,
+      "step": 6127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7288164241339527e-05,
+      "loss": 0.5913,
+      "step": 6128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7287212787868582e-05,
+      "loss": 0.5954,
+      "step": 6129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7286261193708663e-05,
+      "loss": 0.5989,
+      "step": 6130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7285309458878137e-05,
+      "loss": 0.6053,
+      "step": 6131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728435758339538e-05,
+      "loss": 0.625,
+      "step": 6132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283405567278773e-05,
+      "loss": 0.5746,
+      "step": 6133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728245341054669e-05,
+      "loss": 0.6039,
+      "step": 6134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281501113217517e-05,
+      "loss": 0.6178,
+      "step": 6135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280548675309638e-05,
+      "loss": 0.6286,
+      "step": 6136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727959609684144e-05,
+      "loss": 0.5564,
+      "step": 6137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278643377831316e-05,
+      "loss": 0.59,
+      "step": 6138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727769051829766e-05,
+      "loss": 0.6155,
+      "step": 6139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7276737518258865e-05,
+      "loss": 0.609,
+      "step": 6140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727578437773333e-05,
+      "loss": 0.5845,
+      "step": 6141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727483109673946e-05,
+      "loss": 0.6256,
+      "step": 6142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727387767529566e-05,
+      "loss": 0.6126,
+      "step": 6143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727292411342033e-05,
+      "loss": 0.5624,
+      "step": 6144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271970411131888e-05,
+      "loss": 0.6267,
+      "step": 6145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271016568448742e-05,
+      "loss": 0.5664,
+      "step": 6146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7270062585389307e-05,
+      "loss": 0.5687,
+      "step": 6147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7269108461972002e-05,
+      "loss": 0.5938,
+      "step": 6148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726815419821525e-05,
+      "loss": 0.5782,
+      "step": 6149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267199794137468e-05,
+      "loss": 0.6132,
+      "step": 6150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726624524975709e-05,
+      "loss": 0.5667,
+      "step": 6151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7265290565092537e-05,
+      "loss": 0.6289,
+      "step": 6152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264335740162244e-05,
+      "loss": 0.6083,
+      "step": 6153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7263380774984643e-05,
+      "loss": 0.5573,
+      "step": 6154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7262425669578177e-05,
+      "loss": 0.5937,
+      "step": 6155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261470423961276e-05,
+      "loss": 0.6206,
+      "step": 6156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7260515038152393e-05,
+      "loss": 0.5587,
+      "step": 6157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259559512169962e-05,
+      "loss": 0.6168,
+      "step": 6158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725860384603244e-05,
+      "loss": 0.5714,
+      "step": 6159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257648039758273e-05,
+      "loss": 0.5739,
+      "step": 6160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256692093365916e-05,
+      "loss": 0.6284,
+      "step": 6161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725573600687382e-05,
+      "loss": 0.566,
+      "step": 6162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254779780300448e-05,
+      "loss": 0.606,
+      "step": 6163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725382341366426e-05,
+      "loss": 0.5717,
+      "step": 6164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725286690698372e-05,
+      "loss": 0.5641,
+      "step": 6165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251910260277294e-05,
+      "loss": 0.5892,
+      "step": 6166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250953473563455e-05,
+      "loss": 0.5639,
+      "step": 6167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7249996546860668e-05,
+      "loss": 0.5586,
+      "step": 6168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7249039480187413e-05,
+      "loss": 0.5898,
+      "step": 6169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7248082273562164e-05,
+      "loss": 0.5715,
+      "step": 6170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7247124927003403e-05,
+      "loss": 0.5544,
+      "step": 6171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7246167440529612e-05,
+      "loss": 0.5865,
+      "step": 6172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724520981415928e-05,
+      "loss": 0.5735,
+      "step": 6173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7244252047910893e-05,
+      "loss": 0.5829,
+      "step": 6174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243294141802936e-05,
+      "loss": 0.5847,
+      "step": 6175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7242336095853912e-05,
+      "loss": 0.5717,
+      "step": 6176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7241377910082313e-05,
+      "loss": 0.6161,
+      "step": 6177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7240419584506636e-05,
+      "loss": 0.6028,
+      "step": 6178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723946111914539e-05,
+      "loss": 0.5911,
+      "step": 6179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723850251401707e-05,
+      "loss": 0.5937,
+      "step": 6180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7237543769140187e-05,
+      "loss": 0.5993,
+      "step": 6181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723658488453325e-05,
+      "loss": 0.5981,
+      "step": 6182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723562586021478e-05,
+      "loss": 0.5817,
+      "step": 6183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723466669620328e-05,
+      "loss": 0.5508,
+      "step": 6184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723370739251727e-05,
+      "loss": 0.5802,
+      "step": 6185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723274794917528e-05,
+      "loss": 0.6164,
+      "step": 6186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723178836619582e-05,
+      "loss": 0.5874,
+      "step": 6187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230828643597423e-05,
+      "loss": 0.6033,
+      "step": 6188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722986878139862e-05,
+      "loss": 0.5747,
+      "step": 6189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722890877961794e-05,
+      "loss": 0.5794,
+      "step": 6190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7227948638273918e-05,
+      "loss": 0.5739,
+      "step": 6191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7226988357385084e-05,
+      "loss": 0.608,
+      "step": 6192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722602793696999e-05,
+      "loss": 0.5777,
+      "step": 6193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7225067377047167e-05,
+      "loss": 0.5696,
+      "step": 6194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224106677635163e-05,
+      "loss": 0.552,
+      "step": 6195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722314583875253e-05,
+      "loss": 0.568,
+      "step": 6196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722218486041781e-05,
+      "loss": 0.5791,
+      "step": 6197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7221223742649567e-05,
+      "loss": 0.5718,
+      "step": 6198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220262485466347e-05,
+      "loss": 0.5685,
+      "step": 6199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721930108888671e-05,
+      "loss": 0.5917,
+      "step": 6200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721833955292922e-05,
+      "loss": 0.5849,
+      "step": 6201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721737787761244e-05,
+      "loss": 0.5853,
+      "step": 6202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721641606295493e-05,
+      "loss": 0.6254,
+      "step": 6203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721545410897527e-05,
+      "loss": 0.5437,
+      "step": 6204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7214492015692025e-05,
+      "loss": 0.5958,
+      "step": 6205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213529783123776e-05,
+      "loss": 0.563,
+      "step": 6206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7212567411289087e-05,
+      "loss": 0.6438,
+      "step": 6207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211604900206552e-05,
+      "loss": 0.6152,
+      "step": 6208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7210642249894742e-05,
+      "loss": 0.5699,
+      "step": 6209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720967946037225e-05,
+      "loss": 0.6025,
+      "step": 6210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7208716531657662e-05,
+      "loss": 0.5606,
+      "step": 6211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207753463769566e-05,
+      "loss": 0.5783,
+      "step": 6212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206790256726557e-05,
+      "loss": 0.5805,
+      "step": 6213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7205826910547233e-05,
+      "loss": 0.5802,
+      "step": 6214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720486342525019e-05,
+      "loss": 0.5773,
+      "step": 6215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7203899800854033e-05,
+      "loss": 0.614,
+      "step": 6216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7202936037377358e-05,
+      "loss": 0.5872,
+      "step": 6217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720197213483878e-05,
+      "loss": 0.5894,
+      "step": 6218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201008093256907e-05,
+      "loss": 0.2692,
+      "step": 6219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200043912650346e-05,
+      "loss": 0.5731,
+      "step": 6220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7199079593037713e-05,
+      "loss": 0.5711,
+      "step": 6221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7198115134437632e-05,
+      "loss": 0.5674,
+      "step": 6222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7197150536868715e-05,
+      "loss": 0.6356,
+      "step": 6223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719618580034959e-05,
+      "loss": 0.5875,
+      "step": 6224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195220924898883e-05,
+      "loss": 0.576,
+      "step": 6225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194255910535217e-05,
+      "loss": 0.5746,
+      "step": 6226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193290757277226e-05,
+      "loss": 0.5925,
+      "step": 6227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719232546514354e-05,
+      "loss": 0.613,
+      "step": 6228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7191360034152806e-05,
+      "loss": 0.5595,
+      "step": 6229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719039446432365e-05,
+      "loss": 0.5822,
+      "step": 6230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7189428755674723e-05,
+      "loss": 0.6055,
+      "step": 6231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718846290822466e-05,
+      "loss": 0.5979,
+      "step": 6232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187496921992117e-05,
+      "loss": 0.554,
+      "step": 6233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7186530796995737e-05,
+      "loss": 0.5984,
+      "step": 6234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7185564533254178e-05,
+      "loss": 0.5644,
+      "step": 6235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7184598130786088e-05,
+      "loss": 0.6025,
+      "step": 6236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718363158961013e-05,
+      "loss": 0.5834,
+      "step": 6237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182664909744965e-05,
+      "loss": 0.6205,
+      "step": 6238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181698091209253e-05,
+      "loss": 0.5627,
+      "step": 6239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718073113402166e-05,
+      "loss": 0.6209,
+      "step": 6240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7179764038200853e-05,
+      "loss": 0.5849,
+      "step": 6241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717879680376551e-05,
+      "loss": 0.5863,
+      "step": 6242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7177829430734295e-05,
+      "loss": 0.6373,
+      "step": 6243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717686191912589e-05,
+      "loss": 0.5886,
+      "step": 6244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7175894268958974e-05,
+      "loss": 0.5917,
+      "step": 6245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717492648025223e-05,
+      "loss": 0.5711,
+      "step": 6246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7173958553024335e-05,
+      "loss": 0.6495,
+      "step": 6247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7172990487293984e-05,
+      "loss": 0.5712,
+      "step": 6248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7172022283079862e-05,
+      "loss": 0.5995,
+      "step": 6249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7171053940400664e-05,
+      "loss": 0.6044,
+      "step": 6250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717008545927509e-05,
+      "loss": 0.5882,
+      "step": 6251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7169116839721828e-05,
+      "loss": 0.5587,
+      "step": 6252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716814808175958e-05,
+      "loss": 0.6066,
+      "step": 6253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716717918540706e-05,
+      "loss": 0.5668,
+      "step": 6254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7166210150682956e-05,
+      "loss": 0.5701,
+      "step": 6255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716524097760599e-05,
+      "loss": 0.5869,
+      "step": 6256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716427166619487e-05,
+      "loss": 0.6074,
+      "step": 6257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716330221646831e-05,
+      "loss": 0.5482,
+      "step": 6258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7162332628445024e-05,
+      "loss": 0.5869,
+      "step": 6259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161362902143734e-05,
+      "loss": 0.5754,
+      "step": 6260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160393037583157e-05,
+      "loss": 0.5787,
+      "step": 6261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159423034782022e-05,
+      "loss": 0.5892,
+      "step": 6262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7158452893759056e-05,
+      "loss": 0.5923,
+      "step": 6263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7157482614532988e-05,
+      "loss": 0.6174,
+      "step": 6264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715651219712255e-05,
+      "loss": 0.5767,
+      "step": 6265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715554164154648e-05,
+      "loss": 0.5844,
+      "step": 6266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154570947823506e-05,
+      "loss": 0.6142,
+      "step": 6267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153600115972385e-05,
+      "loss": 0.61,
+      "step": 6268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152629146011846e-05,
+      "loss": 0.5799,
+      "step": 6269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151658037960638e-05,
+      "loss": 0.5887,
+      "step": 6270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7150686791837516e-05,
+      "loss": 0.5682,
+      "step": 6271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7149715407661224e-05,
+      "loss": 0.6232,
+      "step": 6272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7148743885450518e-05,
+      "loss": 0.5983,
+      "step": 6273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147772225224155e-05,
+      "loss": 0.6259,
+      "step": 6274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146800427000893e-05,
+      "loss": 0.5819,
+      "step": 6275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145828490799497e-05,
+      "loss": 0.5888,
+      "step": 6276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144856416638727e-05,
+      "loss": 0.5785,
+      "step": 6277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7143884204537352e-05,
+      "loss": 0.6176,
+      "step": 6278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142911854514144e-05,
+      "loss": 0.6002,
+      "step": 6279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7141939366587873e-05,
+      "loss": 0.5932,
+      "step": 6280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7140966740777314e-05,
+      "loss": 0.6029,
+      "step": 6281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139993977101242e-05,
+      "loss": 0.5684,
+      "step": 6282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139021075578446e-05,
+      "loss": 0.5895,
+      "step": 6283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7138048036227703e-05,
+      "loss": 0.6008,
+      "step": 6284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71370748590678e-05,
+      "loss": 0.5801,
+      "step": 6285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7136101544117526e-05,
+      "loss": 0.5846,
+      "step": 6286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7135128091395666e-05,
+      "loss": 0.6047,
+      "step": 6287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7134154500921025e-05,
+      "loss": 0.579,
+      "step": 6288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713318077271239e-05,
+      "loss": 0.6008,
+      "step": 6289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7132206906788565e-05,
+      "loss": 0.5971,
+      "step": 6290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131232903168352e-05,
+      "loss": 0.5849,
+      "step": 6291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713025876187055e-05,
+      "loss": 0.6317,
+      "step": 6292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129284482913973e-05,
+      "loss": 0.574,
+      "step": 6293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7128310066317425e-05,
+      "loss": 0.5886,
+      "step": 6294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712733551209972e-05,
+      "loss": 0.5834,
+      "step": 6295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7126360820279676e-05,
+      "loss": 0.6012,
+      "step": 6296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712538599087611e-05,
+      "loss": 0.5974,
+      "step": 6297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124411023907835e-05,
+      "loss": 0.5691,
+      "step": 6298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7123435919393686e-05,
+      "loss": 0.5677,
+      "step": 6299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712246067735248e-05,
+      "loss": 0.6139,
+      "step": 6300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121485297803046e-05,
+      "loss": 0.5844,
+      "step": 6301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7120509780764218e-05,
+      "loss": 0.5748,
+      "step": 6302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711953412625483e-05,
+      "loss": 0.5988,
+      "step": 6303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118558334293713e-05,
+      "loss": 0.5791,
+      "step": 6304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117582404899714e-05,
+      "loss": 0.6463,
+      "step": 6305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7116606338091664e-05,
+      "loss": 0.6259,
+      "step": 6306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7115630133888416e-05,
+      "loss": 0.5998,
+      "step": 6307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114653792308814e-05,
+      "loss": 0.5986,
+      "step": 6308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711367731337171e-05,
+      "loss": 0.6247,
+      "step": 6309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7112700697095955e-05,
+      "loss": 0.6207,
+      "step": 6310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71117239435004e-05,
+      "loss": 0.5654,
+      "step": 6311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71107470526039e-05,
+      "loss": 0.6183,
+      "step": 6312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109770024425328e-05,
+      "loss": 0.5798,
+      "step": 6313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7108792858983535e-05,
+      "loss": 0.5972,
+      "step": 6314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7107815556297392e-05,
+      "loss": 0.5811,
+      "step": 6315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7106838116385763e-05,
+      "loss": 0.6013,
+      "step": 6316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7105860539267523e-05,
+      "loss": 0.6107,
+      "step": 6317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7104882824961542e-05,
+      "loss": 0.6274,
+      "step": 6318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71039049734867e-05,
+      "loss": 0.5697,
+      "step": 6319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7102926984861877e-05,
+      "loss": 0.6162,
+      "step": 6320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7101948859105943e-05,
+      "loss": 0.5947,
+      "step": 6321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100970596237793e-05,
+      "loss": 0.5639,
+      "step": 6322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709999219627631e-05,
+      "loss": 0.579,
+      "step": 6323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7099013659240385e-05,
+      "loss": 0.6059,
+      "step": 6324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7098034985148907e-05,
+      "loss": 0.6144,
+      "step": 6325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097056174020773e-05,
+      "loss": 0.553,
+      "step": 6326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709607722587488e-05,
+      "loss": 0.5795,
+      "step": 6327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7095098140730123e-05,
+      "loss": 0.6002,
+      "step": 6328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709411891860541e-05,
+      "loss": 0.6009,
+      "step": 6329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093139559519645e-05,
+      "loss": 0.5627,
+      "step": 6330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709216006349174e-05,
+      "loss": 0.5911,
+      "step": 6331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709118043054059e-05,
+      "loss": 0.572,
+      "step": 6332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7090200660685127e-05,
+      "loss": 0.6138,
+      "step": 6333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708922075394426e-05,
+      "loss": 0.5546,
+      "step": 6334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088240710336903e-05,
+      "loss": 0.561,
+      "step": 6335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708726052988198e-05,
+      "loss": 0.6274,
+      "step": 6336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086280212598417e-05,
+      "loss": 0.5885,
+      "step": 6337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085299758505134e-05,
+      "loss": 0.5757,
+      "step": 6338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7084319167621065e-05,
+      "loss": 0.61,
+      "step": 6339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083338439965142e-05,
+      "loss": 0.5703,
+      "step": 6340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7082357575556296e-05,
+      "loss": 0.5909,
+      "step": 6341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081376574413465e-05,
+      "loss": 0.5917,
+      "step": 6342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080395436555593e-05,
+      "loss": 0.5957,
+      "step": 6343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7079414162001617e-05,
+      "loss": 0.5835,
+      "step": 6344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707843275077048e-05,
+      "loss": 0.6117,
+      "step": 6345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707745120288113e-05,
+      "loss": 0.6116,
+      "step": 6346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076469518352525e-05,
+      "loss": 0.5549,
+      "step": 6347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075487697203607e-05,
+      "loss": 0.598,
+      "step": 6348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707450573945334e-05,
+      "loss": 0.5749,
+      "step": 6349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7073523645120677e-05,
+      "loss": 0.5917,
+      "step": 6350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7072541414224578e-05,
+      "loss": 0.6036,
+      "step": 6351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707155904678401e-05,
+      "loss": 0.6134,
+      "step": 6352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7070576542817937e-05,
+      "loss": 0.62,
+      "step": 6353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7069593902345328e-05,
+      "loss": 0.5843,
+      "step": 6354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706861112538515e-05,
+      "loss": 0.5928,
+      "step": 6355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067628211956384e-05,
+      "loss": 0.5341,
+      "step": 6356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066645162078003e-05,
+      "loss": 0.5833,
+      "step": 6357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706566197576898e-05,
+      "loss": 0.5791,
+      "step": 6358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706467865304831e-05,
+      "loss": 0.5906,
+      "step": 6359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063695193934963e-05,
+      "loss": 0.6073,
+      "step": 6360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7062711598447936e-05,
+      "loss": 0.5835,
+      "step": 6361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061727866606215e-05,
+      "loss": 0.5903,
+      "step": 6362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060743998428796e-05,
+      "loss": 0.5778,
+      "step": 6363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7059759993934664e-05,
+      "loss": 0.618,
+      "step": 6364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058775853142825e-05,
+      "loss": 0.5562,
+      "step": 6365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057791576072278e-05,
+      "loss": 0.5699,
+      "step": 6366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056807162742024e-05,
+      "loss": 0.5665,
+      "step": 6367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055822613171068e-05,
+      "loss": 0.6079,
+      "step": 6368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705483792737842e-05,
+      "loss": 0.5711,
+      "step": 6369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705385310538309e-05,
+      "loss": 0.6211,
+      "step": 6370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705286814720409e-05,
+      "loss": 0.5918,
+      "step": 6371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051883052860435e-05,
+      "loss": 0.5941,
+      "step": 6372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050897822371147e-05,
+      "loss": 0.6225,
+      "step": 6373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7049912455755243e-05,
+      "loss": 0.6264,
+      "step": 6374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704892695303175e-05,
+      "loss": 0.5909,
+      "step": 6375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047941314219695e-05,
+      "loss": 0.5662,
+      "step": 6376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046955539338107e-05,
+      "loss": 0.614,
+      "step": 6377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045969628406013e-05,
+      "loss": 0.5809,
+      "step": 6378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704498358144245e-05,
+      "loss": 0.595,
+      "step": 6379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043997398466453e-05,
+      "loss": 0.5835,
+      "step": 6380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704301107949707e-05,
+      "loss": 0.6306,
+      "step": 6381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042024624553333e-05,
+      "loss": 0.5643,
+      "step": 6382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041038033654295e-05,
+      "loss": 0.5971,
+      "step": 6383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7040051306818992e-05,
+      "loss": 0.574,
+      "step": 6384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039064444066488e-05,
+      "loss": 0.5961,
+      "step": 6385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038077445415825e-05,
+      "loss": 0.6193,
+      "step": 6386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7037090310886065e-05,
+      "loss": 0.6055,
+      "step": 6387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036103040496258e-05,
+      "loss": 0.5599,
+      "step": 6388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7035115634265474e-05,
+      "loss": 0.576,
+      "step": 6389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703412809221277e-05,
+      "loss": 0.5686,
+      "step": 6390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7033140414357216e-05,
+      "loss": 0.5988,
+      "step": 6391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7032152600717875e-05,
+      "loss": 0.5913,
+      "step": 6392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703116465131382e-05,
+      "loss": 0.5958,
+      "step": 6393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7030176566164126e-05,
+      "loss": 0.6527,
+      "step": 6394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029188345287868e-05,
+      "loss": 0.6248,
+      "step": 6395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028199988704125e-05,
+      "loss": 0.6104,
+      "step": 6396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7027211496431982e-05,
+      "loss": 0.6119,
+      "step": 6397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702622286849052e-05,
+      "loss": 0.5983,
+      "step": 6398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7025234104898824e-05,
+      "loss": 0.5834,
+      "step": 6399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024245205675986e-05,
+      "loss": 0.608,
+      "step": 6400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023256170841097e-05,
+      "loss": 0.5624,
+      "step": 6401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702226700041325e-05,
+      "loss": 0.5857,
+      "step": 6402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702127769441155e-05,
+      "loss": 0.5945,
+      "step": 6403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7020288252855084e-05,
+      "loss": 0.5785,
+      "step": 6404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019298675762966e-05,
+      "loss": 0.6064,
+      "step": 6405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7018308963154294e-05,
+      "loss": 0.5595,
+      "step": 6406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017319115048178e-05,
+      "loss": 0.5878,
+      "step": 6407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016329131463726e-05,
+      "loss": 0.595,
+      "step": 6408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7015339012420057e-05,
+      "loss": 0.5889,
+      "step": 6409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701434875793628e-05,
+      "loss": 0.5939,
+      "step": 6410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7013358368031517e-05,
+      "loss": 0.5634,
+      "step": 6411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7012367842724887e-05,
+      "loss": 0.5921,
+      "step": 6412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011377182035517e-05,
+      "loss": 0.6014,
+      "step": 6413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7010386385982527e-05,
+      "loss": 0.5948,
+      "step": 6414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7009395454585046e-05,
+      "loss": 0.6286,
+      "step": 6415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7008404387862213e-05,
+      "loss": 0.5583,
+      "step": 6416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700741318583315e-05,
+      "loss": 0.5874,
+      "step": 6417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006421848517004e-05,
+      "loss": 0.6061,
+      "step": 6418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700543037593291e-05,
+      "loss": 0.5806,
+      "step": 6419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700443876810001e-05,
+      "loss": 0.5715,
+      "step": 6420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7003447025037446e-05,
+      "loss": 0.6016,
+      "step": 6421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700245514676437e-05,
+      "loss": 0.5489,
+      "step": 6422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001463133299925e-05,
+      "loss": 0.589,
+      "step": 6423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000470984663267e-05,
+      "loss": 0.5572,
+      "step": 6424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6999478700873552e-05,
+      "loss": 0.6022,
+      "step": 6425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699848628194993e-05,
+      "loss": 0.6079,
+      "step": 6426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699749372791157e-05,
+      "loss": 0.5854,
+      "step": 6427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699650103877763e-05,
+      "loss": 0.5894,
+      "step": 6428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995508214567275e-05,
+      "loss": 0.5831,
+      "step": 6429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994515255299676e-05,
+      "loss": 0.6006,
+      "step": 6430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6993522160994e-05,
+      "loss": 0.5757,
+      "step": 6431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699252893166942e-05,
+      "loss": 0.597,
+      "step": 6432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991535567345115e-05,
+      "loss": 0.5755,
+      "step": 6433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699054206804026e-05,
+      "loss": 0.5693,
+      "step": 6434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698954843377403e-05,
+      "loss": 0.6088,
+      "step": 6435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698855466456562e-05,
+      "loss": 0.6205,
+      "step": 6436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6987560760434213e-05,
+      "loss": 0.5831,
+      "step": 6437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986566721398992e-05,
+      "loss": 0.5661,
+      "step": 6438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698557254747915e-05,
+      "loss": 0.6084,
+      "step": 6439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6984578238693888e-05,
+      "loss": 0.5564,
+      "step": 6440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983583795062392e-05,
+      "loss": 0.5524,
+      "step": 6441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6982589216603867e-05,
+      "loss": 0.5861,
+      "step": 6442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698159450333752e-05,
+      "loss": 0.5852,
+      "step": 6443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698059965528254e-05,
+      "loss": 0.5677,
+      "step": 6444
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979604672458142e-05,
+      "loss": 0.6076,
+      "step": 6445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6978609554883544e-05,
+      "loss": 0.5798,
+      "step": 6446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977614302577943e-05,
+      "loss": 0.5861,
+      "step": 6447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976618915560564e-05,
+      "loss": 0.6346,
+      "step": 6448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6975623393850615e-05,
+      "loss": 0.5781,
+      "step": 6449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974627737467327e-05,
+      "loss": 0.5707,
+      "step": 6450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6973631946429917e-05,
+      "loss": 0.6122,
+      "step": 6451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972636020757605e-05,
+      "loss": 0.6175,
+      "step": 6452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697163996046963e-05,
+      "loss": 0.6137,
+      "step": 6453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697064376558521e-05,
+      "loss": 0.6244,
+      "step": 6454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696964743612359e-05,
+      "loss": 0.6058,
+      "step": 6455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696865097210399e-05,
+      "loss": 0.6202,
+      "step": 6456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6967654373545664e-05,
+      "loss": 0.6326,
+      "step": 6457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966657640467846e-05,
+      "loss": 0.5775,
+      "step": 6458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965660772889778e-05,
+      "loss": 0.5658,
+      "step": 6459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964663770830705e-05,
+      "loss": 0.5822,
+      "step": 6460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6963666634309877e-05,
+      "loss": 0.5825,
+      "step": 6461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962669363346548e-05,
+      "loss": 0.6182,
+      "step": 6462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961671957959967e-05,
+      "loss": 0.5829,
+      "step": 6463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696067441816939e-05,
+      "loss": 0.5865,
+      "step": 6464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695967674399408e-05,
+      "loss": 0.5646,
+      "step": 6465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958678935453298e-05,
+      "loss": 0.5868,
+      "step": 6466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69576809925663e-05,
+      "loss": 0.5716,
+      "step": 6467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956682915352362e-05,
+      "loss": 0.5935,
+      "step": 6468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695568470383075e-05,
+      "loss": 0.5889,
+      "step": 6469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695468635802074e-05,
+      "loss": 0.6018,
+      "step": 6470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695368787794159e-05,
+      "loss": 0.5628,
+      "step": 6471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69526892636126e-05,
+      "loss": 0.5861,
+      "step": 6472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951690515053032e-05,
+      "loss": 0.5766,
+      "step": 6473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695069163228218e-05,
+      "loss": 0.5994,
+      "step": 6474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694969261531932e-05,
+      "loss": 0.5863,
+      "step": 6475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694869346418374e-05,
+      "loss": 0.5745,
+      "step": 6476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947694178894735e-05,
+      "loss": 0.5557,
+      "step": 6477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946694759471592e-05,
+      "loss": 0.6285,
+      "step": 6478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694569520593361e-05,
+      "loss": 0.5815,
+      "step": 6479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944695518300087e-05,
+      "loss": 0.5708,
+      "step": 6480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943695696590322e-05,
+      "loss": 0.5904,
+      "step": 6481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942695740823617e-05,
+      "loss": 0.5998,
+      "step": 6482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6941695651019278e-05,
+      "loss": 0.5673,
+      "step": 6483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6940695427196613e-05,
+      "loss": 0.5661,
+      "step": 6484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939695069374932e-05,
+      "loss": 0.5888,
+      "step": 6485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693869457757355e-05,
+      "loss": 0.6046,
+      "step": 6486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6937693951811783e-05,
+      "loss": 0.5604,
+      "step": 6487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6936693192108947e-05,
+      "loss": 0.6017,
+      "step": 6488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693569229848436e-05,
+      "loss": 0.5746,
+      "step": 6489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934691270957354e-05,
+      "loss": 0.5885,
+      "step": 6490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6933690109547248e-05,
+      "loss": 0.5703,
+      "step": 6491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932688814273377e-05,
+      "loss": 0.5761,
+      "step": 6492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931687385155067e-05,
+      "loss": 0.588,
+      "step": 6493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693068582221165e-05,
+      "loss": 0.6273,
+      "step": 6494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692968412546247e-05,
+      "loss": 0.5709,
+      "step": 6495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692868229492686e-05,
+      "loss": 0.5808,
+      "step": 6496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927680330624165e-05,
+      "loss": 0.5933,
+      "step": 6497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926678232573724e-05,
+      "loss": 0.591,
+      "step": 6498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6925676000794894e-05,
+      "loss": 0.5653,
+      "step": 6499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6924673635307013e-05,
+      "loss": 0.5813,
+      "step": 6500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692367113612944e-05,
+      "loss": 0.5924,
+      "step": 6501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6922668503281525e-05,
+      "loss": 0.6243,
+      "step": 6502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921665736782633e-05,
+      "loss": 0.5908,
+      "step": 6503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692066283665211e-05,
+      "loss": 0.5818,
+      "step": 6504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6919659802909332e-05,
+      "loss": 0.5716,
+      "step": 6505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6918656635573656e-05,
+      "loss": 0.5729,
+      "step": 6506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6917653334664456e-05,
+      "loss": 0.5841,
+      "step": 6507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691664990020109e-05,
+      "loss": 0.5847,
+      "step": 6508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6915646332202948e-05,
+      "loss": 0.5987,
+      "step": 6509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691464263068939e-05,
+      "loss": 0.6245,
+      "step": 6510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69136387956798e-05,
+      "loss": 0.5584,
+      "step": 6511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6912634827193556e-05,
+      "loss": 0.6002,
+      "step": 6512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6911630725250045e-05,
+      "loss": 0.5695,
+      "step": 6513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691062648986865e-05,
+      "loss": 0.6037,
+      "step": 6514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909622121068755e-05,
+      "loss": 0.608,
+      "step": 6515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6908617618869762e-05,
+      "loss": 0.6177,
+      "step": 6516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6907612983291048e-05,
+      "loss": 0.5545,
+      "step": 6517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6906608214352025e-05,
+      "loss": 0.5914,
+      "step": 6518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6905603312072082e-05,
+      "loss": 0.5831,
+      "step": 6519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904598276470617e-05,
+      "loss": 0.5993,
+      "step": 6520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903593107567044e-05,
+      "loss": 0.6014,
+      "step": 6521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690258780538076e-05,
+      "loss": 0.5677,
+      "step": 6522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690158236993118e-05,
+      "loss": 0.6155,
+      "step": 6523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690057680123771e-05,
+      "loss": 0.5627,
+      "step": 6524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689957109931977e-05,
+      "loss": 0.5765,
+      "step": 6525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689856526419677e-05,
+      "loss": 0.6059,
+      "step": 6526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689755929588813e-05,
+      "loss": 0.6079,
+      "step": 6527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6896553194413274e-05,
+      "loss": 0.5839,
+      "step": 6528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6895546959791627e-05,
+      "loss": 0.6046,
+      "step": 6529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689454059204261e-05,
+      "loss": 0.5902,
+      "step": 6530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893534091185658e-05,
+      "loss": 0.5643,
+      "step": 6531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68925274572402e-05,
+      "loss": 0.5548,
+      "step": 6532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6891520690225672e-05,
+      "loss": 0.5636,
+      "step": 6533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689051379016151e-05,
+      "loss": 0.5889,
+      "step": 6534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6889506757067157e-05,
+      "loss": 0.5745,
+      "step": 6535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6888499590962046e-05,
+      "loss": 0.2714,
+      "step": 6536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887492291865628e-05,
+      "loss": 0.6346,
+      "step": 6537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688648485979735e-05,
+      "loss": 0.576,
+      "step": 6538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688547729477666e-05,
+      "loss": 0.589,
+      "step": 6539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884469596823015e-05,
+      "loss": 0.6038,
+      "step": 6540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6883461765955863e-05,
+      "loss": 0.5479,
+      "step": 6541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6882453802194666e-05,
+      "loss": 0.5858,
+      "step": 6542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6881445705558882e-05,
+      "loss": 0.6026,
+      "step": 6543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880437476067973e-05,
+      "loss": 0.5833,
+      "step": 6544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687942911374141e-05,
+      "loss": 0.5826,
+      "step": 6545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6878420618598653e-05,
+      "loss": 0.6047,
+      "step": 6546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6877411990659176e-05,
+      "loss": 0.5958,
+      "step": 6547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876403229942453e-05,
+      "loss": 0.6242,
+      "step": 6548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875394336467956e-05,
+      "loss": 0.6059,
+      "step": 6549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687438531025517e-05,
+      "loss": 0.5684,
+      "step": 6550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6873376151323562e-05,
+      "loss": 0.6319,
+      "step": 6551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687236685969263e-05,
+      "loss": 0.5375,
+      "step": 6552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687135743538185e-05,
+      "loss": 0.5968,
+      "step": 6553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6870347878410712e-05,
+      "loss": 0.5811,
+      "step": 6554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6869338188798713e-05,
+      "loss": 0.5724,
+      "step": 6555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6868328366565338e-05,
+      "loss": 0.5745,
+      "step": 6556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867318411730088e-05,
+      "loss": 0.5742,
+      "step": 6557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686630832431246e-05,
+      "loss": 0.5917,
+      "step": 6558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865298104331956e-05,
+      "loss": 0.6027,
+      "step": 6559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6864287751808075e-05,
+      "loss": 0.581,
+      "step": 6560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6863277266760327e-05,
+      "loss": 0.5634,
+      "step": 6561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6862266649208226e-05,
+      "loss": 0.5981,
+      "step": 6562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861255899171272e-05,
+      "loss": 0.5637,
+      "step": 6563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6860245016668986e-05,
+      "loss": 0.5949,
+      "step": 6564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6859234001720882e-05,
+      "loss": 0.5926,
+      "step": 6565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6858222854346485e-05,
+      "loss": 0.576,
+      "step": 6566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6857211574565303e-05,
+      "loss": 0.561,
+      "step": 6567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6856200162396876e-05,
+      "loss": 0.5827,
+      "step": 6568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855188617860717e-05,
+      "loss": 0.6114,
+      "step": 6569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6854176940976365e-05,
+      "loss": 0.5918,
+      "step": 6570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685316513176335e-05,
+      "loss": 0.5657,
+      "step": 6571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6852153190241197e-05,
+      "loss": 0.5367,
+      "step": 6572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851141116429453e-05,
+      "loss": 0.5626,
+      "step": 6573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6850128910347657e-05,
+      "loss": 0.6066,
+      "step": 6574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6849116572015347e-05,
+      "loss": 0.5573,
+      "step": 6575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6848104101452067e-05,
+      "loss": 0.2494,
+      "step": 6576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684709149867737e-05,
+      "loss": 0.5822,
+      "step": 6577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6846078763710795e-05,
+      "loss": 0.5831,
+      "step": 6578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845065896571906e-05,
+      "loss": 0.6178,
+      "step": 6579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684405289728025e-05,
+      "loss": 0.61,
+      "step": 6580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843039765855388e-05,
+      "loss": 0.5669,
+      "step": 6581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6842026502316874e-05,
+      "loss": 0.5836,
+      "step": 6582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684101310668428e-05,
+      "loss": 0.5801,
+      "step": 6583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6839999578977162e-05,
+      "loss": 0.593,
+      "step": 6584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6838985919215094e-05,
+      "loss": 0.5661,
+      "step": 6585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683797212741764e-05,
+      "loss": 0.2596,
+      "step": 6586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6836958203604376e-05,
+      "loss": 0.5777,
+      "step": 6587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6835944147794875e-05,
+      "loss": 0.615,
+      "step": 6588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683492996000872e-05,
+      "loss": 0.5652,
+      "step": 6589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6833915640265485e-05,
+      "loss": 0.6118,
+      "step": 6590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683290118858476e-05,
+      "loss": 0.5536,
+      "step": 6591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831886604986122e-05,
+      "loss": 0.6014,
+      "step": 6592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683087188948916e-05,
+      "loss": 0.5821,
+      "step": 6593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682985704211347e-05,
+      "loss": 0.5793,
+      "step": 6594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682884206287864e-05,
+      "loss": 0.5514,
+      "step": 6595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827826951804268e-05,
+      "loss": 0.5602,
+      "step": 6596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6826811708909954e-05,
+      "loss": 0.5914,
+      "step": 6597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682579633421529e-05,
+      "loss": 0.5325,
+      "step": 6598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682478082773989e-05,
+      "loss": 0.6078,
+      "step": 6599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823765189503348e-05,
+      "loss": 0.5715,
+      "step": 6600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822749419525286e-05,
+      "loss": 0.5569,
+      "step": 6601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821733517825304e-05,
+      "loss": 0.565,
+      "step": 6602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6820717484423018e-05,
+      "loss": 0.6142,
+      "step": 6603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6819701319338045e-05,
+      "loss": 0.5734,
+      "step": 6604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818685022590004e-05,
+      "loss": 0.5647,
+      "step": 6605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6817668594198514e-05,
+      "loss": 0.5755,
+      "step": 6606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6816652034183197e-05,
+      "loss": 0.5769,
+      "step": 6607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6815635342563686e-05,
+      "loss": 0.6005,
+      "step": 6608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814618519359603e-05,
+      "loss": 0.6147,
+      "step": 6609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681360156459058e-05,
+      "loss": 0.6175,
+      "step": 6610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681258447827625e-05,
+      "loss": 0.6049,
+      "step": 6611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811567260436255e-05,
+      "loss": 0.5829,
+      "step": 6612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810549911090226e-05,
+      "loss": 0.6022,
+      "step": 6613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6809532430257805e-05,
+      "loss": 0.5584,
+      "step": 6614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808514817958645e-05,
+      "loss": 0.5552,
+      "step": 6615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680749707421238e-05,
+      "loss": 0.5681,
+      "step": 6616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6806479199038667e-05,
+      "loss": 0.5831,
+      "step": 6617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6805461192457158e-05,
+      "loss": 0.5867,
+      "step": 6618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68044430544875e-05,
+      "loss": 0.5941,
+      "step": 6619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6803424785149352e-05,
+      "loss": 0.5789,
+      "step": 6620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6802406384462376e-05,
+      "loss": 0.5706,
+      "step": 6621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801387852446232e-05,
+      "loss": 0.6216,
+      "step": 6622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6800369189120585e-05,
+      "loss": 0.6001,
+      "step": 6623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799350394505097e-05,
+      "loss": 0.5745,
+      "step": 6624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798331468619443e-05,
+      "loss": 0.5906,
+      "step": 6625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6797312411483292e-05,
+      "loss": 0.6153,
+      "step": 6626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796293223116316e-05,
+      "loss": 0.5669,
+      "step": 6627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6795273903538198e-05,
+      "loss": 0.5504,
+      "step": 6628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679425445276861e-05,
+      "loss": 0.5422,
+      "step": 6629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679323487082724e-05,
+      "loss": 0.571,
+      "step": 6630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679221515773377e-05,
+      "loss": 0.5724,
+      "step": 6631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6791195313507886e-05,
+      "loss": 0.5751,
+      "step": 6632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6790175338169277e-05,
+      "loss": 0.599,
+      "step": 6633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6789155231737632e-05,
+      "loss": 0.5734,
+      "step": 6634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788134994232653e-05,
+      "loss": 0.6145,
+      "step": 6635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787114625674035e-05,
+      "loss": 0.5843,
+      "step": 6636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6786094126081473e-05,
+      "loss": 0.545,
+      "step": 6637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6785073495474673e-05,
+      "loss": 0.6064,
+      "step": 6638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678405273387334e-05,
+      "loss": 0.592,
+      "step": 6639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6783031841297175e-05,
+      "loss": 0.5705,
+      "step": 6640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782010817765896e-05,
+      "loss": 0.6108,
+      "step": 6641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678098966329921e-05,
+      "loss": 0.5498,
+      "step": 6642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6779968377916832e-05,
+      "loss": 0.5681,
+      "step": 6643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6778946961638482e-05,
+      "loss": 0.6098,
+      "step": 6644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6777925414483876e-05,
+      "loss": 0.6403,
+      "step": 6645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776903736472742e-05,
+      "loss": 0.6021,
+      "step": 6646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6775881927624796e-05,
+      "loss": 0.5739,
+      "step": 6647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774859987959774e-05,
+      "loss": 0.5682,
+      "step": 6648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67738379174974e-05,
+      "loss": 0.5736,
+      "step": 6649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.5712,
+      "step": 6650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6771793384259537e-05,
+      "loss": 0.5738,
+      "step": 6651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6770770921523522e-05,
+      "loss": 0.272,
+      "step": 6652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67697483280691e-05,
+      "loss": 0.5531,
+      "step": 6653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768725603916015e-05,
+      "loss": 0.5907,
+      "step": 6654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676770274908401e-05,
+      "loss": 0.5515,
+      "step": 6655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676667976359284e-05,
+      "loss": 0.6052,
+      "step": 6656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765656647462245e-05,
+      "loss": 0.613,
+      "step": 6657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764633400711984e-05,
+      "loss": 0.5678,
+      "step": 6658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6763610023361813e-05,
+      "loss": 0.5661,
+      "step": 6659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762586515431484e-05,
+      "loss": 0.6138,
+      "step": 6660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676156287694076e-05,
+      "loss": 0.6028,
+      "step": 6661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6760539107909405e-05,
+      "loss": 0.5595,
+      "step": 6662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6759515208357188e-05,
+      "loss": 0.582,
+      "step": 6663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6758491178303865e-05,
+      "loss": 0.588,
+      "step": 6664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757467017769216e-05,
+      "loss": 0.5503,
+      "step": 6665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675644272677301e-05,
+      "loss": 0.585,
+      "step": 6666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6755418305335026e-05,
+      "loss": 0.607,
+      "step": 6667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754393753475032e-05,
+      "loss": 0.6155,
+      "step": 6668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675336907121282e-05,
+      "loss": 0.5976,
+      "step": 6669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752344258568167e-05,
+      "loss": 0.594,
+      "step": 6670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6751319315560863e-05,
+      "loss": 0.2429,
+      "step": 6671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750294242210685e-05,
+      "loss": 0.6003,
+      "step": 6672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749269038537435e-05,
+      "loss": 0.615,
+      "step": 6673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6748243704560902e-05,
+      "loss": 0.5991,
+      "step": 6674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674721824030088e-05,
+      "loss": 0.5945,
+      "step": 6675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6746192645777164e-05,
+      "loss": 0.5962,
+      "step": 6676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6745166921009563e-05,
+      "loss": 0.5799,
+      "step": 6677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6744141066017873e-05,
+      "loss": 0.5657,
+      "step": 6678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67431150808219e-05,
+      "loss": 0.5916,
+      "step": 6679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6742088965441456e-05,
+      "loss": 0.5808,
+      "step": 6680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6741062719896345e-05,
+      "loss": 0.5713,
+      "step": 6681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740036344206384e-05,
+      "loss": 0.6228,
+      "step": 6682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673900983839139e-05,
+      "loss": 0.5872,
+      "step": 6683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673798320247118e-05,
+      "loss": 0.5886,
+      "step": 6684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6736956436465573e-05,
+      "loss": 0.5675,
+      "step": 6685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6735929540394392e-05,
+      "loss": 0.5814,
+      "step": 6686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734902514277467e-05,
+      "loss": 0.5649,
+      "step": 6687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673387535813462e-05,
+      "loss": 0.5459,
+      "step": 6688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6732848071985683e-05,
+      "loss": 0.5323,
+      "step": 6689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731820655850492e-05,
+      "loss": 0.5728,
+      "step": 6690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6730793109748877e-05,
+      "loss": 0.6361,
+      "step": 6691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729765433700684e-05,
+      "loss": 0.6076,
+      "step": 6692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6728737627725745e-05,
+      "loss": 0.5942,
+      "step": 6693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6727709691843915e-05,
+      "loss": 0.5966,
+      "step": 6694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6726681626075026e-05,
+      "loss": 0.575,
+      "step": 6695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6725653430438935e-05,
+      "loss": 0.5673,
+      "step": 6696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724625104955484e-05,
+      "loss": 0.6167,
+      "step": 6697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672359664964454e-05,
+      "loss": 0.5953,
+      "step": 6698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6722568064525947e-05,
+      "loss": 0.5714,
+      "step": 6699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6721539349619567e-05,
+      "loss": 0.5713,
+      "step": 6700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672051050494526e-05,
+      "loss": 0.5977,
+      "step": 6701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671948153052289e-05,
+      "loss": 0.5934,
+      "step": 6702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671845242637232e-05,
+      "loss": 0.5851,
+      "step": 6703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6717423192513424e-05,
+      "loss": 0.5916,
+      "step": 6704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671639382896607e-05,
+      "loss": 0.5831,
+      "step": 6705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715364335750126e-05,
+      "loss": 0.5763,
+      "step": 6706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6714334712885476e-05,
+      "loss": 0.5658,
+      "step": 6707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6713304960391992e-05,
+      "loss": 0.566,
+      "step": 6708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671227507828956e-05,
+      "loss": 0.5612,
+      "step": 6709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711245066598054e-05,
+      "loss": 0.5773,
+      "step": 6710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6710214925337372e-05,
+      "loss": 0.5823,
+      "step": 6711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709184654527392e-05,
+      "loss": 0.5777,
+      "step": 6712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670815425418801e-05,
+      "loss": 0.5861,
+      "step": 6713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6707123724339114e-05,
+      "loss": 0.5665,
+      "step": 6714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670609306500061e-05,
+      "loss": 0.6031,
+      "step": 6715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670506227619239e-05,
+      "loss": 0.5614,
+      "step": 6716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6704031357934347e-05,
+      "loss": 0.584,
+      "step": 6717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67030003102464e-05,
+      "loss": 0.6077,
+      "step": 6718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6701969133148442e-05,
+      "loss": 0.5954,
+      "step": 6719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6700937826660384e-05,
+      "loss": 0.6041,
+      "step": 6720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6699906390802142e-05,
+      "loss": 0.5559,
+      "step": 6721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669887482559362e-05,
+      "loss": 0.5748,
+      "step": 6722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6697843131054746e-05,
+      "loss": 0.5582,
+      "step": 6723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669681130720543e-05,
+      "loss": 0.5582,
+      "step": 6724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669577935406559e-05,
+      "loss": 0.5721,
+      "step": 6725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6694747271655155e-05,
+      "loss": 0.5919,
+      "step": 6726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669371505999405e-05,
+      "loss": 0.5872,
+      "step": 6727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66926827191022e-05,
+      "loss": 0.5889,
+      "step": 6728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669165024899954e-05,
+      "loss": 0.6059,
+      "step": 6729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690617649706e-05,
+      "loss": 0.5259,
+      "step": 6730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6689584921241514e-05,
+      "loss": 0.5736,
+      "step": 6731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6688552063626025e-05,
+      "loss": 0.5907,
+      "step": 6732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687519076879473e-05,
+      "loss": 0.268,
+      "step": 6733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6686485961021795e-05,
+      "loss": 0.598,
+      "step": 6734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6685452716072946e-05,
+      "loss": 0.6238,
+      "step": 6735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6684419342052868e-05,
+      "loss": 0.5975,
+      "step": 6736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668338583898151e-05,
+      "loss": 0.5626,
+      "step": 6737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668235220687883e-05,
+      "loss": 0.5762,
+      "step": 6738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668131844576478e-05,
+      "loss": 0.5927,
+      "step": 6739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668028455565932e-05,
+      "loss": 0.5861,
+      "step": 6740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667925053658241e-05,
+      "loss": 0.5514,
+      "step": 6741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6678216388554012e-05,
+      "loss": 0.5769,
+      "step": 6742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6677182111594095e-05,
+      "loss": 0.6104,
+      "step": 6743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6676147705722627e-05,
+      "loss": 0.5659,
+      "step": 6744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675113170959568e-05,
+      "loss": 0.5726,
+      "step": 6745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674078507324905e-05,
+      "loss": 0.607,
+      "step": 6746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667304371483861e-05,
+      "loss": 0.5822,
+      "step": 6747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6672008793520656e-05,
+      "loss": 0.5594,
+      "step": 6748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670973743391025e-05,
+      "loss": 0.6087,
+      "step": 6749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669938564469707e-05,
+      "loss": 0.5873,
+      "step": 6750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6668903256776673e-05,
+      "loss": 0.5846,
+      "step": 6751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6667867820331927e-05,
+      "loss": 0.5913,
+      "step": 6752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6666832255155447e-05,
+      "loss": 0.5973,
+      "step": 6753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6665796561267234e-05,
+      "loss": 0.5909,
+      "step": 6754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666476073868728e-05,
+      "loss": 0.5889,
+      "step": 6755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6663724787435584e-05,
+      "loss": 0.5627,
+      "step": 6756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662688707532143e-05,
+      "loss": 0.5482,
+      "step": 6757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666165249899696e-05,
+      "loss": 0.6069,
+      "step": 6758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666061616185005e-05,
+      "loss": 0.5626,
+      "step": 6759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665957969611141e-05,
+      "loss": 0.6135,
+      "step": 6760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6658543101801055e-05,
+      "loss": 0.5599,
+      "step": 6761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6657506378938994e-05,
+      "loss": 0.5774,
+      "step": 6762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665646952754525e-05,
+      "loss": 0.599,
+      "step": 6763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665543254763983e-05,
+      "loss": 0.5507,
+      "step": 6764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6654395439242762e-05,
+      "loss": 0.605,
+      "step": 6765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653358202374068e-05,
+      "loss": 0.6159,
+      "step": 6766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665232083705377e-05,
+      "loss": 0.5589,
+      "step": 6767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66512833433019e-05,
+      "loss": 0.6085,
+      "step": 6768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650245721138483e-05,
+      "loss": 0.5846,
+      "step": 6769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6649207970583557e-05,
+      "loss": 0.5593,
+      "step": 6770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648170091657153e-05,
+      "loss": 0.5779,
+      "step": 6771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664713208437931e-05,
+      "loss": 0.5776,
+      "step": 6772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664609394877007e-05,
+      "loss": 0.5845,
+      "step": 6773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664505568484947e-05,
+      "loss": 0.6036,
+      "step": 6774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664401729263756e-05,
+      "loss": 0.5696,
+      "step": 6775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6642978772154387e-05,
+      "loss": 0.5737,
+      "step": 6776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664194012342e-05,
+      "loss": 0.5823,
+      "step": 6777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640901346454447e-05,
+      "loss": 0.583,
+      "step": 6778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663986244127779e-05,
+      "loss": 0.5916,
+      "step": 6779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638823407910085e-05,
+      "loss": 0.5907,
+      "step": 6780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6637784246371393e-05,
+      "loss": 0.5363,
+      "step": 6781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663674495668177e-05,
+      "loss": 0.5771,
+      "step": 6782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635705538861288e-05,
+      "loss": 0.5796,
+      "step": 6783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634665992930008e-05,
+      "loss": 0.621,
+      "step": 6784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6633626318908003e-05,
+      "loss": 0.5683,
+      "step": 6785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6632586516815346e-05,
+      "loss": 0.5769,
+      "step": 6786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6631546586672113e-05,
+      "loss": 0.5592,
+      "step": 6787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630506528498377e-05,
+      "loss": 0.5943,
+      "step": 6788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662946634231422e-05,
+      "loss": 0.6006,
+      "step": 6789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662842602813972e-05,
+      "loss": 0.5797,
+      "step": 6790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6627385585994972e-05,
+      "loss": 0.5515,
+      "step": 6791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626345015900055e-05,
+      "loss": 0.5786,
+      "step": 6792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662530431787506e-05,
+      "loss": 0.5798,
+      "step": 6793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624263491940074e-05,
+      "loss": 0.5591,
+      "step": 6794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66232225381152e-05,
+      "loss": 0.5549,
+      "step": 6795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6622181456420532e-05,
+      "loss": 0.5405,
+      "step": 6796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6621140246876167e-05,
+      "loss": 0.5611,
+      "step": 6797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662009890950221e-05,
+      "loss": 0.5571,
+      "step": 6798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6619057444318766e-05,
+      "loss": 0.5633,
+      "step": 6799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6618015851345938e-05,
+      "loss": 0.5622,
+      "step": 6800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6616974130603837e-05,
+      "loss": 0.5848,
+      "step": 6801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615932282112575e-05,
+      "loss": 0.6121,
+      "step": 6802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6614890305892266e-05,
+      "loss": 0.61,
+      "step": 6803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6613848201963027e-05,
+      "loss": 0.6084,
+      "step": 6804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612805970344978e-05,
+      "loss": 0.5824,
+      "step": 6805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661176361105824e-05,
+      "loss": 0.6,
+      "step": 6806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661072112412293e-05,
+      "loss": 0.5797,
+      "step": 6807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660967850955919e-05,
+      "loss": 0.5912,
+      "step": 6808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660863576738714e-05,
+      "loss": 0.5749,
+      "step": 6809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6607592897626904e-05,
+      "loss": 0.5959,
+      "step": 6810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606549900298633e-05,
+      "loss": 0.6278,
+      "step": 6811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660550677542245e-05,
+      "loss": 0.6066,
+      "step": 6812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604463523018493e-05,
+      "loss": 0.5756,
+      "step": 6813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6603420143106913e-05,
+      "loss": 0.6073,
+      "step": 6814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6602376635707852e-05,
+      "loss": 0.6079,
+      "step": 6815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660133300084145e-05,
+      "loss": 0.6004,
+      "step": 6816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6600289238527857e-05,
+      "loss": 0.6196,
+      "step": 6817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659924534878723e-05,
+      "loss": 0.5594,
+      "step": 6818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6598201331639714e-05,
+      "loss": 0.5668,
+      "step": 6819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6597157187105475e-05,
+      "loss": 0.5834,
+      "step": 6820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659611291520466e-05,
+      "loss": 0.584,
+      "step": 6821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6595068515957444e-05,
+      "loss": 0.5653,
+      "step": 6822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594023989383977e-05,
+      "loss": 0.5779,
+      "step": 6823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6592979335504436e-05,
+      "loss": 0.5638,
+      "step": 6824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659193455433898e-05,
+      "loss": 0.5698,
+      "step": 6825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590889645907787e-05,
+      "loss": 0.6136,
+      "step": 6826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658984461023102e-05,
+      "loss": 0.5711,
+      "step": 6827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588799447328874e-05,
+      "loss": 0.5944,
+      "step": 6828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587754157221507e-05,
+      "loss": 0.5832,
+      "step": 6829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658670873992911e-05,
+      "loss": 0.5666,
+      "step": 6830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6585663195471865e-05,
+      "loss": 0.5997,
+      "step": 6831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6584617523869956e-05,
+      "loss": 0.5665,
+      "step": 6832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583571725143574e-05,
+      "loss": 0.6179,
+      "step": 6833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6582525799312904e-05,
+      "loss": 0.5807,
+      "step": 6834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581479746398145e-05,
+      "loss": 0.5872,
+      "step": 6835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580433566419482e-05,
+      "loss": 0.5684,
+      "step": 6836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657938725939713e-05,
+      "loss": 0.5804,
+      "step": 6837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657834082535127e-05,
+      "loss": 0.5977,
+      "step": 6838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577294264302125e-05,
+      "loss": 0.5673,
+      "step": 6839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657624757626988e-05,
+      "loss": 0.5912,
+      "step": 6840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575200761274758e-05,
+      "loss": 0.6149,
+      "step": 6841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657415381933696e-05,
+      "loss": 0.6063,
+      "step": 6842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6573106750476706e-05,
+      "loss": 0.5855,
+      "step": 6843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6572059554714204e-05,
+      "loss": 0.5835,
+      "step": 6844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6571012232069678e-05,
+      "loss": 0.5517,
+      "step": 6845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656996478256334e-05,
+      "loss": 0.593,
+      "step": 6846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656891720621542e-05,
+      "loss": 0.5768,
+      "step": 6847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567869503046135e-05,
+      "loss": 0.5906,
+      "step": 6848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656682167307572e-05,
+      "loss": 0.6033,
+      "step": 6849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65657737163244e-05,
+      "loss": 0.6076,
+      "step": 6850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656472563281241e-05,
+      "loss": 0.5673,
+      "step": 6851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656367742255998e-05,
+      "loss": 0.5877,
+      "step": 6852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6562629085587355e-05,
+      "loss": 0.6081,
+      "step": 6853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561580621914764e-05,
+      "loss": 0.6066,
+      "step": 6854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560532031562456e-05,
+      "loss": 0.6024,
+      "step": 6855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655948331455068e-05,
+      "loss": 0.5815,
+      "step": 6856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655843447089967e-05,
+      "loss": 0.5977,
+      "step": 6857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6557385500629686e-05,
+      "loss": 0.5992,
+      "step": 6858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655633640376097e-05,
+      "loss": 0.5625,
+      "step": 6859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6555287180313787e-05,
+      "loss": 0.6012,
+      "step": 6860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655423783030839e-05,
+      "loss": 0.5668,
+      "step": 6861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655318835376503e-05,
+      "loss": 0.6115,
+      "step": 6862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552138750703984e-05,
+      "loss": 0.5781,
+      "step": 6863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65510890211455e-05,
+      "loss": 0.5879,
+      "step": 6864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655003916510985e-05,
+      "loss": 0.5751,
+      "step": 6865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6548989182617312e-05,
+      "loss": 0.5876,
+      "step": 6866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6547939073688148e-05,
+      "loss": 0.6125,
+      "step": 6867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6546888838342626e-05,
+      "loss": 0.5534,
+      "step": 6868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6545838476601035e-05,
+      "loss": 0.5749,
+      "step": 6869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544787988483647e-05,
+      "loss": 0.5734,
+      "step": 6870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6543737374010742e-05,
+      "loss": 0.5812,
+      "step": 6871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654268663320261e-05,
+      "loss": 0.5793,
+      "step": 6872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654163576607953e-05,
+      "loss": 0.5858,
+      "step": 6873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540584772661786e-05,
+      "loss": 0.5877,
+      "step": 6874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6539533652969683e-05,
+      "loss": 0.584,
+      "step": 6875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538482407023504e-05,
+      "loss": 0.5438,
+      "step": 6876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653743103484355e-05,
+      "loss": 0.5815,
+      "step": 6877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653637953645011e-05,
+      "loss": 0.5966,
+      "step": 6878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65353279118635e-05,
+      "loss": 0.5801,
+      "step": 6879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534276161104003e-05,
+      "loss": 0.5768,
+      "step": 6880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6533224284191943e-05,
+      "loss": 0.602,
+      "step": 6881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532172281147617e-05,
+      "loss": 0.5709,
+      "step": 6882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653112015199134e-05,
+      "loss": 0.572,
+      "step": 6883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653006789674342e-05,
+      "loss": 0.5825,
+      "step": 6884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529015515424178e-05,
+      "loss": 0.5901,
+      "step": 6885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6527963008053925e-05,
+      "loss": 0.6032,
+      "step": 6886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652691037465299e-05,
+      "loss": 0.594,
+      "step": 6887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6525857615241686e-05,
+      "loss": 0.6057,
+      "step": 6888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524804729840344e-05,
+      "loss": 0.5594,
+      "step": 6889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652375171846929e-05,
+      "loss": 0.5662,
+      "step": 6890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652269858114885e-05,
+      "loss": 0.5914,
+      "step": 6891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652164531789936e-05,
+      "loss": 0.5997,
+      "step": 6892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6520591928741158e-05,
+      "loss": 0.5558,
+      "step": 6893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651953841369457e-05,
+      "loss": 0.6233,
+      "step": 6894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651848477277995e-05,
+      "loss": 0.5669,
+      "step": 6895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517431006017627e-05,
+      "loss": 0.5866,
+      "step": 6896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651637711342795e-05,
+      "loss": 0.5752,
+      "step": 6897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651532309503127e-05,
+      "loss": 0.6068,
+      "step": 6898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651426895084793e-05,
+      "loss": 0.5833,
+      "step": 6899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651321468089828e-05,
+      "loss": 0.5921,
+      "step": 6900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6512160285202683e-05,
+      "loss": 0.6011,
+      "step": 6901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511105763781493e-05,
+      "loss": 0.5648,
+      "step": 6902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651005111665506e-05,
+      "loss": 0.579,
+      "step": 6903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6508996343843755e-05,
+      "loss": 0.6036,
+      "step": 6904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6507941445367935e-05,
+      "loss": 0.5584,
+      "step": 6905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6506886421247975e-05,
+      "loss": 0.5688,
+      "step": 6906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6505831271504236e-05,
+      "loss": 0.2469,
+      "step": 6907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504775996157092e-05,
+      "loss": 0.5423,
+      "step": 6908
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6503720595226912e-05,
+      "loss": 0.574,
+      "step": 6909
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502665068734082e-05,
+      "loss": 0.5673,
+      "step": 6910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6501609416698966e-05,
+      "loss": 0.5861,
+      "step": 6911
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650055363914196e-05,
+      "loss": 0.6224,
+      "step": 6912
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6499497736083434e-05,
+      "loss": 0.6286,
+      "step": 6913
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649844170754378e-05,
+      "loss": 0.6402,
+      "step": 6914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6497385553543388e-05,
+      "loss": 0.5739,
+      "step": 6915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496329274102643e-05,
+      "loss": 0.6109,
+      "step": 6916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495272869241942e-05,
+      "loss": 0.5727,
+      "step": 6917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649421633898168e-05,
+      "loss": 0.5783,
+      "step": 6918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649315968334225e-05,
+      "loss": 0.5705,
+      "step": 6919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6492102902344056e-05,
+      "loss": 0.5944,
+      "step": 6920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64910459960075e-05,
+      "loss": 0.6065,
+      "step": 6921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648998896435299e-05,
+      "loss": 0.5686,
+      "step": 6922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648893180740093e-05,
+      "loss": 0.574,
+      "step": 6923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6487874525171724e-05,
+      "loss": 0.6132,
+      "step": 6924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486817117685795e-05,
+      "loss": 0.573,
+      "step": 6925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6485759584963554e-05,
+      "loss": 0.5753,
+      "step": 6926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648470192702541e-05,
+      "loss": 0.5923,
+      "step": 6927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64836441438918e-05,
+      "loss": 0.5878,
+      "step": 6928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6482586235583128e-05,
+      "loss": 0.5964,
+      "step": 6929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481528202119826e-05,
+      "loss": 0.5775,
+      "step": 6930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6480470043522322e-05,
+      "loss": 0.5804,
+      "step": 6931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6479411759811038e-05,
+      "loss": 0.5805,
+      "step": 6932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6478353351006417e-05,
+      "loss": 0.5814,
+      "step": 6933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6477294817128885e-05,
+      "loss": 0.5929,
+      "step": 6934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6476236158198878e-05,
+      "loss": 0.5846,
+      "step": 6935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475177374236837e-05,
+      "loss": 0.5726,
+      "step": 6936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6474118465263205e-05,
+      "loss": 0.5891,
+      "step": 6937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6473059431298424e-05,
+      "loss": 0.5865,
+      "step": 6938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6472000272362937e-05,
+      "loss": 0.5667,
+      "step": 6939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6470940988477195e-05,
+      "loss": 0.5906,
+      "step": 6940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469881579661652e-05,
+      "loss": 0.5759,
+      "step": 6941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6468822045936756e-05,
+      "loss": 0.5694,
+      "step": 6942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6467762387322964e-05,
+      "loss": 0.5687,
+      "step": 6943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6466702603840733e-05,
+      "loss": 0.5681,
+      "step": 6944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646564269551053e-05,
+      "loss": 0.614,
+      "step": 6945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646458266235281e-05,
+      "loss": 0.6118,
+      "step": 6946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463522504388045e-05,
+      "loss": 0.576,
+      "step": 6947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6462462221636694e-05,
+      "loss": 0.5801,
+      "step": 6948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6461401814119233e-05,
+      "loss": 0.6025,
+      "step": 6949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6460341281856137e-05,
+      "loss": 0.5478,
+      "step": 6950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6459280624867876e-05,
+      "loss": 0.5923,
+      "step": 6951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645821984317493e-05,
+      "loss": 0.6191,
+      "step": 6952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6457158936797776e-05,
+      "loss": 0.6051,
+      "step": 6953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64560979057569e-05,
+      "loss": 0.5698,
+      "step": 6954
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645503675007278e-05,
+      "loss": 0.5718,
+      "step": 6955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6453975469765913e-05,
+      "loss": 0.6198,
+      "step": 6956
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645291406485678e-05,
+      "loss": 0.5579,
+      "step": 6957
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6451852535365876e-05,
+      "loss": 0.5825,
+      "step": 6958
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6450790881313693e-05,
+      "loss": 0.5712,
+      "step": 6959
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449729102720735e-05,
+      "loss": 0.5776,
+      "step": 6960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6448667199607492e-05,
+      "loss": 0.605,
+      "step": 6961
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6447605171994467e-05,
+      "loss": 0.5918,
+      "step": 6962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644654301990217e-05,
+      "loss": 0.5768,
+      "step": 6963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64454807433511e-05,
+      "loss": 0.5522,
+      "step": 6964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444418342361766e-05,
+      "loss": 0.5817,
+      "step": 6965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6443355816954688e-05,
+      "loss": 0.5533,
+      "step": 6966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442293167150366e-05,
+      "loss": 0.5969,
+      "step": 6967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441230392969325e-05,
+      "loss": 0.5905,
+      "step": 6968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6440167494432083e-05,
+      "loss": 0.5948,
+      "step": 6969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6439104471559157e-05,
+      "loss": 0.5759,
+      "step": 6970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6438041324371074e-05,
+      "loss": 0.5915,
+      "step": 6971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6436978052888353e-05,
+      "loss": 0.5711,
+      "step": 6972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643591465713153e-05,
+      "loss": 0.5737,
+      "step": 6973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6434851137121126e-05,
+      "loss": 0.6115,
+      "step": 6974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643378749287768e-05,
+      "loss": 0.6093,
+      "step": 6975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643272372442173e-05,
+      "loss": 0.5393,
+      "step": 6976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6431659831773803e-05,
+      "loss": 0.6029,
+      "step": 6977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643059581495445e-05,
+      "loss": 0.5711,
+      "step": 6978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429531673984205e-05,
+      "loss": 0.2485,
+      "step": 6979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428467408883614e-05,
+      "loss": 0.6092,
+      "step": 6980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642740301967323e-05,
+      "loss": 0.5988,
+      "step": 6981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64263385063736e-05,
+      "loss": 0.5725,
+      "step": 6982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425273869005266e-05,
+      "loss": 0.6266,
+      "step": 6983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424209107588796e-05,
+      "loss": 0.5792,
+      "step": 6984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423144222144737e-05,
+      "loss": 0.5923,
+      "step": 6985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422079212693655e-05,
+      "loss": 0.5847,
+      "step": 6986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6421014079256106e-05,
+      "loss": 0.6079,
+      "step": 6987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6419948821852653e-05,
+      "loss": 0.5957,
+      "step": 6988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6418883440503868e-05,
+      "loss": 0.5787,
+      "step": 6989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6417817935230318e-05,
+      "loss": 0.5669,
+      "step": 6990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641675230605257e-05,
+      "loss": 0.5476,
+      "step": 6991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64156865529912e-05,
+      "loss": 0.5854,
+      "step": 6992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6414620676066782e-05,
+      "loss": 0.5588,
+      "step": 6993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64135546752999e-05,
+      "loss": 0.554,
+      "step": 6994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641248855071113e-05,
+      "loss": 0.5907,
+      "step": 6995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6411422302321048e-05,
+      "loss": 0.5676,
+      "step": 6996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6410355930150252e-05,
+      "loss": 0.5734,
+      "step": 6997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409289434219323e-05,
+      "loss": 0.5713,
+      "step": 6998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640822281454885e-05,
+      "loss": 0.5735,
+      "step": 6999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6407156071159432e-05,
+      "loss": 0.5962,
+      "step": 7000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406089204071655e-05,
+      "loss": 0.5775,
+      "step": 7001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640502221330612e-05,
+      "loss": 0.6237,
+      "step": 7002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403955098883434e-05,
+      "loss": 0.5621,
+      "step": 7003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6402887860824185e-05,
+      "loss": 0.5495,
+      "step": 7004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6401820499148988e-05,
+      "loss": 0.5873,
+      "step": 7005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6400753013878444e-05,
+      "loss": 0.5868,
+      "step": 7006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399685405033168e-05,
+      "loss": 0.575,
+      "step": 7007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6398617672633765e-05,
+      "loss": 0.5945,
+      "step": 7008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6397549816700854e-05,
+      "loss": 0.5753,
+      "step": 7009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639648183725505e-05,
+      "loss": 0.583,
+      "step": 7010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6395413734316966e-05,
+      "loss": 0.599,
+      "step": 7011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394345507907233e-05,
+      "loss": 0.5498,
+      "step": 7012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639327715804647e-05,
+      "loss": 0.5838,
+      "step": 7013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392208684755296e-05,
+      "loss": 0.5559,
+      "step": 7014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6391140088054353e-05,
+      "loss": 0.5408,
+      "step": 7015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6390071367964263e-05,
+      "loss": 0.5793,
+      "step": 7016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6389002524505658e-05,
+      "loss": 0.601,
+      "step": 7017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6387933557699177e-05,
+      "loss": 0.5787,
+      "step": 7018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638686446756546e-05,
+      "loss": 0.622,
+      "step": 7019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385795254125142e-05,
+      "loss": 0.611,
+      "step": 7020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6384725917398865e-05,
+      "loss": 0.5827,
+      "step": 7021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383656457407275e-05,
+      "loss": 0.5756,
+      "step": 7022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382586874171025e-05,
+      "loss": 0.5992,
+      "step": 7023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6381517167710757e-05,
+      "loss": 0.5831,
+      "step": 7024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6380447338047124e-05,
+      "loss": 0.5544,
+      "step": 7025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637937738520079e-05,
+      "loss": 0.5729,
+      "step": 7026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378307309192397e-05,
+      "loss": 0.5724,
+      "step": 7027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377237110042615e-05,
+      "loss": 0.5697,
+      "step": 7028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6376166787772104e-05,
+      "loss": 0.5585,
+      "step": 7029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6375096342401523e-05,
+      "loss": 0.5941,
+      "step": 7030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6374025773951542e-05,
+      "loss": 0.5734,
+      "step": 7031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372955082442832e-05,
+      "loss": 0.5466,
+      "step": 7032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6371884267896058e-05,
+      "loss": 0.562,
+      "step": 7033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6370813330331897e-05,
+      "loss": 0.6257,
+      "step": 7034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369742269771024e-05,
+      "loss": 0.5736,
+      "step": 7035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636867108623412e-05,
+      "loss": 0.5488,
+      "step": 7036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636759977974186e-05,
+      "loss": 0.6003,
+      "step": 7037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636652835031493e-05,
+      "loss": 0.6103,
+      "step": 7038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636545679797402e-05,
+      "loss": 0.5766,
+      "step": 7039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364385122739812e-05,
+      "loss": 0.584,
+      "step": 7040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363313324632995e-05,
+      "loss": 0.5497,
+      "step": 7041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362241403674266e-05,
+      "loss": 0.5884,
+      "step": 7042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6361169359884316e-05,
+      "loss": 0.6026,
+      "step": 7043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6360097193283846e-05,
+      "loss": 0.5526,
+      "step": 7044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359024903893553e-05,
+      "loss": 0.5471,
+      "step": 7045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6357952491734136e-05,
+      "loss": 0.5968,
+      "step": 7046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63568799568263e-05,
+      "loss": 0.5659,
+      "step": 7047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635580729919076e-05,
+      "loss": 0.5555,
+      "step": 7048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635473451884822e-05,
+      "loss": 0.5938,
+      "step": 7049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353661615819386e-05,
+      "loss": 0.5859,
+      "step": 7050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352588590124978e-05,
+      "loss": 0.5829,
+      "step": 7051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635151544178571e-05,
+      "loss": 0.6042,
+      "step": 7052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6350442170822303e-05,
+      "loss": 0.6043,
+      "step": 7053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6349368777255473e-05,
+      "loss": 0.5986,
+      "step": 7054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348295261105946e-05,
+      "loss": 0.5507,
+      "step": 7055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347221622394452e-05,
+      "loss": 0.5893,
+      "step": 7056
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634614786114171e-05,
+      "loss": 0.5798,
+      "step": 7057
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345073977368455e-05,
+      "loss": 0.5575,
+      "step": 7058
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343999971095422e-05,
+      "loss": 0.6068,
+      "step": 7059
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6342925842343343e-05,
+      "loss": 0.5802,
+      "step": 7060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6341851591132957e-05,
+      "loss": 0.5659,
+      "step": 7061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6340777217485e-05,
+      "loss": 0.5821,
+      "step": 7062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6339702721420222e-05,
+      "loss": 0.5745,
+      "step": 7063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6338628102959358e-05,
+      "loss": 0.5921,
+      "step": 7064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337553362123165e-05,
+      "loss": 0.5891,
+      "step": 7065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336478498932382e-05,
+      "loss": 0.6049,
+      "step": 7066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6335403513407765e-05,
+      "loss": 0.5454,
+      "step": 7067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6334328405570072e-05,
+      "loss": 0.569,
+      "step": 7068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333253175440052e-05,
+      "loss": 0.598,
+      "step": 7069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6332177823038472e-05,
+      "loss": 0.5489,
+      "step": 7070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331102348386087e-05,
+      "loss": 0.563,
+      "step": 7071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633002675150366e-05,
+      "loss": 0.5717,
+      "step": 7072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6328951032411958e-05,
+      "loss": 0.5372,
+      "step": 7073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632787519113175e-05,
+      "loss": 0.5512,
+      "step": 7074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6326799227683806e-05,
+      "loss": 0.5953,
+      "step": 7075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63257231420889e-05,
+      "loss": 0.5717,
+      "step": 7076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6324646934367806e-05,
+      "loss": 0.5678,
+      "step": 7077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323570604541304e-05,
+      "loss": 0.613,
+      "step": 7078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632249415263017e-05,
+      "loss": 0.5864,
+      "step": 7079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632141757865519e-05,
+      "loss": 0.5784,
+      "step": 7080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320340882637143e-05,
+      "loss": 0.5932,
+      "step": 7081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6319264064596823e-05,
+      "loss": 0.547,
+      "step": 7082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6318187124555013e-05,
+      "loss": 0.5892,
+      "step": 7083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631711006253251e-05,
+      "loss": 0.5674,
+      "step": 7084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6316032878550107e-05,
+      "loss": 0.5834,
+      "step": 7085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6314955572628597e-05,
+      "loss": 0.5878,
+      "step": 7086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313878144788783e-05,
+      "loss": 0.5699,
+      "step": 7087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631280059505146e-05,
+      "loss": 0.5604,
+      "step": 7088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6311722923437442e-05,
+      "loss": 0.5952,
+      "step": 7089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310645129967525e-05,
+      "loss": 0.5883,
+      "step": 7090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630956721466252e-05,
+      "loss": 0.5988,
+      "step": 7091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630848917754324e-05,
+      "loss": 0.5905,
+      "step": 7092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307411018630496e-05,
+      "loss": 0.5552,
+      "step": 7093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6306332737945103e-05,
+      "loss": 0.5561,
+      "step": 7094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630525433550788e-05,
+      "loss": 0.5685,
+      "step": 7095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6304175811339643e-05,
+      "loss": 0.6015,
+      "step": 7096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6303097165461217e-05,
+      "loss": 0.629,
+      "step": 7097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630201839789343e-05,
+      "loss": 0.5876,
+      "step": 7098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300939508657103e-05,
+      "loss": 0.6103,
+      "step": 7099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629986049777307e-05,
+      "loss": 0.5983,
+      "step": 7100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629878136526216e-05,
+      "loss": 0.5724,
+      "step": 7101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297702111145207e-05,
+      "loss": 0.5693,
+      "step": 7102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6296622735443047e-05,
+      "loss": 0.5631,
+      "step": 7103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295543238176525e-05,
+      "loss": 0.5754,
+      "step": 7104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6294463619366473e-05,
+      "loss": 0.6066,
+      "step": 7105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629338387903374e-05,
+      "loss": 0.5729,
+      "step": 7106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629230401719917e-05,
+      "loss": 0.5461,
+      "step": 7107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629122403388361e-05,
+      "loss": 0.556,
+      "step": 7108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629014392910791e-05,
+      "loss": 0.5272,
+      "step": 7109
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6289063702892932e-05,
+      "loss": 0.596,
+      "step": 7110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6287983355259517e-05,
+      "loss": 0.5897,
+      "step": 7111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6286902886228537e-05,
+      "loss": 0.5828,
+      "step": 7112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285822295820835e-05,
+      "loss": 0.591,
+      "step": 7113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6284741584057287e-05,
+      "loss": 0.5902,
+      "step": 7114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283660750958752e-05,
+      "loss": 0.5794,
+      "step": 7115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6282579796546095e-05,
+      "loss": 0.5802,
+      "step": 7116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628149872084019e-05,
+      "loss": 0.6077,
+      "step": 7117
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628041752386191e-05,
+      "loss": 0.5734,
+      "step": 7118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6279336205632122e-05,
+      "loss": 0.5386,
+      "step": 7119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278254766171703e-05,
+      "loss": 0.6054,
+      "step": 7120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277173205501537e-05,
+      "loss": 0.6152,
+      "step": 7121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276091523642504e-05,
+      "loss": 0.5748,
+      "step": 7122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6275009720615483e-05,
+      "loss": 0.5615,
+      "step": 7123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627392779644136e-05,
+      "loss": 0.5966,
+      "step": 7124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272845751141026e-05,
+      "loss": 0.5659,
+      "step": 7125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271763584735373e-05,
+      "loss": 0.5599,
+      "step": 7126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270681297245286e-05,
+      "loss": 0.5676,
+      "step": 7127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626959888869167e-05,
+      "loss": 0.579,
+      "step": 7128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626851635909541e-05,
+      "loss": 0.2602,
+      "step": 7129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626743370847742e-05,
+      "loss": 0.5808,
+      "step": 7130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626635093685859e-05,
+      "loss": 0.6064,
+      "step": 7131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626526804425983e-05,
+      "loss": 0.565,
+      "step": 7132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264185030702046e-05,
+      "loss": 0.5731,
+      "step": 7133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6263101896206147e-05,
+      "loss": 0.5664,
+      "step": 7134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262018640793042e-05,
+      "loss": 0.5894,
+      "step": 7135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6260935264483647e-05,
+      "loss": 0.6174,
+      "step": 7136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625985176729888e-05,
+      "loss": 0.5881,
+      "step": 7137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625876814925965e-05,
+      "loss": 0.5809,
+      "step": 7138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625768441038689e-05,
+      "loss": 0.5716,
+      "step": 7139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256600550701513e-05,
+      "loss": 0.5631,
+      "step": 7140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625551657022445e-05,
+      "loss": 0.2539,
+      "step": 7141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6254432468976626e-05,
+      "loss": 0.5842,
+      "step": 7142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625334824697898e-05,
+      "loss": 0.5591,
+      "step": 7143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6252263904252424e-05,
+      "loss": 0.5726,
+      "step": 7144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251179440817915e-05,
+      "loss": 0.6083,
+      "step": 7145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250094856696376e-05,
+      "loss": 0.5857,
+      "step": 7146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624901015190875e-05,
+      "loss": 0.5665,
+      "step": 7147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624792532647598e-05,
+      "loss": 0.5577,
+      "step": 7148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246840380419008e-05,
+      "loss": 0.5556,
+      "step": 7149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624575531375878e-05,
+      "loss": 0.6162,
+      "step": 7150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6244670126516248e-05,
+      "loss": 0.591,
+      "step": 7151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243584818712358e-05,
+      "loss": 0.5864,
+      "step": 7152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624249939036807e-05,
+      "loss": 0.5796,
+      "step": 7153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624141384150433e-05,
+      "loss": 0.5972,
+      "step": 7154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6240328172142104e-05,
+      "loss": 0.5534,
+      "step": 7155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239242382302354e-05,
+      "loss": 0.5855,
+      "step": 7156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623815647200603e-05,
+      "loss": 0.6204,
+      "step": 7157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623707044127411e-05,
+      "loss": 0.5599,
+      "step": 7158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6235984290127556e-05,
+      "loss": 0.5833,
+      "step": 7159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.6166,
+      "step": 7160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233811626674426e-05,
+      "loss": 0.5569,
+      "step": 7161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6232725114409797e-05,
+      "loss": 0.5616,
+      "step": 7162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6231638481814424e-05,
+      "loss": 0.5972,
+      "step": 7163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6230551728909293e-05,
+      "loss": 0.6044,
+      "step": 7164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229464855715376e-05,
+      "loss": 0.5876,
+      "step": 7165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228377862253664e-05,
+      "loss": 0.5511,
+      "step": 7166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6227290748545136e-05,
+      "loss": 0.6008,
+      "step": 7167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6226203514610787e-05,
+      "loss": 0.5904,
+      "step": 7168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6225116160471603e-05,
+      "loss": 0.5848,
+      "step": 7169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622402868614858e-05,
+      "loss": 0.5368,
+      "step": 7170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222941091662706e-05,
+      "loss": 0.5611,
+      "step": 7171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221853377034988e-05,
+      "loss": 0.5651,
+      "step": 7172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622076554228642e-05,
+      "loss": 0.5373,
+      "step": 7173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6219677587438007e-05,
+      "loss": 0.5982,
+      "step": 7174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621858951251075e-05,
+      "loss": 0.5768,
+      "step": 7175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6217501317525653e-05,
+      "loss": 0.5974,
+      "step": 7176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216413002503736e-05,
+      "loss": 0.5569,
+      "step": 7177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215324567466e-05,
+      "loss": 0.5728,
+      "step": 7178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6214236012433464e-05,
+      "loss": 0.5926,
+      "step": 7179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621314733742714e-05,
+      "loss": 0.6068,
+      "step": 7180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621205854246805e-05,
+      "loss": 0.5682,
+      "step": 7181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6210969627577213e-05,
+      "loss": 0.6171,
+      "step": 7182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620988059277565e-05,
+      "loss": 0.5886,
+      "step": 7183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208791438084392e-05,
+      "loss": 0.5561,
+      "step": 7184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620770216352446e-05,
+      "loss": 0.5848,
+      "step": 7185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206612769116885e-05,
+      "loss": 0.5765,
+      "step": 7186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6205523254882702e-05,
+      "loss": 0.5488,
+      "step": 7187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6204433620842947e-05,
+      "loss": 0.5651,
+      "step": 7188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.620334386701865e-05,
+      "loss": 0.5898,
+      "step": 7189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202253993430853e-05,
+      "loss": 0.6073,
+      "step": 7190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6201164000100602e-05,
+      "loss": 0.5397,
+      "step": 7191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6200073887048936e-05,
+      "loss": 0.5967,
+      "step": 7192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6198983654296904e-05,
+      "loss": 0.5762,
+      "step": 7193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619789330186555e-05,
+      "loss": 0.5691,
+      "step": 7194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6196802829775928e-05,
+      "loss": 0.5606,
+      "step": 7195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619571223804909e-05,
+      "loss": 0.6009,
+      "step": 7196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6194621526706093e-05,
+      "loss": 0.6044,
+      "step": 7197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619353069576799e-05,
+      "loss": 0.594,
+      "step": 7198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6192439745255844e-05,
+      "loss": 0.5818,
+      "step": 7199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6191348675190718e-05,
+      "loss": 0.564,
+      "step": 7200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190257485593677e-05,
+      "loss": 0.5877,
+      "step": 7201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189166176485784e-05,
+      "loss": 0.5566,
+      "step": 7202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618807474788811e-05,
+      "loss": 0.5648,
+      "step": 7203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618698319982173e-05,
+      "loss": 0.5488,
+      "step": 7204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618589153230771e-05,
+      "loss": 0.6104,
+      "step": 7205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6184799745367135e-05,
+      "loss": 0.5711,
+      "step": 7206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183707839021077e-05,
+      "loss": 0.5984,
+      "step": 7207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6182615813290618e-05,
+      "loss": 0.6023,
+      "step": 7208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6181523668196842e-05,
+      "loss": 0.5908,
+      "step": 7209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6180431403760836e-05,
+      "loss": 0.2613,
+      "step": 7210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179339020003685e-05,
+      "loss": 0.6019,
+      "step": 7211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6178246516946477e-05,
+      "loss": 0.6009,
+      "step": 7212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177153894610306e-05,
+      "loss": 0.5964,
+      "step": 7213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617606115301627e-05,
+      "loss": 0.5823,
+      "step": 7214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6174968292185465e-05,
+      "loss": 0.5462,
+      "step": 7215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6173875312138983e-05,
+      "loss": 0.5438,
+      "step": 7216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617278221289793e-05,
+      "loss": 0.566,
+      "step": 7217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6171688994483413e-05,
+      "loss": 0.5691,
+      "step": 7218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6170595656916535e-05,
+      "loss": 0.5723,
+      "step": 7219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169502200218403e-05,
+      "loss": 0.5775,
+      "step": 7220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168408624410125e-05,
+      "loss": 0.5539,
+      "step": 7221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616731492951282e-05,
+      "loss": 0.577,
+      "step": 7222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61662211155476e-05,
+      "loss": 0.5202,
+      "step": 7223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6165127182535586e-05,
+      "loss": 0.5864,
+      "step": 7224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164033130497894e-05,
+      "loss": 0.589,
+      "step": 7225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6162938959455647e-05,
+      "loss": 0.5327,
+      "step": 7226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161844669429968e-05,
+      "loss": 0.5799,
+      "step": 7227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616075026044199e-05,
+      "loss": 0.5683,
+      "step": 7228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615965573251283e-05,
+      "loss": 0.5379,
+      "step": 7229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615856108566363e-05,
+      "loss": 0.5914,
+      "step": 7230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615746631991552e-05,
+      "loss": 0.2553,
+      "step": 7231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6156371435289636e-05,
+      "loss": 0.5657,
+      "step": 7232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6155276431807116e-05,
+      "loss": 0.5507,
+      "step": 7233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61541813094891e-05,
+      "loss": 0.5927,
+      "step": 7234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6153086068356732e-05,
+      "loss": 0.5887,
+      "step": 7235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6151990708431157e-05,
+      "loss": 0.5923,
+      "step": 7236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615089522973352e-05,
+      "loss": 0.5624,
+      "step": 7237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6149799632284972e-05,
+      "loss": 0.5848,
+      "step": 7238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6148703916106666e-05,
+      "loss": 0.5748,
+      "step": 7239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6147608081219753e-05,
+      "loss": 0.5537,
+      "step": 7240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6146512127645393e-05,
+      "loss": 0.5778,
+      "step": 7241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6145416055404744e-05,
+      "loss": 0.5902,
+      "step": 7242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6144319864518966e-05,
+      "loss": 0.5796,
+      "step": 7243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6143223555009222e-05,
+      "loss": 0.5848,
+      "step": 7244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142127126896682e-05,
+      "loss": 0.5568,
+      "step": 7245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6141030580202507e-05,
+      "loss": 0.585,
+      "step": 7246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613993391494787e-05,
+      "loss": 0.5699,
+      "step": 7247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6138837131153948e-05,
+      "loss": 0.599,
+      "step": 7248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6137740228841908e-05,
+      "loss": 0.5819,
+      "step": 7249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6136643208032937e-05,
+      "loss": 0.552,
+      "step": 7250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6135546068748203e-05,
+      "loss": 0.5957,
+      "step": 7251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6134448811008897e-05,
+      "loss": 0.5727,
+      "step": 7252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6133351434836198e-05,
+      "loss": 0.626,
+      "step": 7253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6132253940251294e-05,
+      "loss": 0.5437,
+      "step": 7254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6131156327275372e-05,
+      "loss": 0.5691,
+      "step": 7255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6130058595929625e-05,
+      "loss": 0.5789,
+      "step": 7256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6128960746235242e-05,
+      "loss": 0.5966,
+      "step": 7257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612786277821343e-05,
+      "loss": 0.5828,
+      "step": 7258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612676469188537e-05,
+      "loss": 0.5483,
+      "step": 7259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6125666487272268e-05,
+      "loss": 0.5865,
+      "step": 7260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6124568164395336e-05,
+      "loss": 0.6137,
+      "step": 7261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6123469723275766e-05,
+      "loss": 0.5867,
+      "step": 7262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612237116393477e-05,
+      "loss": 0.56,
+      "step": 7263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612127248639356e-05,
+      "loss": 0.5447,
+      "step": 7264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.612017369067334e-05,
+      "loss": 0.5751,
+      "step": 7265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611907477679533e-05,
+      "loss": 0.5509,
+      "step": 7266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611797574478074e-05,
+      "loss": 0.5736,
+      "step": 7267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61168765946508e-05,
+      "loss": 0.5391,
+      "step": 7268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6115777326426717e-05,
+      "loss": 0.6019,
+      "step": 7269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114677940129716e-05,
+      "loss": 0.5635,
+      "step": 7270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611357843578103e-05,
+      "loss": 0.5696,
+      "step": 7271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112478813401882e-05,
+      "loss": 0.6053,
+      "step": 7272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61113790730135e-05,
+      "loss": 0.6015,
+      "step": 7273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6110279214637115e-05,
+      "loss": 0.5662,
+      "step": 7274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6109179238293967e-05,
+      "loss": 0.5813,
+      "step": 7275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108079144005286e-05,
+      "loss": 0.598,
+      "step": 7276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106978931792315e-05,
+      "loss": 0.5644,
+      "step": 7277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6105878601676296e-05,
+      "loss": 0.5739,
+      "step": 7278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6104778153678467e-05,
+      "loss": 0.2405,
+      "step": 7279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6103677587820076e-05,
+      "loss": 0.5884,
+      "step": 7280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610257690412237e-05,
+      "loss": 0.5748,
+      "step": 7281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101476102606602e-05,
+      "loss": 0.2444,
+      "step": 7282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610037518329402e-05,
+      "loss": 0.5502,
+      "step": 7283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6099274146205885e-05,
+      "loss": 0.5536,
+      "step": 7284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098172991363446e-05,
+      "loss": 0.6111,
+      "step": 7285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097071718787974e-05,
+      "loss": 0.5926,
+      "step": 7286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6095970328500716e-05,
+      "loss": 0.5851,
+      "step": 7287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6094868820522947e-05,
+      "loss": 0.6094,
+      "step": 7288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6093767194875923e-05,
+      "loss": 0.5764,
+      "step": 7289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092665451580925e-05,
+      "loss": 0.5964,
+      "step": 7290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609156359065921e-05,
+      "loss": 0.5604,
+      "step": 7291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6090461612132064e-05,
+      "loss": 0.5632,
+      "step": 7292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089359516020752e-05,
+      "loss": 0.5916,
+      "step": 7293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6088257302346556e-05,
+      "loss": 0.5973,
+      "step": 7294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6087154971130753e-05,
+      "loss": 0.6049,
+      "step": 7295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6086052522394625e-05,
+      "loss": 0.5879,
+      "step": 7296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608494995615946e-05,
+      "loss": 0.5771,
+      "step": 7297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6083847272446544e-05,
+      "loss": 0.5815,
+      "step": 7298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608274447127716e-05,
+      "loss": 0.5629,
+      "step": 7299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6081641552672604e-05,
+      "loss": 0.5749,
+      "step": 7300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608053851665417e-05,
+      "loss": 0.5697,
+      "step": 7301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6079435363243147e-05,
+      "loss": 0.5442,
+      "step": 7302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607833209246084e-05,
+      "loss": 0.5739,
+      "step": 7303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6077228704328545e-05,
+      "loss": 0.5592,
+      "step": 7304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076125198867573e-05,
+      "loss": 0.5506,
+      "step": 7305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075021576099215e-05,
+      "loss": 0.5786,
+      "step": 7306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6073917836044784e-05,
+      "loss": 0.541,
+      "step": 7307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607281397872559e-05,
+      "loss": 0.5767,
+      "step": 7308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6071710004162945e-05,
+      "loss": 0.5739,
+      "step": 7309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607060591237816e-05,
+      "loss": 0.5547,
+      "step": 7310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6069501703392553e-05,
+      "loss": 0.5708,
+      "step": 7311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606839737722744e-05,
+      "loss": 0.5979,
+      "step": 7312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067292933904144e-05,
+      "loss": 0.5828,
+      "step": 7313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6066188373443985e-05,
+      "loss": 0.5738,
+      "step": 7314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606508369586829e-05,
+      "loss": 0.5987,
+      "step": 7315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6063978901198387e-05,
+      "loss": 0.5836,
+      "step": 7316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6062873989455602e-05,
+      "loss": 0.5698,
+      "step": 7317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061768960661274e-05,
+      "loss": 0.5759,
+      "step": 7318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6060663814836726e-05,
+      "loss": 0.5657,
+      "step": 7319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60595585520033e-05,
+      "loss": 0.5479,
+      "step": 7320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605845317218234e-05,
+      "loss": 0.6124,
+      "step": 7321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605734767539518e-05,
+      "loss": 0.5509,
+      "step": 7322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6056242061663164e-05,
+      "loss": 0.5223,
+      "step": 7323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6055136331007635e-05,
+      "loss": 0.5915,
+      "step": 7324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6054030483449944e-05,
+      "loss": 0.5881,
+      "step": 7325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605292451901144e-05,
+      "loss": 0.6022,
+      "step": 7326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051818437713478e-05,
+      "loss": 0.5829,
+      "step": 7327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605071223957741e-05,
+      "loss": 0.5583,
+      "step": 7328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604960592462459e-05,
+      "loss": 0.5871,
+      "step": 7329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6048499492876378e-05,
+      "loss": 0.5806,
+      "step": 7330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6047392944354136e-05,
+      "loss": 0.5999,
+      "step": 7331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6046286279079227e-05,
+      "loss": 0.5536,
+      "step": 7332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045179497073017e-05,
+      "loss": 0.5867,
+      "step": 7333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044072598356875e-05,
+      "loss": 0.5733,
+      "step": 7334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604296558295217e-05,
+      "loss": 0.5578,
+      "step": 7335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6041858450880274e-05,
+      "loss": 0.6024,
+      "step": 7336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604075120216256e-05,
+      "loss": 0.5522,
+      "step": 7337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039643836820412e-05,
+      "loss": 0.571,
+      "step": 7338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6038536354875194e-05,
+      "loss": 0.5602,
+      "step": 7339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6037428756348306e-05,
+      "loss": 0.5515,
+      "step": 7340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603632104126112e-05,
+      "loss": 0.5726,
+      "step": 7341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6035213209635024e-05,
+      "loss": 0.5559,
+      "step": 7342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603410526149141e-05,
+      "loss": 0.5469,
+      "step": 7343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6032997196851662e-05,
+      "loss": 0.574,
+      "step": 7344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6031889015737176e-05,
+      "loss": 0.5689,
+      "step": 7345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6030780718169344e-05,
+      "loss": 0.5451,
+      "step": 7346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602967230416957e-05,
+      "loss": 0.5917,
+      "step": 7347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6028563773759244e-05,
+      "loss": 0.5654,
+      "step": 7348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6027455126959775e-05,
+      "loss": 0.5758,
+      "step": 7349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 0.5847,
+      "step": 7350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602523748427902e-05,
+      "loss": 0.561,
+      "step": 7351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024128488440547e-05,
+      "loss": 0.5518,
+      "step": 7352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602301937629856e-05,
+      "loss": 0.576,
+      "step": 7353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6021910147874464e-05,
+      "loss": 0.5982,
+      "step": 7354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6020800803189682e-05,
+      "loss": 0.5912,
+      "step": 7355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6019691342265628e-05,
+      "loss": 0.5767,
+      "step": 7356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6018581765123724e-05,
+      "loss": 0.5543,
+      "step": 7357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601747207178539e-05,
+      "loss": 0.6003,
+      "step": 7358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016362262272047e-05,
+      "loss": 0.5509,
+      "step": 7359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6015252336605124e-05,
+      "loss": 0.5825,
+      "step": 7360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6014142294806054e-05,
+      "loss": 0.5943,
+      "step": 7361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6013032136896264e-05,
+      "loss": 0.5869,
+      "step": 7362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6011921862897186e-05,
+      "loss": 0.5818,
+      "step": 7363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6010811472830253e-05,
+      "loss": 0.595,
+      "step": 7364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009700966716907e-05,
+      "loss": 0.5578,
+      "step": 7365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6008590344578587e-05,
+      "loss": 0.5787,
+      "step": 7366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6007479606436733e-05,
+      "loss": 0.5857,
+      "step": 7367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6006368752312792e-05,
+      "loss": 0.5726,
+      "step": 7368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6005257782228208e-05,
+      "loss": 0.5678,
+      "step": 7369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600414669620443e-05,
+      "loss": 0.5773,
+      "step": 7370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600303549426291e-05,
+      "loss": 0.5441,
+      "step": 7371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.60019241764251e-05,
+      "loss": 0.5922,
+      "step": 7372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6000812742712458e-05,
+      "loss": 0.5379,
+      "step": 7373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5999701193146434e-05,
+      "loss": 0.5596,
+      "step": 7374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59985895277485e-05,
+      "loss": 0.5815,
+      "step": 7375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5997477746540112e-05,
+      "loss": 0.5818,
+      "step": 7376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599636584954273e-05,
+      "loss": 0.5535,
+      "step": 7377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599525383677783e-05,
+      "loss": 0.5853,
+      "step": 7378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599414170826687e-05,
+      "loss": 0.5736,
+      "step": 7379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5993029464031327e-05,
+      "loss": 0.5618,
+      "step": 7380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5991917104092677e-05,
+      "loss": 0.5713,
+      "step": 7381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5990804628472393e-05,
+      "loss": 0.5402,
+      "step": 7382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989692037191948e-05,
+      "loss": 0.5627,
+      "step": 7383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5988579330272828e-05,
+      "loss": 0.586,
+      "step": 7384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598746650773651e-05,
+      "loss": 0.5459,
+      "step": 7385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598635356960449e-05,
+      "loss": 0.5535,
+      "step": 7386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985240515898244e-05,
+      "loss": 0.5997,
+      "step": 7387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5984127346639264e-05,
+      "loss": 0.6055,
+      "step": 7388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598301406184904e-05,
+      "loss": 0.5237,
+      "step": 7389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981900661549063e-05,
+      "loss": 0.5838,
+      "step": 7390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5980787145760834e-05,
+      "loss": 0.5701,
+      "step": 7391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597967351450585e-05,
+      "loss": 0.5869,
+      "step": 7392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5978559767805606e-05,
+      "loss": 0.611,
+      "step": 7393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5977445905681607e-05,
+      "loss": 0.5807,
+      "step": 7394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976331928155366e-05,
+      "loss": 0.5666,
+      "step": 7395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5975217835248377e-05,
+      "loss": 0.2438,
+      "step": 7396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5974103626982155e-05,
+      "loss": 0.5928,
+      "step": 7397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5972989303378207e-05,
+      "loss": 0.573,
+      "step": 7398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5971874864458056e-05,
+      "loss": 0.5472,
+      "step": 7399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597076031024321e-05,
+      "loss": 0.558,
+      "step": 7400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5969645640755187e-05,
+      "loss": 0.5732,
+      "step": 7401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5968530856015508e-05,
+      "loss": 0.5471,
+      "step": 7402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5967415956045695e-05,
+      "loss": 0.5871,
+      "step": 7403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5966300940867277e-05,
+      "loss": 0.5781,
+      "step": 7404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965185810501775e-05,
+      "loss": 0.5946,
+      "step": 7405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5964070564970718e-05,
+      "loss": 0.5633,
+      "step": 7406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962955204295646e-05,
+      "loss": 0.595,
+      "step": 7407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5961839728498083e-05,
+      "loss": 0.5735,
+      "step": 7408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5960724137599567e-05,
+      "loss": 0.5596,
+      "step": 7409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5959608431621634e-05,
+      "loss": 0.5821,
+      "step": 7410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5958492610585832e-05,
+      "loss": 0.5829,
+      "step": 7411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5957376674513696e-05,
+      "loss": 0.5844,
+      "step": 7412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595626062342677e-05,
+      "loss": 0.5705,
+      "step": 7413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955144457346606e-05,
+      "loss": 0.5864,
+      "step": 7414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595402817629475e-05,
+      "loss": 0.5757,
+      "step": 7415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5952911780292756e-05,
+      "loss": 0.5697,
+      "step": 7416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595179526936217e-05,
+      "loss": 0.6097,
+      "step": 7417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5950678643524557e-05,
+      "loss": 0.5493,
+      "step": 7418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949561902801472e-05,
+      "loss": 0.5944,
+      "step": 7419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594844504721447e-05,
+      "loss": 0.579,
+      "step": 7420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5947328076785118e-05,
+      "loss": 0.6242,
+      "step": 7421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5946210991534983e-05,
+      "loss": 0.5764,
+      "step": 7422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5945093791485623e-05,
+      "loss": 0.5996,
+      "step": 7423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594397647665862e-05,
+      "loss": 0.5804,
+      "step": 7424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594285904707553e-05,
+      "loss": 0.5819,
+      "step": 7425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594174150275794e-05,
+      "loss": 0.58,
+      "step": 7426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940623843727415e-05,
+      "loss": 0.5786,
+      "step": 7427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593950607000554e-05,
+      "loss": 0.5874,
+      "step": 7428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5938388181613893e-05,
+      "loss": 0.569,
+      "step": 7429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5937270178574056e-05,
+      "loss": 0.5558,
+      "step": 7430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5936152060907612e-05,
+      "loss": 0.5881,
+      "step": 7431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593503382863615e-05,
+      "loss": 0.5803,
+      "step": 7432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5933915481781256e-05,
+      "loss": 0.5836,
+      "step": 7433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5932797020364525e-05,
+      "loss": 0.6288,
+      "step": 7434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593167844440755e-05,
+      "loss": 0.5912,
+      "step": 7435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593055975393192e-05,
+      "loss": 0.5454,
+      "step": 7436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5929440948959244e-05,
+      "loss": 0.5608,
+      "step": 7437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592832202951111e-05,
+      "loss": 0.5667,
+      "step": 7438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592720299560913e-05,
+      "loss": 0.5937,
+      "step": 7439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5926083847274904e-05,
+      "loss": 0.5689,
+      "step": 7440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5924964584530033e-05,
+      "loss": 0.5997,
+      "step": 7441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923845207396137e-05,
+      "loss": 0.5872,
+      "step": 7442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592272571589482e-05,
+      "loss": 0.5627,
+      "step": 7443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59216061100477e-05,
+      "loss": 0.5738,
+      "step": 7444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920486389876383e-05,
+      "loss": 0.6003,
+      "step": 7445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5919366555402498e-05,
+      "loss": 0.5611,
+      "step": 7446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591824660664766e-05,
+      "loss": 0.6091,
+      "step": 7447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917126543633486e-05,
+      "loss": 0.5788,
+      "step": 7448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591600636638161e-05,
+      "loss": 0.5557,
+      "step": 7449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5914886074913648e-05,
+      "loss": 0.5706,
+      "step": 7450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591376566925124e-05,
+      "loss": 0.5762,
+      "step": 7451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591264514941601e-05,
+      "loss": 0.5838,
+      "step": 7452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5911524515429588e-05,
+      "loss": 0.5741,
+      "step": 7453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591040376731362e-05,
+      "loss": 0.6117,
+      "step": 7454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590928290508973e-05,
+      "loss": 0.6164,
+      "step": 7455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5908161928779566e-05,
+      "loss": 0.5684,
+      "step": 7456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590704083840477e-05,
+      "loss": 0.5787,
+      "step": 7457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5905919633986985e-05,
+      "loss": 0.5922,
+      "step": 7458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5904798315547852e-05,
+      "loss": 0.57,
+      "step": 7459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5903676883109032e-05,
+      "loss": 0.5895,
+      "step": 7460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590255533669216e-05,
+      "loss": 0.5791,
+      "step": 7461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59014336763189e-05,
+      "loss": 0.5584,
+      "step": 7462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590031190201091e-05,
+      "loss": 0.5972,
+      "step": 7463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5899190013789833e-05,
+      "loss": 0.559,
+      "step": 7464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589806801167734e-05,
+      "loss": 0.5969,
+      "step": 7465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589694589569509e-05,
+      "loss": 0.602,
+      "step": 7466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5895823665864744e-05,
+      "loss": 0.5358,
+      "step": 7467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5894701322207974e-05,
+      "loss": 0.5664,
+      "step": 7468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5893578864746442e-05,
+      "loss": 0.5833,
+      "step": 7469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892456293501823e-05,
+      "loss": 0.558,
+      "step": 7470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891333608495786e-05,
+      "loss": 0.5537,
+      "step": 7471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5890210809750006e-05,
+      "loss": 0.562,
+      "step": 7472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5889087897286166e-05,
+      "loss": 0.6084,
+      "step": 7473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5887964871125938e-05,
+      "loss": 0.5669,
+      "step": 7474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886841731291008e-05,
+      "loss": 0.5636,
+      "step": 7475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885718477803054e-05,
+      "loss": 0.5723,
+      "step": 7476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588459511068377e-05,
+      "loss": 0.565,
+      "step": 7477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588347162995484e-05,
+      "loss": 0.5909,
+      "step": 7478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5882348035637956e-05,
+      "loss": 0.5693,
+      "step": 7479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5881224327754807e-05,
+      "loss": 0.6005,
+      "step": 7480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880100506327084e-05,
+      "loss": 0.5796,
+      "step": 7481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5878976571376496e-05,
+      "loss": 0.5794,
+      "step": 7482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.5718,
+      "step": 7483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.58767283609935e-05,
+      "loss": 0.5583,
+      "step": 7484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875604085604496e-05,
+      "loss": 0.5873,
+      "step": 7485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5874479696779427e-05,
+      "loss": 0.5217,
+      "step": 7486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5873355194540008e-05,
+      "loss": 0.5765,
+      "step": 7487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587223057890794e-05,
+      "loss": 0.5691,
+      "step": 7488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5871105849904943e-05,
+      "loss": 0.5748,
+      "step": 7489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5869981007552725e-05,
+      "loss": 0.568,
+      "step": 7490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5868856051873005e-05,
+      "loss": 0.6181,
+      "step": 7491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5867730982887504e-05,
+      "loss": 0.5974,
+      "step": 7492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586660580061794e-05,
+      "loss": 0.5851,
+      "step": 7493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5865480505086032e-05,
+      "loss": 0.5639,
+      "step": 7494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5864355096313515e-05,
+      "loss": 0.5917,
+      "step": 7495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586322957432211e-05,
+      "loss": 0.5802,
+      "step": 7496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586210393913355e-05,
+      "loss": 0.5506,
+      "step": 7497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5860978190769563e-05,
+      "loss": 0.586,
+      "step": 7498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585985232925188e-05,
+      "loss": 0.5458,
+      "step": 7499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5858726354602248e-05,
+      "loss": 0.5688,
+      "step": 7500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5857600266842397e-05,
+      "loss": 0.5876,
+      "step": 7501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5856474065994074e-05,
+      "loss": 0.6007,
+      "step": 7502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5855347752079015e-05,
+      "loss": 0.5876,
+      "step": 7503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585422132511897e-05,
+      "loss": 0.5794,
+      "step": 7504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853094785135682e-05,
+      "loss": 0.5421,
+      "step": 7505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5851968132150898e-05,
+      "loss": 0.6015,
+      "step": 7506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585084136618638e-05,
+      "loss": 0.5597,
+      "step": 7507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5849714487263872e-05,
+      "loss": 0.5525,
+      "step": 7508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5848587495405136e-05,
+      "loss": 0.575,
+      "step": 7509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5847460390631925e-05,
+      "loss": 0.572,
+      "step": 7510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5846333172965996e-05,
+      "loss": 0.5621,
+      "step": 7511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5845205842429122e-05,
+      "loss": 0.5648,
+      "step": 7512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5844078399043065e-05,
+      "loss": 0.5904,
+      "step": 7513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5842950842829584e-05,
+      "loss": 0.5699,
+      "step": 7514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841823173810457e-05,
+      "loss": 0.5281,
+      "step": 7515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584069539200745e-05,
+      "loss": 0.5775,
+      "step": 7516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5839567497442338e-05,
+      "loss": 0.5581,
+      "step": 7517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838439490136893e-05,
+      "loss": 0.5388,
+      "step": 7518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5837311370112898e-05,
+      "loss": 0.5779,
+      "step": 7519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836183137392128e-05,
+      "loss": 0.5743,
+      "step": 7520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5835054791996366e-05,
+      "loss": 0.5555,
+      "step": 7521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58339263339474e-05,
+      "loss": 0.2443,
+      "step": 7522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5832797763267013e-05,
+      "loss": 0.583,
+      "step": 7523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5831669079976995e-05,
+      "loss": 0.5823,
+      "step": 7524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5830540284099133e-05,
+      "loss": 0.5874,
+      "step": 7525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829411375655227e-05,
+      "loss": 0.5804,
+      "step": 7526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5828282354667064e-05,
+      "loss": 0.6045,
+      "step": 7527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827153221156445e-05,
+      "loss": 0.5859,
+      "step": 7528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826023975145173e-05,
+      "loss": 0.596,
+      "step": 7529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582489461665504e-05,
+      "loss": 0.5597,
+      "step": 7530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582376514570786e-05,
+      "loss": 0.5792,
+      "step": 7531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5822635562325433e-05,
+      "loss": 0.5954,
+      "step": 7532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582150586652957e-05,
+      "loss": 0.5414,
+      "step": 7533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5820376058342077e-05,
+      "loss": 0.6244,
+      "step": 7534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581924613778477e-05,
+      "loss": 0.5893,
+      "step": 7535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818116104879468e-05,
+      "loss": 0.5639,
+      "step": 7536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816985959647973e-05,
+      "loss": 0.566,
+      "step": 7537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581585570211212e-05,
+      "loss": 0.5693,
+      "step": 7538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581472533229372e-05,
+      "loss": 0.5722,
+      "step": 7539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58135948502146e-05,
+      "loss": 0.518,
+      "step": 7540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5812464255896585e-05,
+      "loss": 0.5984,
+      "step": 7541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5811333549361503e-05,
+      "loss": 0.5439,
+      "step": 7542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5810202730631185e-05,
+      "loss": 0.5962,
+      "step": 7543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580907179972746e-05,
+      "loss": 0.5832,
+      "step": 7544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580794075667216e-05,
+      "loss": 0.5876,
+      "step": 7545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580680960148713e-05,
+      "loss": 0.5979,
+      "step": 7546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58056783341942e-05,
+      "loss": 0.5671,
+      "step": 7547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804546954815213e-05,
+      "loss": 0.5912,
+      "step": 7548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5803415463372013e-05,
+      "loss": 0.5435,
+      "step": 7549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580228385988644e-05,
+      "loss": 0.5991,
+      "step": 7550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5801152144380353e-05,
+      "loss": 0.5913,
+      "step": 7551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5800020316875587e-05,
+      "loss": 0.5589,
+      "step": 7552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5798888377394004e-05,
+      "loss": 0.5714,
+      "step": 7553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579775632595745e-05,
+      "loss": 0.602,
+      "step": 7554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5796624162587787e-05,
+      "loss": 0.5573,
+      "step": 7555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5795491887306867e-05,
+      "loss": 0.6019,
+      "step": 7556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794359500136557e-05,
+      "loss": 0.5294,
+      "step": 7557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579322700109871e-05,
+      "loss": 0.5639,
+      "step": 7558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5792094390215198e-05,
+      "loss": 0.5706,
+      "step": 7559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790961667507884e-05,
+      "loss": 0.548,
+      "step": 7560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578982883299864e-05,
+      "loss": 0.5971,
+      "step": 7561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5788695886709332e-05,
+      "loss": 0.583,
+      "step": 7562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5787562828661833e-05,
+      "loss": 0.5738,
+      "step": 7563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5786429658878023e-05,
+      "loss": 0.6104,
+      "step": 7564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5785296377379777e-05,
+      "loss": 0.5594,
+      "step": 7565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5784162984188974e-05,
+      "loss": 0.5563,
+      "step": 7566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5783029479327495e-05,
+      "loss": 0.5949,
+      "step": 7567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578189586281723e-05,
+      "loss": 0.598,
+      "step": 7568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5780762134680054e-05,
+      "loss": 0.5543,
+      "step": 7569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5779628294937862e-05,
+      "loss": 0.566,
+      "step": 7570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5778494343612544e-05,
+      "loss": 0.5717,
+      "step": 7571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5777360280725987e-05,
+      "loss": 0.5297,
+      "step": 7572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577622610630009e-05,
+      "loss": 0.603,
+      "step": 7573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5775091820356755e-05,
+      "loss": 0.6057,
+      "step": 7574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5773957422917873e-05,
+      "loss": 0.6179,
+      "step": 7575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5772822914005345e-05,
+      "loss": 0.5935,
+      "step": 7576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577168829364108e-05,
+      "loss": 0.5466,
+      "step": 7577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5770553561846975e-05,
+      "loss": 0.6081,
+      "step": 7578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769418718644946e-05,
+      "loss": 0.5809,
+      "step": 7579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5768283764056898e-05,
+      "loss": 0.5843,
+      "step": 7580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767148698104743e-05,
+      "loss": 0.5648,
+      "step": 7581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5766013520810398e-05,
+      "loss": 0.5809,
+      "step": 7582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576487823219577e-05,
+      "loss": 0.5682,
+      "step": 7583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576374283228279e-05,
+      "loss": 0.6045,
+      "step": 7584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5762607321093368e-05,
+      "loss": 0.5533,
+      "step": 7585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5761471698649433e-05,
+      "loss": 0.5555,
+      "step": 7586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5760335964972905e-05,
+      "loss": 0.5649,
+      "step": 7587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5759200120085717e-05,
+      "loss": 0.5611,
+      "step": 7588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575806416400979e-05,
+      "loss": 0.5747,
+      "step": 7589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575692809676706e-05,
+      "loss": 0.5739,
+      "step": 7590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575579191837946e-05,
+      "loss": 0.5806,
+      "step": 7591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5754655628868926e-05,
+      "loss": 0.5811,
+      "step": 7592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575351922825739e-05,
+      "loss": 0.5795,
+      "step": 7593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57523827165668e-05,
+      "loss": 0.5725,
+      "step": 7594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575124609381909e-05,
+      "loss": 0.5433,
+      "step": 7595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575010936003621e-05,
+      "loss": 0.5713,
+      "step": 7596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57489725152401e-05,
+      "loss": 0.5456,
+      "step": 7597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5747835559452716e-05,
+      "loss": 0.5957,
+      "step": 7598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5746698492696007e-05,
+      "loss": 0.5972,
+      "step": 7599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574556131499192e-05,
+      "loss": 0.545,
+      "step": 7600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5744424026362413e-05,
+      "loss": 0.5666,
+      "step": 7601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5743286626829437e-05,
+      "loss": 0.5619,
+      "step": 7602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5742149116414962e-05,
+      "loss": 0.5811,
+      "step": 7603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574101149514094e-05,
+      "loss": 0.5601,
+      "step": 7604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573987376302934e-05,
+      "loss": 0.584,
+      "step": 7605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5738735920102124e-05,
+      "loss": 0.5525,
+      "step": 7606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573759796638126e-05,
+      "loss": 0.5869,
+      "step": 7607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736459901888717e-05,
+      "loss": 0.5909,
+      "step": 7608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5735321726646474e-05,
+      "loss": 0.5916,
+      "step": 7609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5734183440676493e-05,
+      "loss": 0.5825,
+      "step": 7610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573304504400076e-05,
+      "loss": 0.5877,
+      "step": 7611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731906536641245e-05,
+      "loss": 0.5545,
+      "step": 7612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5730767918619935e-05,
+      "loss": 0.5813,
+      "step": 7613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572962918995881e-05,
+      "loss": 0.5878,
+      "step": 7614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5728490350679853e-05,
+      "loss": 0.586,
+      "step": 7615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727351400805054e-05,
+      "loss": 0.5687,
+      "step": 7616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57262123403564e-05,
+      "loss": 0.5813,
+      "step": 7617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572507316935588e-05,
+      "loss": 0.5513,
+      "step": 7618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723933887825492e-05,
+      "loss": 0.5749,
+      "step": 7619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722794495787227e-05,
+      "loss": 0.5793,
+      "step": 7620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5721654993263085e-05,
+      "loss": 0.5679,
+      "step": 7621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5720515380275066e-05,
+      "loss": 0.5764,
+      "step": 7622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571937565684517e-05,
+      "loss": 0.5421,
+      "step": 7623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57182358229954e-05,
+      "loss": 0.5999,
+      "step": 7624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5717095878747764e-05,
+      "loss": 0.5757,
+      "step": 7625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5715955824124267e-05,
+      "loss": 0.5691,
+      "step": 7626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5714815659146923e-05,
+      "loss": 0.5667,
+      "step": 7627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5713675383837744e-05,
+      "loss": 0.5862,
+      "step": 7628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712534998218742e-05,
+      "loss": 0.57,
+      "step": 7629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5711394502311934e-05,
+      "loss": 0.5921,
+      "step": 7630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571025389613934e-05,
+      "loss": 0.5716,
+      "step": 7631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570911317972298e-05,
+      "loss": 0.5806,
+      "step": 7632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570797235308488e-05,
+      "loss": 0.5692,
+      "step": 7633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5706831416247062e-05,
+      "loss": 0.5879,
+      "step": 7634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5705690369231552e-05,
+      "loss": 0.6145,
+      "step": 7635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704549212060383e-05,
+      "loss": 0.5934,
+      "step": 7636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570340794475558e-05,
+      "loss": 0.5821,
+      "step": 7637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570226656733919e-05,
+      "loss": 0.5749,
+      "step": 7638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570112507983323e-05,
+      "loss": 0.5991,
+      "step": 7639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5699983482259756e-05,
+      "loss": 0.5949,
+      "step": 7640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5698841774640797e-05,
+      "loss": 0.5841,
+      "step": 7641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5697699956998398e-05,
+      "loss": 0.5818,
+      "step": 7642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696558029354604e-05,
+      "loss": 0.5471,
+      "step": 7643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569541599173146e-05,
+      "loss": 0.551,
+      "step": 7644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694273844151015e-05,
+      "loss": 0.6124,
+      "step": 7645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569313158663532e-05,
+      "loss": 0.597,
+      "step": 7646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691989219206426e-05,
+      "loss": 0.5733,
+      "step": 7647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5690846741886395e-05,
+      "loss": 0.5599,
+      "step": 7648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5689704154697274e-05,
+      "loss": 0.5469,
+      "step": 7649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688561457661125e-05,
+      "loss": 0.2464,
+      "step": 7650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5687418650800012e-05,
+      "loss": 0.5479,
+      "step": 7651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5686275734136e-05,
+      "loss": 0.5565,
+      "step": 7652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568513270769115e-05,
+      "loss": 0.5858,
+      "step": 7653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568398957148753e-05,
+      "loss": 0.5386,
+      "step": 7654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568284632554721e-05,
+      "loss": 0.5566,
+      "step": 7655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5681702969892266e-05,
+      "loss": 0.5701,
+      "step": 7656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5680559504544768e-05,
+      "loss": 0.5795,
+      "step": 7657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567941592952679e-05,
+      "loss": 0.5689,
+      "step": 7658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5678272244860415e-05,
+      "loss": 0.6086,
+      "step": 7659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567712845056772e-05,
+      "loss": 0.5898,
+      "step": 7660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567598454667079e-05,
+      "loss": 0.5677,
+      "step": 7661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567484053319171e-05,
+      "loss": 0.6049,
+      "step": 7662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567369641015256e-05,
+      "loss": 0.548,
+      "step": 7663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567255217757544e-05,
+      "loss": 0.5561,
+      "step": 7664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671407835482426e-05,
+      "loss": 0.5627,
+      "step": 7665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5670263383895627e-05,
+      "loss": 0.5638,
+      "step": 7666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669118822837127e-05,
+      "loss": 0.2479,
+      "step": 7667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5667974152329025e-05,
+      "loss": 0.545,
+      "step": 7668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5666829372393424e-05,
+      "loss": 0.593,
+      "step": 7669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5665684483052425e-05,
+      "loss": 0.6057,
+      "step": 7670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566453948432813e-05,
+      "loss": 0.5731,
+      "step": 7671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663394376242643e-05,
+      "loss": 0.5736,
+      "step": 7672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662249158818072e-05,
+      "loss": 0.604,
+      "step": 7673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566110383207653e-05,
+      "loss": 0.5992,
+      "step": 7674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565995839604013e-05,
+      "loss": 0.5772,
+      "step": 7675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565881285073098e-05,
+      "loss": 0.5967,
+      "step": 7676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56576671961712e-05,
+      "loss": 0.5771,
+      "step": 7677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5656521432382912e-05,
+      "loss": 0.549,
+      "step": 7678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565537555938823e-05,
+      "loss": 0.5916,
+      "step": 7679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565422957720928e-05,
+      "loss": 0.5492,
+      "step": 7680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5653083485868183e-05,
+      "loss": 0.5785,
+      "step": 7681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565193728538707e-05,
+      "loss": 0.5803,
+      "step": 7682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5650790975788066e-05,
+      "loss": 0.5578,
+      "step": 7683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5649644557093306e-05,
+      "loss": 0.6104,
+      "step": 7684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5648498029324923e-05,
+      "loss": 0.5801,
+      "step": 7685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5647351392505047e-05,
+      "loss": 0.5432,
+      "step": 7686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564620464665582e-05,
+      "loss": 0.5977,
+      "step": 7687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645057791799385e-05,
+      "loss": 0.5526,
+      "step": 7688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5643910827957876e-05,
+      "loss": 0.5676,
+      "step": 7689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5642763755153438e-05,
+      "loss": 0.5993,
+      "step": 7690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5641616573408217e-05,
+      "loss": 0.5457,
+      "step": 7691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640469282744364e-05,
+      "loss": 0.5355,
+      "step": 7692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5639321883184027e-05,
+      "loss": 0.5857,
+      "step": 7693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563817437474936e-05,
+      "loss": 0.5593,
+      "step": 7694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563702675746251e-05,
+      "loss": 0.5468,
+      "step": 7695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563587903134564e-05,
+      "loss": 0.6124,
+      "step": 7696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634731196420906e-05,
+      "loss": 0.5614,
+      "step": 7697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5633583252710473e-05,
+      "loss": 0.5864,
+      "step": 7698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5632435200236497e-05,
+      "loss": 0.539,
+      "step": 7699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5631287039021143e-05,
+      "loss": 0.5942,
+      "step": 7700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563013876908658e-05,
+      "loss": 0.5603,
+      "step": 7701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5628990390454973e-05,
+      "loss": 0.5655,
+      "step": 7702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5627841903148502e-05,
+      "loss": 0.5561,
+      "step": 7703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626693307189334e-05,
+      "loss": 0.5769,
+      "step": 7704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5625544602599647e-05,
+      "loss": 0.5692,
+      "step": 7705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624395789401614e-05,
+      "loss": 0.5597,
+      "step": 7706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623246867617415e-05,
+      "loss": 0.5757,
+      "step": 7707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562209783726923e-05,
+      "loss": 0.5712,
+      "step": 7708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562094869837925e-05,
+      "loss": 0.6019,
+      "step": 7709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5619799450969655e-05,
+      "loss": 0.5844,
+      "step": 7710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561865009506263e-05,
+      "loss": 0.591,
+      "step": 7711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561750063068037e-05,
+      "loss": 0.5848,
+      "step": 7712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5616351057845064e-05,
+      "loss": 0.536,
+      "step": 7713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5615201376578914e-05,
+      "loss": 0.5615,
+      "step": 7714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56140515869041e-05,
+      "loss": 0.5634,
+      "step": 7715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5612901688842833e-05,
+      "loss": 0.5724,
+      "step": 7716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561175168241731e-05,
+      "loss": 0.2534,
+      "step": 7717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5610601567649733e-05,
+      "loss": 0.5634,
+      "step": 7718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5609451344562306e-05,
+      "loss": 0.5705,
+      "step": 7719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5608301013177234e-05,
+      "loss": 0.5616,
+      "step": 7720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560715057351673e-05,
+      "loss": 0.5801,
+      "step": 7721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5606000025603e-05,
+      "loss": 0.5871,
+      "step": 7722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560484936945826e-05,
+      "loss": 0.5596,
+      "step": 7723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5603698605104724e-05,
+      "loss": 0.5638,
+      "step": 7724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602547732564607e-05,
+      "loss": 0.6077,
+      "step": 7725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560139675186013e-05,
+      "loss": 0.5921,
+      "step": 7726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560024566301352e-05,
+      "loss": 0.5589,
+      "step": 7727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5599094466046985e-05,
+      "loss": 0.5466,
+      "step": 7728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559794316098277e-05,
+      "loss": 0.5911,
+      "step": 7729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5596791747843083e-05,
+      "loss": 0.5731,
+      "step": 7730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5595640226650167e-05,
+      "loss": 0.5915,
+      "step": 7731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5594488597426248e-05,
+      "loss": 0.5784,
+      "step": 7732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5593336860193563e-05,
+      "loss": 0.5312,
+      "step": 7733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559218501497434e-05,
+      "loss": 0.5648,
+      "step": 7734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591033061790827e-05,
+      "loss": 0.5736,
+      "step": 7735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558988100066526e-05,
+      "loss": 0.5994,
+      "step": 7736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5588728831619877e-05,
+      "loss": 0.5981,
+      "step": 7737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5587576554676927e-05,
+      "loss": 0.5504,
+      "step": 7738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586424169858657e-05,
+      "loss": 0.5777,
+      "step": 7739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558527167718731e-05,
+      "loss": 0.5984,
+      "step": 7740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558411907668514e-05,
+      "loss": 0.5388,
+      "step": 7741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55829663683744e-05,
+      "loss": 0.5744,
+      "step": 7742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5581813552277343e-05,
+      "loss": 0.5712,
+      "step": 7743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580660628416226e-05,
+      "loss": 0.5707,
+      "step": 7744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579507596813304e-05,
+      "loss": 0.5776,
+      "step": 7745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5578354457490844e-05,
+      "loss": 0.5629,
+      "step": 7746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5577201210471106e-05,
+      "loss": 0.5607,
+      "step": 7747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5576047855776357e-05,
+      "loss": 0.5857,
+      "step": 7748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5574894393428856e-05,
+      "loss": 0.5802,
+      "step": 7749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5573740823450885e-05,
+      "loss": 0.5399,
+      "step": 7750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5572587145864703e-05,
+      "loss": 0.5671,
+      "step": 7751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557143336069259e-05,
+      "loss": 0.5564,
+      "step": 7752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557027946795682e-05,
+      "loss": 0.5565,
+      "step": 7753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569125467679672e-05,
+      "loss": 0.6008,
+      "step": 7754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556797135988342e-05,
+      "loss": 0.5897,
+      "step": 7755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5566817144590348e-05,
+      "loss": 0.5711,
+      "step": 7756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565662821822748e-05,
+      "loss": 0.5958,
+      "step": 7757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5564508391602893e-05,
+      "loss": 0.5917,
+      "step": 7758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563353853953077e-05,
+      "loss": 0.5715,
+      "step": 7759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5562199208895586e-05,
+      "loss": 0.5933,
+      "step": 7760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556104445645272e-05,
+      "loss": 0.5328,
+      "step": 7761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559889596646765e-05,
+      "loss": 0.6052,
+      "step": 7762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555873462950002e-05,
+      "loss": 0.5387,
+      "step": 7763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557579555034782e-05,
+      "loss": 0.5939,
+      "step": 7764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556424373273357e-05,
+      "loss": 0.2614,
+      "step": 7765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555526908423804e-05,
+      "loss": 0.5813,
+      "step": 7766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555411368795113e-05,
+      "loss": 0.5424,
+      "step": 7767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555295818443495e-05,
+      "loss": 0.5927,
+      "step": 7768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55518025737118e-05,
+      "loss": 0.6044,
+      "step": 7769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5550646855803986e-05,
+      "loss": 0.5648,
+      "step": 7770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5549491030733828e-05,
+      "loss": 0.4982,
+      "step": 7771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5548335098523634e-05,
+      "loss": 0.5043,
+      "step": 7772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547179059195726e-05,
+      "loss": 0.4946,
+      "step": 7773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554602291277242e-05,
+      "loss": 0.5248,
+      "step": 7774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554486665927604e-05,
+      "loss": 0.495,
+      "step": 7775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5543710298728906e-05,
+      "loss": 0.4941,
+      "step": 7776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5542553831153347e-05,
+      "loss": 0.5084,
+      "step": 7777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5541397256571682e-05,
+      "loss": 0.54,
+      "step": 7778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554024057500625e-05,
+      "loss": 0.4942,
+      "step": 7779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553908378647937e-05,
+      "loss": 0.5007,
+      "step": 7780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553792689101339e-05,
+      "loss": 0.5122,
+      "step": 7781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536769888630636e-05,
+      "loss": 0.4943,
+      "step": 7782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5535612779353446e-05,
+      "loss": 0.1644,
+      "step": 7783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534455563204162e-05,
+      "loss": 0.4896,
+      "step": 7784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533298240205124e-05,
+      "loss": 0.5345,
+      "step": 7785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5532140810378673e-05,
+      "loss": 0.4925,
+      "step": 7786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553098327374716e-05,
+      "loss": 0.4888,
+      "step": 7787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5529825630332934e-05,
+      "loss": 0.47,
+      "step": 7788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5528667880158338e-05,
+      "loss": 0.5133,
+      "step": 7789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552751002324573e-05,
+      "loss": 0.4984,
+      "step": 7790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5526352059617455e-05,
+      "loss": 0.4891,
+      "step": 7791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5525193989295884e-05,
+      "loss": 0.4888,
+      "step": 7792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524035812303358e-05,
+      "loss": 0.4608,
+      "step": 7793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552287752866225e-05,
+      "loss": 0.4529,
+      "step": 7794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521719138394914e-05,
+      "loss": 0.4733,
+      "step": 7795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5520560641523715e-05,
+      "loss": 0.5237,
+      "step": 7796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5519402038071025e-05,
+      "loss": 0.4647,
+      "step": 7797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551824332805921e-05,
+      "loss": 0.4694,
+      "step": 7798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5517084511510638e-05,
+      "loss": 0.503,
+      "step": 7799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5515925588447683e-05,
+      "loss": 0.4983,
+      "step": 7800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551476655889272e-05,
+      "loss": 0.5257,
+      "step": 7801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5513607422868127e-05,
+      "loss": 0.4831,
+      "step": 7802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5512448180396278e-05,
+      "loss": 0.5104,
+      "step": 7803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5511288831499554e-05,
+      "loss": 0.4807,
+      "step": 7804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551012937620034e-05,
+      "loss": 0.472,
+      "step": 7805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5508969814521026e-05,
+      "loss": 0.4984,
+      "step": 7806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550781014648399e-05,
+      "loss": 0.4894,
+      "step": 7807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5506650372111628e-05,
+      "loss": 0.4967,
+      "step": 7808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550549049142632e-05,
+      "loss": 0.4787,
+      "step": 7809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5504330504450474e-05,
+      "loss": 0.5056,
+      "step": 7810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550317041120647e-05,
+      "loss": 0.4949,
+      "step": 7811
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5502010211716718e-05,
+      "loss": 0.4903,
+      "step": 7812
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550084990600361e-05,
+      "loss": 0.4811,
+      "step": 7813
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5499689494089548e-05,
+      "loss": 0.485,
+      "step": 7814
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5498528975996933e-05,
+      "loss": 0.5203,
+      "step": 7815
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5497368351748176e-05,
+      "loss": 0.4923,
+      "step": 7816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5496207621365686e-05,
+      "loss": 0.4723,
+      "step": 7817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495046784871865e-05,
+      "loss": 0.4885,
+      "step": 7818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5493885842289125e-05,
+      "loss": 0.5057,
+      "step": 7819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549272479363988e-05,
+      "loss": 0.4892,
+      "step": 7820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5491563638946552e-05,
+      "loss": 0.4783,
+      "step": 7821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5490402378231553e-05,
+      "loss": 0.5135,
+      "step": 7822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489241011517303e-05,
+      "loss": 0.5101,
+      "step": 7823
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5488079538826224e-05,
+      "loss": 0.5034,
+      "step": 7824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5486917960180742e-05,
+      "loss": 0.5058,
+      "step": 7825
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5485756275603277e-05,
+      "loss": 0.4954,
+      "step": 7826
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5484594485116265e-05,
+      "loss": 0.4904,
+      "step": 7827
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5483432588742128e-05,
+      "loss": 0.5004,
+      "step": 7828
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5482270586503303e-05,
+      "loss": 0.4982,
+      "step": 7829
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548110847842222e-05,
+      "loss": 0.4893,
+      "step": 7830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479946264521318e-05,
+      "loss": 0.4795,
+      "step": 7831
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547878394482304e-05,
+      "loss": 0.5034,
+      "step": 7832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547762151934981e-05,
+      "loss": 0.5361,
+      "step": 7833
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5476458988124088e-05,
+      "loss": 0.5033,
+      "step": 7834
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547529635116831e-05,
+      "loss": 0.4943,
+      "step": 7835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547413360850492e-05,
+      "loss": 0.484,
+      "step": 7836
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5472970760156373e-05,
+      "loss": 0.4804,
+      "step": 7837
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547180780614511e-05,
+      "loss": 0.5102,
+      "step": 7838
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5470644746493593e-05,
+      "loss": 0.4704,
+      "step": 7839
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.4475,
+      "step": 7840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54683183103596e-05,
+      "loss": 0.4763,
+      "step": 7841
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5467154933922042e-05,
+      "loss": 0.4962,
+      "step": 7842
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5465991451934052e-05,
+      "loss": 0.5147,
+      "step": 7843
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54648278644181e-05,
+      "loss": 0.512,
+      "step": 7844
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463664171396643e-05,
+      "loss": 0.4694,
+      "step": 7845
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5462500372892153e-05,
+      "loss": 0.499,
+      "step": 7846
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5461336468927095e-05,
+      "loss": 0.4751,
+      "step": 7847
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460172459523944e-05,
+      "loss": 0.5081,
+      "step": 7848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5459008344705165e-05,
+      "loss": 0.5144,
+      "step": 7849
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457844124493243e-05,
+      "loss": 0.5047,
+      "step": 7850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5456679798910644e-05,
+      "loss": 0.518,
+      "step": 7851
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5455515367979856e-05,
+      "loss": 0.4891,
+      "step": 7852
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454350831723354e-05,
+      "loss": 0.5315,
+      "step": 7853
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453186190163628e-05,
+      "loss": 0.4634,
+      "step": 7854
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452021443323155e-05,
+      "loss": 0.4919,
+      "step": 7855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5450856591224417e-05,
+      "loss": 0.4765,
+      "step": 7856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5449691633889924e-05,
+      "loss": 0.5114,
+      "step": 7857
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5448526571342145e-05,
+      "loss": 0.5138,
+      "step": 7858
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5447361403603583e-05,
+      "loss": 0.4586,
+      "step": 7859
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544619613069673e-05,
+      "loss": 0.4999,
+      "step": 7860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544503075264409e-05,
+      "loss": 0.4808,
+      "step": 7861
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5443865269468154e-05,
+      "loss": 0.508,
+      "step": 7862
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5442699681191426e-05,
+      "loss": 0.5375,
+      "step": 7863
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544153398783641e-05,
+      "loss": 0.4852,
+      "step": 7864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544036818942561e-05,
+      "loss": 0.495,
+      "step": 7865
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5439202285981533e-05,
+      "loss": 0.4802,
+      "step": 7866
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543803627752669e-05,
+      "loss": 0.4968,
+      "step": 7867
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543687016408359e-05,
+      "loss": 0.5198,
+      "step": 7868
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543570394567475e-05,
+      "loss": 0.4819,
+      "step": 7869
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543453762232268e-05,
+      "loss": 0.4863,
+      "step": 7870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5433371194049903e-05,
+      "loss": 0.4955,
+      "step": 7871
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5432204660878936e-05,
+      "loss": 0.494,
+      "step": 7872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5431038022832297e-05,
+      "loss": 0.5166,
+      "step": 7873
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5429871279932514e-05,
+      "loss": 0.4745,
+      "step": 7874
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542870443220211e-05,
+      "loss": 0.4834,
+      "step": 7875
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5427537479663617e-05,
+      "loss": 0.5181,
+      "step": 7876
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5426370422339555e-05,
+      "loss": 0.5095,
+      "step": 7877
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542520326025247e-05,
+      "loss": 0.5085,
+      "step": 7878
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5424035993424878e-05,
+      "loss": 0.5182,
+      "step": 7879
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542286862187933e-05,
+      "loss": 0.5333,
+      "step": 7880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5421701145638355e-05,
+      "loss": 0.5164,
+      "step": 7881
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5420533564724495e-05,
+      "loss": 0.4754,
+      "step": 7882
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541936587916029e-05,
+      "loss": 0.4765,
+      "step": 7883
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541819808896829e-05,
+      "loss": 0.4982,
+      "step": 7884
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541703019417103e-05,
+      "loss": 0.511,
+      "step": 7885
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5415862194791068e-05,
+      "loss": 0.4996,
+      "step": 7886
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5414694090850945e-05,
+      "loss": 0.5252,
+      "step": 7887
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5413525882373224e-05,
+      "loss": 0.5032,
+      "step": 7888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5412357569380447e-05,
+      "loss": 0.5196,
+      "step": 7889
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5411189151895175e-05,
+      "loss": 0.4904,
+      "step": 7890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5410020629939966e-05,
+      "loss": 0.4808,
+      "step": 7891
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540885200353738e-05,
+      "loss": 0.5045,
+      "step": 7892
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5407683272709974e-05,
+      "loss": 0.5443,
+      "step": 7893
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540651443748032e-05,
+      "loss": 0.5032,
+      "step": 7894
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5405345497870977e-05,
+      "loss": 0.468,
+      "step": 7895
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5404176453904518e-05,
+      "loss": 0.4981,
+      "step": 7896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5403007305603508e-05,
+      "loss": 0.4883,
+      "step": 7897
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5401838052990523e-05,
+      "loss": 0.4848,
+      "step": 7898
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400668696088138e-05,
+      "loss": 0.4715,
+      "step": 7899
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5399499234918924e-05,
+      "loss": 0.4934,
+      "step": 7900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539832966950546e-05,
+      "loss": 0.4918,
+      "step": 7901
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539715999987033e-05,
+      "loss": 0.4587,
+      "step": 7902
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539599022603611e-05,
+      "loss": 0.4722,
+      "step": 7903
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539482034802539e-05,
+      "loss": 0.4721,
+      "step": 7904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393650365860748e-05,
+      "loss": 0.5127,
+      "step": 7905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392480279564784e-05,
+      "loss": 0.4718,
+      "step": 7906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5391310089160077e-05,
+      "loss": 0.4779,
+      "step": 7907
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5390139794669225e-05,
+      "loss": 0.4806,
+      "step": 7908
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538896939611482e-05,
+      "loss": 0.471,
+      "step": 7909
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387798893519454e-05,
+      "loss": 0.4892,
+      "step": 7910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386628286905735e-05,
+      "loss": 0.489,
+      "step": 7911
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5385457576296253e-05,
+      "loss": 0.4805,
+      "step": 7912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538428676171362e-05,
+      "loss": 0.4842,
+      "step": 7913
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5383115843180427e-05,
+      "loss": 0.4865,
+      "step": 7914
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5381944820719296e-05,
+      "loss": 0.5124,
+      "step": 7915
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538077369435282e-05,
+      "loss": 0.5143,
+      "step": 7916
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537960246410362e-05,
+      "loss": 0.4935,
+      "step": 7917
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378431129994302e-05,
+      "loss": 0.5167,
+      "step": 7918
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537725969204748e-05,
+      "loss": 0.5124,
+      "step": 7919
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5376088150285777e-05,
+      "loss": 0.4888,
+      "step": 7920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53749165047318e-05,
+      "loss": 0.1725,
+      "step": 7921
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5373744755408182e-05,
+      "loss": 0.5336,
+      "step": 7922
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5372572902337533e-05,
+      "loss": 0.5085,
+      "step": 7923
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371400945542486e-05,
+      "loss": 0.5149,
+      "step": 7924
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5370228885045662e-05,
+      "loss": 0.4957,
+      "step": 7925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536905672086969e-05,
+      "loss": 0.4843,
+      "step": 7926
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5367884453037203e-05,
+      "loss": 0.5163,
+      "step": 7927
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536671208157083e-05,
+      "loss": 0.5001,
+      "step": 7928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365539606493206e-05,
+      "loss": 0.4777,
+      "step": 7929
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364367027826968e-05,
+      "loss": 0.5064,
+      "step": 7930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363194345594754e-05,
+      "loss": 0.4906,
+      "step": 7931
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5362021559819203e-05,
+      "loss": 0.4973,
+      "step": 7932
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5360848670522955e-05,
+      "loss": 0.4896,
+      "step": 7933
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5359675677728665e-05,
+      "loss": 0.5035,
+      "step": 7934
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358502581458964e-05,
+      "loss": 0.5114,
+      "step": 7935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357329381736508e-05,
+      "loss": 0.4858,
+      "step": 7936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356156078583946e-05,
+      "loss": 0.4854,
+      "step": 7937
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535498267202393e-05,
+      "loss": 0.5297,
+      "step": 7938
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5353809162079117e-05,
+      "loss": 0.5257,
+      "step": 7939
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535263554877216e-05,
+      "loss": 0.4981,
+      "step": 7940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535146183212572e-05,
+      "loss": 0.4862,
+      "step": 7941
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535028801216245e-05,
+      "loss": 0.5103,
+      "step": 7942
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534911408890502e-05,
+      "loss": 0.492,
+      "step": 7943
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534794006237609e-05,
+      "loss": 0.5028,
+      "step": 7944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346765932598326e-05,
+      "loss": 0.4804,
+      "step": 7945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5345591699594397e-05,
+      "loss": 0.5097,
+      "step": 7946
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5344417363386975e-05,
+      "loss": 0.5109,
+      "step": 7947
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5343242923998732e-05,
+      "loss": 0.4844,
+      "step": 7948
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534206838145234e-05,
+      "loss": 0.4811,
+      "step": 7949
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534089373577047e-05,
+      "loss": 0.4807,
+      "step": 7950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5339718986975808e-05,
+      "loss": 0.4816,
+      "step": 7951
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5338544135091034e-05,
+      "loss": 0.4718,
+      "step": 7952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337369180138827e-05,
+      "loss": 0.4939,
+      "step": 7953
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5336194122141867e-05,
+      "loss": 0.5261,
+      "step": 7954
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533501896112285e-05,
+      "loss": 0.4719,
+      "step": 7955
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533384369710446e-05,
+      "loss": 0.492,
+      "step": 7956
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533266833010938e-05,
+      "loss": 0.4736,
+      "step": 7957
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533149286016031e-05,
+      "loss": 0.5091,
+      "step": 7958
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533031728727994e-05,
+      "loss": 0.4891,
+      "step": 7959
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329141611490968e-05,
+      "loss": 0.4986,
+      "step": 7960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5327965832816093e-05,
+      "loss": 0.5193,
+      "step": 7961
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532678995127801e-05,
+      "loss": 0.5067,
+      "step": 7962
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5325613966899426e-05,
+      "loss": 0.5246,
+      "step": 7963
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532443787970304e-05,
+      "loss": 0.5258,
+      "step": 7964
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5323261689711566e-05,
+      "loss": 0.486,
+      "step": 7965
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5322085396947706e-05,
+      "loss": 0.4872,
+      "step": 7966
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5320909001434165e-05,
+      "loss": 0.5178,
+      "step": 7967
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5319732503193664e-05,
+      "loss": 0.5235,
+      "step": 7968
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5318555902248914e-05,
+      "loss": 0.5028,
+      "step": 7969
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5317379198622632e-05,
+      "loss": 0.4916,
+      "step": 7970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531620239233753e-05,
+      "loss": 0.5143,
+      "step": 7971
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315025483416333e-05,
+      "loss": 0.502,
+      "step": 7972
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313848471881757e-05,
+      "loss": 0.5051,
+      "step": 7973
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5312671357756536e-05,
+      "loss": 0.4373,
+      "step": 7974
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5311494141063387e-05,
+      "loss": 0.48,
+      "step": 7975
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531031682182504e-05,
+      "loss": 0.4975,
+      "step": 7976
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5309139400064224e-05,
+      "loss": 0.5011,
+      "step": 7977
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5307961875803675e-05,
+      "loss": 0.4963,
+      "step": 7978
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530678424906612e-05,
+      "loss": 0.1555,
+      "step": 7979
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53056065198743e-05,
+      "loss": 0.5027,
+      "step": 7980
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5304428688250948e-05,
+      "loss": 0.479,
+      "step": 7981
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303250754218807e-05,
+      "loss": 0.5043,
+      "step": 7982
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530207271780062e-05,
+      "loss": 0.4965,
+      "step": 7983
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5300894579019124e-05,
+      "loss": 0.4996,
+      "step": 7984
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529971633789707e-05,
+      "loss": 0.4835,
+      "step": 7985
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5298537994457206e-05,
+      "loss": 0.1698,
+      "step": 7986
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5297359548722277e-05,
+      "loss": 0.4802,
+      "step": 7987
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296181000715037e-05,
+      "loss": 0.5055,
+      "step": 7988
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529500235045824e-05,
+      "loss": 0.4916,
+      "step": 7989
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5293823597974638e-05,
+      "loss": 0.5047,
+      "step": 7990
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529264474328699e-05,
+      "loss": 0.5041,
+      "step": 7991
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529146578641806e-05,
+      "loss": 0.4989,
+      "step": 7992
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290286727390604e-05,
+      "loss": 0.4644,
+      "step": 7993
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5289107566227385e-05,
+      "loss": 0.5036,
+      "step": 7994
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528792830295117e-05,
+      "loss": 0.4918,
+      "step": 7995
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528674893758473e-05,
+      "loss": 0.5043,
+      "step": 7996
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5285569470150825e-05,
+      "loss": 0.4948,
+      "step": 7997
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528438990067223e-05,
+      "loss": 0.4994,
+      "step": 7998
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5283210229171727e-05,
+      "loss": 0.4703,
+      "step": 7999
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5282030455672076e-05,
+      "loss": 0.4895,
+      "step": 8000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5280850580196066e-05,
+      "loss": 0.4802,
+      "step": 8001
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279670602766467e-05,
+      "loss": 0.5276,
+      "step": 8002
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527849052340607e-05,
+      "loss": 0.5084,
+      "step": 8003
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527731034213765e-05,
+      "loss": 0.4799,
+      "step": 8004
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5276130058983993e-05,
+      "loss": 0.4826,
+      "step": 8005
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527494967396789e-05,
+      "loss": 0.4881,
+      "step": 8006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273769187112124e-05,
+      "loss": 0.5172,
+      "step": 8007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527258859843949e-05,
+      "loss": 0.5209,
+      "step": 8008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527140790797278e-05,
+      "loss": 0.4872,
+      "step": 8009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527022711573479e-05,
+      "loss": 0.4935,
+      "step": 8010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5269046221748314e-05,
+      "loss": 0.5049,
+      "step": 8011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526786522603615e-05,
+      "loss": 0.4983,
+      "step": 8012
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52666841286211e-05,
+      "loss": 0.4645,
+      "step": 8013
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526550292952597e-05,
+      "loss": 0.4953,
+      "step": 8014
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526432162877356e-05,
+      "loss": 0.511,
+      "step": 8015
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5263140226386678e-05,
+      "loss": 0.4966,
+      "step": 8016
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526195872238813e-05,
+      "loss": 0.4747,
+      "step": 8017
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5260777116800732e-05,
+      "loss": 0.5151,
+      "step": 8018
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5259595409647293e-05,
+      "loss": 0.4662,
+      "step": 8019
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525841360095063e-05,
+      "loss": 0.5259,
+      "step": 8020
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257231690733552e-05,
+      "loss": 0.4991,
+      "step": 8021
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256049679018887e-05,
+      "loss": 0.5089,
+      "step": 8022
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254867565829447e-05,
+      "loss": 0.4744,
+      "step": 8023
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253685351188061e-05,
+      "loss": 0.4992,
+      "step": 8024
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252503035117552e-05,
+      "loss": 0.5198,
+      "step": 8025
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5251320617640739e-05,
+      "loss": 0.4927,
+      "step": 8026
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250138098780456e-05,
+      "loss": 0.537,
+      "step": 8027
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5248955478559535e-05,
+      "loss": 0.5055,
+      "step": 8028
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5247772757000806e-05,
+      "loss": 0.5132,
+      "step": 8029
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246589934127096e-05,
+      "loss": 0.4996,
+      "step": 8030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245407009961252e-05,
+      "loss": 0.4916,
+      "step": 8031
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5244223984526105e-05,
+      "loss": 0.4836,
+      "step": 8032
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243040857844498e-05,
+      "loss": 0.5195,
+      "step": 8033
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5241857629939272e-05,
+      "loss": 0.4918,
+      "step": 8034
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5240674300833269e-05,
+      "loss": 0.5141,
+      "step": 8035
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5239490870549336e-05,
+      "loss": 0.5244,
+      "step": 8036
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5238307339110323e-05,
+      "loss": 0.4794,
+      "step": 8037
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237123706539077e-05,
+      "loss": 0.4992,
+      "step": 8038
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5235939972858447e-05,
+      "loss": 0.5121,
+      "step": 8039
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234756138091288e-05,
+      "loss": 0.501,
+      "step": 8040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5233572202260463e-05,
+      "loss": 0.5042,
+      "step": 8041
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5232388165388821e-05,
+      "loss": 0.5059,
+      "step": 8042
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.523120402749922e-05,
+      "loss": 0.494,
+      "step": 8043
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5230019788614527e-05,
+      "loss": 0.4789,
+      "step": 8044
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228835448757605e-05,
+      "loss": 0.4857,
+      "step": 8045
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5227651007951313e-05,
+      "loss": 0.4867,
+      "step": 8046
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5226466466218526e-05,
+      "loss": 0.486,
+      "step": 8047
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225281823582108e-05,
+      "loss": 0.4804,
+      "step": 8048
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5224097080064933e-05,
+      "loss": 0.4771,
+      "step": 8049
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222912235689868e-05,
+      "loss": 0.4936,
+      "step": 8050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221727290479796e-05,
+      "loss": 0.4862,
+      "step": 8051
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522054224445759e-05,
+      "loss": 0.4879,
+      "step": 8052
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521935709764613e-05,
+      "loss": 0.4976,
+      "step": 8053
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5218171850068293e-05,
+      "loss": 0.5341,
+      "step": 8054
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5216986501746966e-05,
+      "loss": 0.4836,
+      "step": 8055
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5215801052705033e-05,
+      "loss": 0.5332,
+      "step": 8056
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521461550296538e-05,
+      "loss": 0.4887,
+      "step": 8057
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5213429852550894e-05,
+      "loss": 0.5152,
+      "step": 8058
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521224410148447e-05,
+      "loss": 0.4763,
+      "step": 8059
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211058249788996e-05,
+      "loss": 0.473,
+      "step": 8060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5209872297487365e-05,
+      "loss": 0.4613,
+      "step": 8061
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520868624460248e-05,
+      "loss": 0.5181,
+      "step": 8062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5207500091157235e-05,
+      "loss": 0.4735,
+      "step": 8063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5206313837174528e-05,
+      "loss": 0.5084,
+      "step": 8064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205127482677265e-05,
+      "loss": 0.4815,
+      "step": 8065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5203941027688349e-05,
+      "loss": 0.5004,
+      "step": 8066
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202754472230687e-05,
+      "loss": 0.4941,
+      "step": 8067
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5201567816327183e-05,
+      "loss": 0.5231,
+      "step": 8068
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200381060000755e-05,
+      "loss": 0.523,
+      "step": 8069
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199194203274306e-05,
+      "loss": 0.512,
+      "step": 8070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198007246170754e-05,
+      "loss": 0.4745,
+      "step": 8071
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5196820188713011e-05,
+      "loss": 0.5124,
+      "step": 8072
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5195633030924003e-05,
+      "loss": 0.4911,
+      "step": 8073
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519444577282664e-05,
+      "loss": 0.474,
+      "step": 8074
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5193258414443852e-05,
+      "loss": 0.4886,
+      "step": 8075
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192070955798555e-05,
+      "loss": 0.4813,
+      "step": 8076
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.519088339691368e-05,
+      "loss": 0.4801,
+      "step": 8077
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189695737812153e-05,
+      "loss": 0.5023,
+      "step": 8078
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5188507978516902e-05,
+      "loss": 0.4819,
+      "step": 8079
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187320119050854e-05,
+      "loss": 0.5226,
+      "step": 8080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186132159436953e-05,
+      "loss": 0.5062,
+      "step": 8081
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5184944099698124e-05,
+      "loss": 0.4852,
+      "step": 8082
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183755939857311e-05,
+      "loss": 0.4898,
+      "step": 8083
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5182567679937447e-05,
+      "loss": 0.513,
+      "step": 8084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518137931996148e-05,
+      "loss": 0.5047,
+      "step": 8085
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180190859952344e-05,
+      "loss": 0.5148,
+      "step": 8086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5179002299932993e-05,
+      "loss": 0.4991,
+      "step": 8087
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177813639926364e-05,
+      "loss": 0.4752,
+      "step": 8088
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5176624879955412e-05,
+      "loss": 0.5426,
+      "step": 8089
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175436020043085e-05,
+      "loss": 0.5112,
+      "step": 8090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5174247060212341e-05,
+      "loss": 0.501,
+      "step": 8091
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173058000486128e-05,
+      "loss": 0.4819,
+      "step": 8092
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171868840887403e-05,
+      "loss": 0.5101,
+      "step": 8093
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5170679581439127e-05,
+      "loss": 0.4993,
+      "step": 8094
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169490222164255e-05,
+      "loss": 0.4618,
+      "step": 8095
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5168300763085757e-05,
+      "loss": 0.4746,
+      "step": 8096
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5167111204226592e-05,
+      "loss": 0.4991,
+      "step": 8097
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165921545609728e-05,
+      "loss": 0.4782,
+      "step": 8098
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164731787258129e-05,
+      "loss": 0.5001,
+      "step": 8099
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163541929194769e-05,
+      "loss": 0.498,
+      "step": 8100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5162351971442619e-05,
+      "loss": 0.4873,
+      "step": 8101
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161161914024649e-05,
+      "loss": 0.518,
+      "step": 8102
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5159971756963839e-05,
+      "loss": 0.515,
+      "step": 8103
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158781500283165e-05,
+      "loss": 0.5126,
+      "step": 8104
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157591144005609e-05,
+      "loss": 0.4709,
+      "step": 8105
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5156400688154147e-05,
+      "loss": 0.51,
+      "step": 8106
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5155210132751763e-05,
+      "loss": 0.4857,
+      "step": 8107
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515401947782145e-05,
+      "loss": 0.5165,
+      "step": 8108
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5152828723386184e-05,
+      "loss": 0.4839,
+      "step": 8109
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515163786946896e-05,
+      "loss": 0.4907,
+      "step": 8110
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515044691609277e-05,
+      "loss": 0.5596,
+      "step": 8111
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5149255863280607e-05,
+      "loss": 0.5161,
+      "step": 8112
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148064711055463e-05,
+      "loss": 0.4967,
+      "step": 8113
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5146873459440334e-05,
+      "loss": 0.5079,
+      "step": 8114
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5145682108458224e-05,
+      "loss": 0.5083,
+      "step": 8115
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5144490658132126e-05,
+      "loss": 0.4917,
+      "step": 8116
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143299108485048e-05,
+      "loss": 0.489,
+      "step": 8117
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514210745953999e-05,
+      "loss": 0.4878,
+      "step": 8118
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5140915711319965e-05,
+      "loss": 0.5122,
+      "step": 8119
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5139723863847977e-05,
+      "loss": 0.4943,
+      "step": 8120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5138531917147035e-05,
+      "loss": 0.4923,
+      "step": 8121
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5137339871240153e-05,
+      "loss": 0.5227,
+      "step": 8122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136147726150345e-05,
+      "loss": 0.4955,
+      "step": 8123
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5134955481900627e-05,
+      "loss": 0.4538,
+      "step": 8124
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5133763138514015e-05,
+      "loss": 0.5094,
+      "step": 8125
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5132570696013528e-05,
+      "loss": 0.4697,
+      "step": 8126
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5131378154422196e-05,
+      "loss": 0.4918,
+      "step": 8127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130185513763027e-05,
+      "loss": 0.4962,
+      "step": 8128
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5128992774059063e-05,
+      "loss": 0.4963,
+      "step": 8129
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127799935333323e-05,
+      "loss": 0.4732,
+      "step": 8130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5126606997608835e-05,
+      "loss": 0.5014,
+      "step": 8131
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5125413960908632e-05,
+      "loss": 0.5113,
+      "step": 8132
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124220825255748e-05,
+      "loss": 0.5207,
+      "step": 8133
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5123027590673221e-05,
+      "loss": 0.4582,
+      "step": 8134
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512183425718408e-05,
+      "loss": 0.5183,
+      "step": 8135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512064082481137e-05,
+      "loss": 0.4968,
+      "step": 8136
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5119447293578131e-05,
+      "loss": 0.4905,
+      "step": 8137
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118253663507405e-05,
+      "loss": 0.5029,
+      "step": 8138
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5117059934622231e-05,
+      "loss": 0.5078,
+      "step": 8139
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115866106945664e-05,
+      "loss": 0.5003,
+      "step": 8140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511467218050075e-05,
+      "loss": 0.5013,
+      "step": 8141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5113478155310537e-05,
+      "loss": 0.4947,
+      "step": 8142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112284031398079e-05,
+      "loss": 0.5341,
+      "step": 8143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5111089808786435e-05,
+      "loss": 0.5004,
+      "step": 8144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109895487498648e-05,
+      "loss": 0.4667,
+      "step": 8145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108701067557787e-05,
+      "loss": 0.5055,
+      "step": 8146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5107506548986905e-05,
+      "loss": 0.4887,
+      "step": 8147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510631193180907e-05,
+      "loss": 0.4948,
+      "step": 8148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5105117216047344e-05,
+      "loss": 0.493,
+      "step": 8149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103922401724789e-05,
+      "loss": 0.4947,
+      "step": 8150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5102727488864473e-05,
+      "loss": 0.4918,
+      "step": 8151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510153247748947e-05,
+      "loss": 0.4915,
+      "step": 8152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5100337367622846e-05,
+      "loss": 0.1715,
+      "step": 8153
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5099142159287679e-05,
+      "loss": 0.4798,
+      "step": 8154
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097946852507042e-05,
+      "loss": 0.5007,
+      "step": 8155
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5096751447304008e-05,
+      "loss": 0.5175,
+      "step": 8156
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5095555943701658e-05,
+      "loss": 0.4777,
+      "step": 8157
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5094360341723074e-05,
+      "loss": 0.5111,
+      "step": 8158
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093164641391342e-05,
+      "loss": 0.5033,
+      "step": 8159
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509196884272954e-05,
+      "loss": 0.5013,
+      "step": 8160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509077294576076e-05,
+      "loss": 0.4934,
+      "step": 8161
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089576950508082e-05,
+      "loss": 0.4944,
+      "step": 8162
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5088380856994608e-05,
+      "loss": 0.5199,
+      "step": 8163
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5087184665243424e-05,
+      "loss": 0.498,
+      "step": 8164
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508598837527762e-05,
+      "loss": 0.528,
+      "step": 8165
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084791987120298e-05,
+      "loss": 0.5298,
+      "step": 8166
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5083595500794554e-05,
+      "loss": 0.5024,
+      "step": 8167
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082398916323486e-05,
+      "loss": 0.5029,
+      "step": 8168
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.50812022337302e-05,
+      "loss": 0.4871,
+      "step": 8169
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5080005453037791e-05,
+      "loss": 0.5055,
+      "step": 8170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5078808574269377e-05,
+      "loss": 0.5152,
+      "step": 8171
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5077611597448055e-05,
+      "loss": 0.5046,
+      "step": 8172
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5076414522596933e-05,
+      "loss": 0.4971,
+      "step": 8173
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5075217349739136e-05,
+      "loss": 0.4839,
+      "step": 8174
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507402007889776e-05,
+      "loss": 0.5119,
+      "step": 8175
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5072822710095929e-05,
+      "loss": 0.4883,
+      "step": 8176
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5071625243356755e-05,
+      "loss": 0.4973,
+      "step": 8177
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070427678703365e-05,
+      "loss": 0.5226,
+      "step": 8178
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5069230016158871e-05,
+      "loss": 0.5204,
+      "step": 8179
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.50680322557464e-05,
+      "loss": 0.4803,
+      "step": 8180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066834397489075e-05,
+      "loss": 0.539,
+      "step": 8181
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5065636441410023e-05,
+      "loss": 0.445,
+      "step": 8182
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5064438387532368e-05,
+      "loss": 0.4942,
+      "step": 8183
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063240235879244e-05,
+      "loss": 0.4712,
+      "step": 8184
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506204198647378e-05,
+      "loss": 0.5161,
+      "step": 8185
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5060843639339116e-05,
+      "loss": 0.4966,
+      "step": 8186
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505964519449838e-05,
+      "loss": 0.4988,
+      "step": 8187
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5058446651974716e-05,
+      "loss": 0.5075,
+      "step": 8188
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057248011791256e-05,
+      "loss": 0.4672,
+      "step": 8189
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505604927397115e-05,
+      "loss": 0.5078,
+      "step": 8190
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5054850438537531e-05,
+      "loss": 0.4941,
+      "step": 8191
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5053651505513553e-05,
+      "loss": 0.5093,
+      "step": 8192
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5052452474922362e-05,
+      "loss": 0.5064,
+      "step": 8193
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5051253346787102e-05,
+      "loss": 0.5271,
+      "step": 8194
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5050054121130925e-05,
+      "loss": 0.5096,
+      "step": 8195
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5048854797976987e-05,
+      "loss": 0.4759,
+      "step": 8196
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.504765537734844e-05,
+      "loss": 0.1667,
+      "step": 8197
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5046455859268436e-05,
+      "loss": 0.5024,
+      "step": 8198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045256243760144e-05,
+      "loss": 0.5153,
+      "step": 8199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5044056530846715e-05,
+      "loss": 0.4659,
+      "step": 8200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5042856720551315e-05,
+      "loss": 0.468,
+      "step": 8201
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041656812897104e-05,
+      "loss": 0.5169,
+      "step": 8202
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5040456807907253e-05,
+      "loss": 0.4882,
+      "step": 8203
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5039256705604928e-05,
+      "loss": 0.4865,
+      "step": 8204
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038056506013297e-05,
+      "loss": 0.5042,
+      "step": 8205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5036856209155531e-05,
+      "loss": 0.5018,
+      "step": 8206
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035655815054807e-05,
+      "loss": 0.4756,
+      "step": 8207
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034455323734293e-05,
+      "loss": 0.4999,
+      "step": 8208
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5033254735217173e-05,
+      "loss": 0.5494,
+      "step": 8209
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5032054049526623e-05,
+      "loss": 0.4989,
+      "step": 8210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5030853266685825e-05,
+      "loss": 0.5111,
+      "step": 8211
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502965238671796e-05,
+      "loss": 0.5067,
+      "step": 8212
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028451409646214e-05,
+      "loss": 0.5097,
+      "step": 8213
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5027250335493771e-05,
+      "loss": 0.5212,
+      "step": 8214
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026049164283824e-05,
+      "loss": 0.5008,
+      "step": 8215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502484789603956e-05,
+      "loss": 0.4775,
+      "step": 8216
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502364653078417e-05,
+      "loss": 0.4933,
+      "step": 8217
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022445068540847e-05,
+      "loss": 0.4909,
+      "step": 8218
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021243509332792e-05,
+      "loss": 0.521,
+      "step": 8219
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5020041853183198e-05,
+      "loss": 0.5239,
+      "step": 8220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5018840100115266e-05,
+      "loss": 0.4859,
+      "step": 8221
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017638250152198e-05,
+      "loss": 0.4497,
+      "step": 8222
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5016436303317197e-05,
+      "loss": 0.4901,
+      "step": 8223
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015234259633467e-05,
+      "loss": 0.495,
+      "step": 8224
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5014032119124215e-05,
+      "loss": 0.5276,
+      "step": 8225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012829881812651e-05,
+      "loss": 0.4874,
+      "step": 8226
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011627547721988e-05,
+      "loss": 0.4644,
+      "step": 8227
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5010425116875433e-05,
+      "loss": 0.5144,
+      "step": 8228
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5009222589296204e-05,
+      "loss": 0.5055,
+      "step": 8229
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5008019965007518e-05,
+      "loss": 0.4644,
+      "step": 8230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006817244032589e-05,
+      "loss": 0.498,
+      "step": 8231
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005614426394643e-05,
+      "loss": 0.491,
+      "step": 8232
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004411512116899e-05,
+      "loss": 0.5051,
+      "step": 8233
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5003208501222579e-05,
+      "loss": 0.5027,
+      "step": 8234
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5002005393734914e-05,
+      "loss": 0.5136,
+      "step": 8235
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000802189677122e-05,
+      "loss": 0.5125,
+      "step": 8236
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4999598889072445e-05,
+      "loss": 0.5009,
+      "step": 8237
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4998395491944101e-05,
+      "loss": 0.5375,
+      "step": 8238
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4997191998315335e-05,
+      "loss": 0.4884,
+      "step": 8239
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995988408209374e-05,
+      "loss": 0.4915,
+      "step": 8240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499478472164946e-05,
+      "loss": 0.4828,
+      "step": 8241
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993580938658827e-05,
+      "loss": 0.5077,
+      "step": 8242
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4992377059260718e-05,
+      "loss": 0.492,
+      "step": 8243
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4991173083478374e-05,
+      "loss": 0.4947,
+      "step": 8244
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989969011335043e-05,
+      "loss": 0.4996,
+      "step": 8245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988764842853969e-05,
+      "loss": 0.4498,
+      "step": 8246
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.49875605780584e-05,
+      "loss": 0.4854,
+      "step": 8247
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4986356216971583e-05,
+      "loss": 0.4811,
+      "step": 8248
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4985151759616776e-05,
+      "loss": 0.4882,
+      "step": 8249
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983947206017225e-05,
+      "loss": 0.4831,
+      "step": 8250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982742556196192e-05,
+      "loss": 0.5368,
+      "step": 8251
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498153781017693e-05,
+      "loss": 0.4893,
+      "step": 8252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980332967982703e-05,
+      "loss": 0.5222,
+      "step": 8253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4979128029636765e-05,
+      "loss": 0.4774,
+      "step": 8254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4977922995162384e-05,
+      "loss": 0.5106,
+      "step": 8255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4976717864582824e-05,
+      "loss": 0.4956,
+      "step": 8256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975512637921352e-05,
+      "loss": 0.4894,
+      "step": 8257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4974307315201235e-05,
+      "loss": 0.4905,
+      "step": 8258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4973101896445743e-05,
+      "loss": 0.4956,
+      "step": 8259
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4971896381678152e-05,
+      "loss": 0.4994,
+      "step": 8260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4970690770921728e-05,
+      "loss": 0.5186,
+      "step": 8261
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4969485064199756e-05,
+      "loss": 0.4877,
+      "step": 8262
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968279261535508e-05,
+      "loss": 0.5014,
+      "step": 8263
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4967073362952267e-05,
+      "loss": 0.4922,
+      "step": 8264
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965867368473308e-05,
+      "loss": 0.5005,
+      "step": 8265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964661278121924e-05,
+      "loss": 0.5212,
+      "step": 8266
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4963455091921391e-05,
+      "loss": 0.5224,
+      "step": 8267
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962248809895001e-05,
+      "loss": 0.4976,
+      "step": 8268
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961042432066042e-05,
+      "loss": 0.4667,
+      "step": 8269
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959835958457807e-05,
+      "loss": 0.4784,
+      "step": 8270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4958629389093582e-05,
+      "loss": 0.4976,
+      "step": 8271
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957422723996667e-05,
+      "loss": 0.4978,
+      "step": 8272
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4956215963190355e-05,
+      "loss": 0.4783,
+      "step": 8273
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4955009106697948e-05,
+      "loss": 0.4845,
+      "step": 8274
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4953802154542743e-05,
+      "loss": 0.4823,
+      "step": 8275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4952595106748043e-05,
+      "loss": 0.4906,
+      "step": 8276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4951387963337147e-05,
+      "loss": 0.5045,
+      "step": 8277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4950180724333367e-05,
+      "loss": 0.5014,
+      "step": 8278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948973389760008e-05,
+      "loss": 0.373,
+      "step": 8279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4947765959640379e-05,
+      "loss": 0.4709,
+      "step": 8280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946558433997792e-05,
+      "loss": 0.4865,
+      "step": 8281
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4945350812855555e-05,
+      "loss": 0.5003,
+      "step": 8282
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944143096236987e-05,
+      "loss": 0.4832,
+      "step": 8283
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4942935284165403e-05,
+      "loss": 0.5031,
+      "step": 8284
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941727376664125e-05,
+      "loss": 0.4798,
+      "step": 8285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940519373756468e-05,
+      "loss": 0.5201,
+      "step": 8286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4939311275465758e-05,
+      "loss": 0.4729,
+      "step": 8287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938103081815316e-05,
+      "loss": 0.5305,
+      "step": 8288
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936894792828471e-05,
+      "loss": 0.5192,
+      "step": 8289
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935686408528548e-05,
+      "loss": 0.5008,
+      "step": 8290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934477928938873e-05,
+      "loss": 0.5015,
+      "step": 8291
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4933269354082784e-05,
+      "loss": 0.4895,
+      "step": 8292
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493206068398361e-05,
+      "loss": 0.505,
+      "step": 8293
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4930851918664691e-05,
+      "loss": 0.499,
+      "step": 8294
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929643058149357e-05,
+      "loss": 0.467,
+      "step": 8295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492843410246095e-05,
+      "loss": 0.4772,
+      "step": 8296
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4927225051622811e-05,
+      "loss": 0.5103,
+      "step": 8297
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492601590565828e-05,
+      "loss": 0.4741,
+      "step": 8298
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4924806664590702e-05,
+      "loss": 0.5082,
+      "step": 8299
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923597328443423e-05,
+      "loss": 0.4697,
+      "step": 8300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4922387897239793e-05,
+      "loss": 0.4804,
+      "step": 8301
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4921178371003157e-05,
+      "loss": 0.4552,
+      "step": 8302
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491996874975687e-05,
+      "loss": 0.4623,
+      "step": 8303
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4918759033524288e-05,
+      "loss": 0.4767,
+      "step": 8304
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491754922232876e-05,
+      "loss": 0.4863,
+      "step": 8305
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4916339316193645e-05,
+      "loss": 0.4893,
+      "step": 8306
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915129315142301e-05,
+      "loss": 0.4876,
+      "step": 8307
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4913919219198095e-05,
+      "loss": 0.4788,
+      "step": 8308
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491270902838438e-05,
+      "loss": 0.4935,
+      "step": 8309
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4911498742724523e-05,
+      "loss": 0.4704,
+      "step": 8310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4910288362241896e-05,
+      "loss": 0.5089,
+      "step": 8311
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909077886959863e-05,
+      "loss": 0.4998,
+      "step": 8312
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490786731690179e-05,
+      "loss": 0.4686,
+      "step": 8313
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4906656652091054e-05,
+      "loss": 0.5105,
+      "step": 8314
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905445892551027e-05,
+      "loss": 0.4705,
+      "step": 8315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4904235038305084e-05,
+      "loss": 0.9019,
+      "step": 8316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.49030240893766e-05,
+      "loss": 0.4995,
+      "step": 8317
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4901813045788957e-05,
+      "loss": 0.4962,
+      "step": 8318
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4900601907565535e-05,
+      "loss": 0.5124,
+      "step": 8319
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899390674729716e-05,
+      "loss": 0.5272,
+      "step": 8320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4898179347304882e-05,
+      "loss": 0.5132,
+      "step": 8321
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896967925314427e-05,
+      "loss": 0.5055,
+      "step": 8322
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4895756408781733e-05,
+      "loss": 0.5279,
+      "step": 8323
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4894544797730186e-05,
+      "loss": 0.4779,
+      "step": 8324
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893333092183184e-05,
+      "loss": 0.5186,
+      "step": 8325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4892121292164122e-05,
+      "loss": 0.4988,
+      "step": 8326
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890909397696392e-05,
+      "loss": 0.4722,
+      "step": 8327
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4889697408803385e-05,
+      "loss": 0.4832,
+      "step": 8328
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888485325508514e-05,
+      "loss": 0.4931,
+      "step": 8329
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488727314783517e-05,
+      "loss": 0.4992,
+      "step": 8330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886060875806758e-05,
+      "loss": 0.5291,
+      "step": 8331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884848509446678e-05,
+      "loss": 0.5118,
+      "step": 8332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4883636048778347e-05,
+      "loss": 0.5283,
+      "step": 8333
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882423493825165e-05,
+      "loss": 0.504,
+      "step": 8334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4881210844610543e-05,
+      "loss": 0.4876,
+      "step": 8335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4879998101157892e-05,
+      "loss": 0.4952,
+      "step": 8336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487878526349063e-05,
+      "loss": 0.476,
+      "step": 8337
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4877572331632169e-05,
+      "loss": 0.4895,
+      "step": 8338
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876359305605923e-05,
+      "loss": 0.5103,
+      "step": 8339
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487514618543532e-05,
+      "loss": 0.4968,
+      "step": 8340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4873932971143771e-05,
+      "loss": 0.5383,
+      "step": 8341
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872719662754707e-05,
+      "loss": 0.5007,
+      "step": 8342
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4871506260291544e-05,
+      "loss": 0.512,
+      "step": 8343
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4870292763777717e-05,
+      "loss": 0.5165,
+      "step": 8344
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869079173236648e-05,
+      "loss": 0.5505,
+      "step": 8345
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867865488691769e-05,
+      "loss": 0.4687,
+      "step": 8346
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866651710166511e-05,
+      "loss": 0.488,
+      "step": 8347
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4865437837684312e-05,
+      "loss": 0.514,
+      "step": 8348
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.48642238712686e-05,
+      "loss": 0.4951,
+      "step": 8349
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4863009810942814e-05,
+      "loss": 0.5229,
+      "step": 8350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4861795656730397e-05,
+      "loss": 0.4855,
+      "step": 8351
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486058140865479e-05,
+      "loss": 0.5179,
+      "step": 8352
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485936706673943e-05,
+      "loss": 0.5044,
+      "step": 8353
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4858152631007765e-05,
+      "loss": 0.4943,
+      "step": 8354
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485693810148324e-05,
+      "loss": 0.4876,
+      "step": 8355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855723478189308e-05,
+      "loss": 0.5474,
+      "step": 8356
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485450876114941e-05,
+      "loss": 0.4694,
+      "step": 8357
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853293950387004e-05,
+      "loss": 0.515,
+      "step": 8358
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4852079045925542e-05,
+      "loss": 0.4932,
+      "step": 8359
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850864047788481e-05,
+      "loss": 0.486,
+      "step": 8360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849648955999274e-05,
+      "loss": 0.5013,
+      "step": 8361
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4848433770581383e-05,
+      "loss": 0.4866,
+      "step": 8362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847218491558266e-05,
+      "loss": 0.5127,
+      "step": 8363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484600311895339e-05,
+      "loss": 0.4923,
+      "step": 8364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4844787652790215e-05,
+      "loss": 0.4965,
+      "step": 8365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4843572093092209e-05,
+      "loss": 0.4762,
+      "step": 8366
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4842356439882841e-05,
+      "loss": 0.4908,
+      "step": 8367
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484114069318558e-05,
+      "loss": 0.5282,
+      "step": 8368
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839924853023895e-05,
+      "loss": 0.5225,
+      "step": 8369
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838708919421261e-05,
+      "loss": 0.4893,
+      "step": 8370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837492892401155e-05,
+      "loss": 0.4556,
+      "step": 8371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4836276771987053e-05,
+      "loss": 0.4855,
+      "step": 8372
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4835060558202432e-05,
+      "loss": 0.4881,
+      "step": 8373
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833844251070775e-05,
+      "loss": 0.5035,
+      "step": 8374
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832627850615563e-05,
+      "loss": 0.5056,
+      "step": 8375
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4831411356860276e-05,
+      "loss": 0.5013,
+      "step": 8376
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830194769828413e-05,
+      "loss": 0.5259,
+      "step": 8377
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4828978089543447e-05,
+      "loss": 0.4905,
+      "step": 8378
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827761316028877e-05,
+      "loss": 0.4664,
+      "step": 8379
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4826544449308186e-05,
+      "loss": 0.49,
+      "step": 8380
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4825327489404877e-05,
+      "loss": 0.5278,
+      "step": 8381
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4824110436342438e-05,
+      "loss": 0.4909,
+      "step": 8382
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4822893290144369e-05,
+      "loss": 0.477,
+      "step": 8383
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4821676050834166e-05,
+      "loss": 0.5055,
+      "step": 8384
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820458718435333e-05,
+      "loss": 0.4728,
+      "step": 8385
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4819241292971372e-05,
+      "loss": 0.5074,
+      "step": 8386
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481802377446578e-05,
+      "loss": 0.5042,
+      "step": 8387
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481680616294207e-05,
+      "loss": 0.4813,
+      "step": 8388
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4815588458423746e-05,
+      "loss": 0.4786,
+      "step": 8389
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814370660934323e-05,
+      "loss": 0.4964,
+      "step": 8390
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4813152770497303e-05,
+      "loss": 0.5094,
+      "step": 8391
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4811934787136206e-05,
+      "loss": 0.502,
+      "step": 8392
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810716710874545e-05,
+      "loss": 0.4691,
+      "step": 8393
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4809498541735836e-05,
+      "loss": 0.4873,
+      "step": 8394
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808280279743594e-05,
+      "loss": 0.4866,
+      "step": 8395
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4807061924921346e-05,
+      "loss": 0.5011,
+      "step": 8396
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4805843477292609e-05,
+      "loss": 0.4974,
+      "step": 8397
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4804624936880906e-05,
+      "loss": 0.5028,
+      "step": 8398
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4803406303709768e-05,
+      "loss": 0.5,
+      "step": 8399
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480218757780272e-05,
+      "loss": 0.4888,
+      "step": 8400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4800968759183288e-05,
+      "loss": 0.4985,
+      "step": 8401
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4799749847875002e-05,
+      "loss": 0.5108,
+      "step": 8402
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.47985308439014e-05,
+      "loss": 0.4904,
+      "step": 8403
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797311747286014e-05,
+      "loss": 0.4971,
+      "step": 8404
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4796092558052379e-05,
+      "loss": 0.505,
+      "step": 8405
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4794873276224033e-05,
+      "loss": 0.527,
+      "step": 8406
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4793653901824519e-05,
+      "loss": 0.4806,
+      "step": 8407
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792434434877376e-05,
+      "loss": 0.5081,
+      "step": 8408
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791214875406145e-05,
+      "loss": 0.461,
+      "step": 8409
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4789995223434376e-05,
+      "loss": 0.5048,
+      "step": 8410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4788775478985615e-05,
+      "loss": 0.4796,
+      "step": 8411
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478755564208341e-05,
+      "loss": 0.5086,
+      "step": 8412
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786335712751308e-05,
+      "loss": 0.4815,
+      "step": 8413
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785115691012866e-05,
+      "loss": 0.5242,
+      "step": 8414
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783895576891639e-05,
+      "loss": 0.4826,
+      "step": 8415
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4782675370411177e-05,
+      "loss": 0.4773,
+      "step": 8416
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478145507159504e-05,
+      "loss": 0.5259,
+      "step": 8417
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780234680466792e-05,
+      "loss": 0.5327,
+      "step": 8418
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477901419704999e-05,
+      "loss": 0.4794,
+      "step": 8419
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4777793621368197e-05,
+      "loss": 0.4931,
+      "step": 8420
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4776572953444977e-05,
+      "loss": 0.462,
+      "step": 8421
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4775352193303897e-05,
+      "loss": 0.4743,
+      "step": 8422
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774131340968532e-05,
+      "loss": 0.4788,
+      "step": 8423
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477291039646244e-05,
+      "loss": 0.4754,
+      "step": 8424
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771689359809202e-05,
+      "loss": 0.4931,
+      "step": 8425
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477046823103239e-05,
+      "loss": 0.5179,
+      "step": 8426
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4769247010155579e-05,
+      "loss": 0.4878,
+      "step": 8427
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768025697202345e-05,
+      "loss": 0.5012,
+      "step": 8428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4766804292196267e-05,
+      "loss": 0.4923,
+      "step": 8429
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4765582795160926e-05,
+      "loss": 0.4944,
+      "step": 8430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4764361206119907e-05,
+      "loss": 0.5077,
+      "step": 8431
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4763139525096787e-05,
+      "loss": 0.4983,
+      "step": 8432
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4761917752115162e-05,
+      "loss": 0.4955,
+      "step": 8433
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4760695887198617e-05,
+      "loss": 0.4956,
+      "step": 8434
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759473930370738e-05,
+      "loss": 0.4656,
+      "step": 8435
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4758251881655117e-05,
+      "loss": 0.5161,
+      "step": 8436
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475702974107535e-05,
+      "loss": 0.469,
+      "step": 8437
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4755807508655034e-05,
+      "loss": 0.4817,
+      "step": 8438
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4754585184417758e-05,
+      "loss": 0.4823,
+      "step": 8439
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753362768387125e-05,
+      "loss": 0.4987,
+      "step": 8440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4752140260586739e-05,
+      "loss": 0.4762,
+      "step": 8441
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750917661040197e-05,
+      "loss": 0.5105,
+      "step": 8442
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47496949697711e-05,
+      "loss": 0.5007,
+      "step": 8443
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4748472186803062e-05,
+      "loss": 0.4675,
+      "step": 8444
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747249312159685e-05,
+      "loss": 0.5146,
+      "step": 8445
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4746026345864581e-05,
+      "loss": 0.4912,
+      "step": 8446
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744803287941359e-05,
+      "loss": 0.4699,
+      "step": 8447
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743580138413631e-05,
+      "loss": 0.4983,
+      "step": 8448
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4742356897305013e-05,
+      "loss": 0.4997,
+      "step": 8449
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474113356463912e-05,
+      "loss": 0.4798,
+      "step": 8450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739910140439572e-05,
+      "loss": 0.5149,
+      "step": 8451
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 0.5044,
+      "step": 8452
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473746301753399e-05,
+      "loss": 0.7269,
+      "step": 8453
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736239318875197e-05,
+      "loss": 0.4977,
+      "step": 8454
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4735015528777239e-05,
+      "loss": 0.4975,
+      "step": 8455
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4733791647263742e-05,
+      "loss": 0.4919,
+      "step": 8456
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732567674358334e-05,
+      "loss": 0.5193,
+      "step": 8457
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4731343610084643e-05,
+      "loss": 0.5143,
+      "step": 8458
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4730119454466305e-05,
+      "loss": 0.5079,
+      "step": 8459
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4728895207526952e-05,
+      "loss": 0.4966,
+      "step": 8460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727670869290221e-05,
+      "loss": 0.4842,
+      "step": 8461
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4726446439779744e-05,
+      "loss": 0.4847,
+      "step": 8462
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725221919019172e-05,
+      "loss": 0.5025,
+      "step": 8463
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4723997307032134e-05,
+      "loss": 0.5089,
+      "step": 8464
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722772603842277e-05,
+      "loss": 0.5477,
+      "step": 8465
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4721547809473245e-05,
+      "loss": 1.1,
+      "step": 8466
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472032292394869e-05,
+      "loss": 0.4855,
+      "step": 8467
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4719097947292249e-05,
+      "loss": 0.521,
+      "step": 8468
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4717872879527578e-05,
+      "loss": 0.5035,
+      "step": 8469
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471664772067833e-05,
+      "loss": 0.5334,
+      "step": 8470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715422470768157e-05,
+      "loss": 0.518,
+      "step": 8471
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714197129820714e-05,
+      "loss": 0.4916,
+      "step": 8472
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4712971697859653e-05,
+      "loss": 0.4911,
+      "step": 8473
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471174617490864e-05,
+      "loss": 0.4895,
+      "step": 8474
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4710520560991333e-05,
+      "loss": 0.5116,
+      "step": 8475
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4709294856131389e-05,
+      "loss": 0.5078,
+      "step": 8476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4708069060352479e-05,
+      "loss": 1.0748,
+      "step": 8477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706843173678265e-05,
+      "loss": 0.5233,
+      "step": 8478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4705617196132414e-05,
+      "loss": 0.5074,
+      "step": 8479
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4704391127738596e-05,
+      "loss": 0.4986,
+      "step": 8480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470316496852048e-05,
+      "loss": 0.507,
+      "step": 8481
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701938718501743e-05,
+      "loss": 0.5223,
+      "step": 8482
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4700712377706055e-05,
+      "loss": 0.4899,
+      "step": 8483
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4699485946157093e-05,
+      "loss": 0.4702,
+      "step": 8484
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698259423878534e-05,
+      "loss": 0.4998,
+      "step": 8485
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697032810894062e-05,
+      "loss": 0.5163,
+      "step": 8486
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695806107227355e-05,
+      "loss": 0.4812,
+      "step": 8487
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4694579312902095e-05,
+      "loss": 0.4651,
+      "step": 8488
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469335242794197e-05,
+      "loss": 0.5409,
+      "step": 8489
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692125452370664e-05,
+      "loss": 0.5355,
+      "step": 8490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4690898386211867e-05,
+      "loss": 0.5076,
+      "step": 8491
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689671229489264e-05,
+      "loss": 0.4659,
+      "step": 8492
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688443982226557e-05,
+      "loss": 0.5008,
+      "step": 8493
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687216644447431e-05,
+      "loss": 0.5049,
+      "step": 8494
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4685989216175583e-05,
+      "loss": 0.4661,
+      "step": 8495
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468476169743471e-05,
+      "loss": 0.4911,
+      "step": 8496
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4683534088248516e-05,
+      "loss": 0.4841,
+      "step": 8497
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4682306388640695e-05,
+      "loss": 0.5281,
+      "step": 8498
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4681078598634952e-05,
+      "loss": 0.5188,
+      "step": 8499
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467985071825499e-05,
+      "loss": 0.5047,
+      "step": 8500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4678622747524518e-05,
+      "loss": 0.504,
+      "step": 8501
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4677394686467238e-05,
+      "loss": 0.506,
+      "step": 8502
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676166535106866e-05,
+      "loss": 0.4743,
+      "step": 8503
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674938293467107e-05,
+      "loss": 0.4695,
+      "step": 8504
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673709961571678e-05,
+      "loss": 0.509,
+      "step": 8505
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4672481539444288e-05,
+      "loss": 0.4984,
+      "step": 8506
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467125302710866e-05,
+      "loss": 0.4906,
+      "step": 8507
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467002442458851e-05,
+      "loss": 0.5187,
+      "step": 8508
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668795731907555e-05,
+      "loss": 0.9958,
+      "step": 8509
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4667566949089516e-05,
+      "loss": 0.5115,
+      "step": 8510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666338076158124e-05,
+      "loss": 0.493,
+      "step": 8511
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4665109113137096e-05,
+      "loss": 0.4791,
+      "step": 8512
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466388006005016e-05,
+      "loss": 0.4991,
+      "step": 8513
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662650916921045e-05,
+      "loss": 0.5213,
+      "step": 8514
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4661421683773485e-05,
+      "loss": 0.4949,
+      "step": 8515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466019236063121e-05,
+      "loss": 0.4838,
+      "step": 8516
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658962947517946e-05,
+      "loss": 0.5238,
+      "step": 8517
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4657733444457441e-05,
+      "loss": 0.4496,
+      "step": 8518
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656503851473422e-05,
+      "loss": 0.5191,
+      "step": 8519
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655274168589635e-05,
+      "loss": 0.4864,
+      "step": 8520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4654044395829812e-05,
+      "loss": 0.5121,
+      "step": 8521
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652814533217706e-05,
+      "loss": 0.5317,
+      "step": 8522
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4651584580777054e-05,
+      "loss": 0.4799,
+      "step": 8523
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4650354538531603e-05,
+      "loss": 0.5218,
+      "step": 8524
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46491244065051e-05,
+      "loss": 0.5289,
+      "step": 8525
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4647894184721299e-05,
+      "loss": 0.4991,
+      "step": 8526
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646663873203946e-05,
+      "loss": 0.503,
+      "step": 8527
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645433471976793e-05,
+      "loss": 0.4878,
+      "step": 8528
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4644202981063597e-05,
+      "loss": 0.5042,
+      "step": 8529
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642972400488118e-05,
+      "loss": 0.5127,
+      "step": 8530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641741730274105e-05,
+      "loss": 0.4843,
+      "step": 8531
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4640510970445325e-05,
+      "loss": 0.5069,
+      "step": 8532
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639280121025537e-05,
+      "loss": 0.495,
+      "step": 8533
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4638049182038505e-05,
+      "loss": 0.5133,
+      "step": 8534
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463681815350799e-05,
+      "loss": 0.4964,
+      "step": 8535
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4635587035457762e-05,
+      "loss": 0.4843,
+      "step": 8536
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463435582791159e-05,
+      "loss": 0.4825,
+      "step": 8537
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463312453089324e-05,
+      "loss": 0.5402,
+      "step": 8538
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463189314442649e-05,
+      "loss": 0.4738,
+      "step": 8539
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630661668535106e-05,
+      "loss": 0.4746,
+      "step": 8540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4629430103242872e-05,
+      "loss": 0.5147,
+      "step": 8541
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4628198448573558e-05,
+      "loss": 0.481,
+      "step": 8542
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4626966704550944e-05,
+      "loss": 0.4823,
+      "step": 8543
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4625734871198812e-05,
+      "loss": 0.5171,
+      "step": 8544
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4624502948540944e-05,
+      "loss": 0.4864,
+      "step": 8545
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4623270936601122e-05,
+      "loss": 0.4786,
+      "step": 8546
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4622038835403135e-05,
+      "loss": 0.4796,
+      "step": 8547
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4620806644970765e-05,
+      "loss": 0.5379,
+      "step": 8548
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619574365327806e-05,
+      "loss": 0.4781,
+      "step": 8549
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618341996498044e-05,
+      "loss": 0.4939,
+      "step": 8550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617109538505276e-05,
+      "loss": 0.4503,
+      "step": 8551
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615876991373297e-05,
+      "loss": 0.4993,
+      "step": 8552
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46146443551259e-05,
+      "loss": 0.4971,
+      "step": 8553
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461341162978688e-05,
+      "loss": 0.4891,
+      "step": 8554
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4612178815380041e-05,
+      "loss": 0.4859,
+      "step": 8555
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461094591192918e-05,
+      "loss": 0.4726,
+      "step": 8556
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4609712919458106e-05,
+      "loss": 0.4682,
+      "step": 8557
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4608479837990616e-05,
+      "loss": 0.5358,
+      "step": 8558
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4607246667550524e-05,
+      "loss": 0.494,
+      "step": 8559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460601340816163e-05,
+      "loss": 0.4872,
+      "step": 8560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460478005984775e-05,
+      "loss": 0.5131,
+      "step": 8561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603546622632692e-05,
+      "loss": 0.4982,
+      "step": 8562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602313096540268e-05,
+      "loss": 0.472,
+      "step": 8563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4601079481594298e-05,
+      "loss": 0.5324,
+      "step": 8564
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4599845777818594e-05,
+      "loss": 0.4792,
+      "step": 8565
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4598611985236975e-05,
+      "loss": 0.4875,
+      "step": 8566
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4597378103873261e-05,
+      "loss": 0.4827,
+      "step": 8567
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4596144133751276e-05,
+      "loss": 0.5009,
+      "step": 8568
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4594910074894837e-05,
+      "loss": 0.5014,
+      "step": 8569
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459367592732778e-05,
+      "loss": 0.5039,
+      "step": 8570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459244169107392e-05,
+      "loss": 0.4898,
+      "step": 8571
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591207366157094e-05,
+      "loss": 0.5119,
+      "step": 8572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4589972952601126e-05,
+      "loss": 0.487,
+      "step": 8573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4588738450429853e-05,
+      "loss": 0.5129,
+      "step": 8574
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587503859667106e-05,
+      "loss": 0.5059,
+      "step": 8575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586269180336723e-05,
+      "loss": 0.4486,
+      "step": 8576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585034412462536e-05,
+      "loss": 0.4763,
+      "step": 8577
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458379955606839e-05,
+      "loss": 0.4844,
+      "step": 8578
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4582564611178122e-05,
+      "loss": 0.4895,
+      "step": 8579
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581329577815571e-05,
+      "loss": 0.4806,
+      "step": 8580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580094456004588e-05,
+      "loss": 0.4999,
+      "step": 8581
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578859245769013e-05,
+      "loss": 0.4745,
+      "step": 8582
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4577623947132698e-05,
+      "loss": 0.5297,
+      "step": 8583
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4576388560119485e-05,
+      "loss": 0.5319,
+      "step": 8584
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575153084753233e-05,
+      "loss": 0.476,
+      "step": 8585
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.457391752105779e-05,
+      "loss": 0.4876,
+      "step": 8586
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4572681869057008e-05,
+      "loss": 0.495,
+      "step": 8587
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571446128774746e-05,
+      "loss": 0.493,
+      "step": 8588
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4570210300234866e-05,
+      "loss": 0.4834,
+      "step": 8589
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568974383461217e-05,
+      "loss": 0.4659,
+      "step": 8590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4567738378477667e-05,
+      "loss": 0.4985,
+      "step": 8591
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566502285308075e-05,
+      "loss": 0.5071,
+      "step": 8592
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4565266103976313e-05,
+      "loss": 0.5095,
+      "step": 8593
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4564029834506235e-05,
+      "loss": 0.4858,
+      "step": 8594
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456279347692172e-05,
+      "loss": 0.4998,
+      "step": 8595
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4561557031246629e-05,
+      "loss": 0.4624,
+      "step": 8596
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4560320497504839e-05,
+      "loss": 0.502,
+      "step": 8597
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455908387572022e-05,
+      "loss": 0.4892,
+      "step": 8598
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557847165916646e-05,
+      "loss": 0.4742,
+      "step": 8599
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556610368117996e-05,
+      "loss": 0.5213,
+      "step": 8600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4555373482348146e-05,
+      "loss": 0.4786,
+      "step": 8601
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4554136508630974e-05,
+      "loss": 0.5187,
+      "step": 8602
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552899446990365e-05,
+      "loss": 0.459,
+      "step": 8603
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4551662297450201e-05,
+      "loss": 0.4883,
+      "step": 8604
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550425060034367e-05,
+      "loss": 0.4827,
+      "step": 8605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4549187734766747e-05,
+      "loss": 0.4748,
+      "step": 8606
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4547950321671231e-05,
+      "loss": 0.4011,
+      "step": 8607
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546712820771714e-05,
+      "loss": 0.5112,
+      "step": 8608
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4545475232092074e-05,
+      "loss": 0.493,
+      "step": 8609
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544237555656217e-05,
+      "loss": 0.5144,
+      "step": 8610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542999791488033e-05,
+      "loss": 0.4973,
+      "step": 8611
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454176193961142e-05,
+      "loss": 0.4795,
+      "step": 8612
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4540524000050273e-05,
+      "loss": 0.484,
+      "step": 8613
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539285972828494e-05,
+      "loss": 0.5044,
+      "step": 8614
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538047857969988e-05,
+      "loss": 0.4814,
+      "step": 8615
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536809655498655e-05,
+      "loss": 0.4845,
+      "step": 8616
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4535571365438398e-05,
+      "loss": 0.5149,
+      "step": 8617
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4534332987813128e-05,
+      "loss": 0.5021,
+      "step": 8618
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533094522646752e-05,
+      "loss": 0.5149,
+      "step": 8619
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453185596996318e-05,
+      "loss": 0.4783,
+      "step": 8620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4530617329786323e-05,
+      "loss": 0.4972,
+      "step": 8621
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4529378602140096e-05,
+      "loss": 0.4874,
+      "step": 8622
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528139787048414e-05,
+      "loss": 0.4999,
+      "step": 8623
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4526900884535194e-05,
+      "loss": 0.5122,
+      "step": 8624
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525661894624353e-05,
+      "loss": 0.5,
+      "step": 8625
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524422817339809e-05,
+      "loss": 0.4967,
+      "step": 8626
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4523183652705492e-05,
+      "loss": 0.4843,
+      "step": 8627
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452194440074532e-05,
+      "loss": 0.5052,
+      "step": 8628
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520705061483218e-05,
+      "loss": 0.4945,
+      "step": 8629
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519465634943115e-05,
+      "loss": 0.4884,
+      "step": 8630
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451822612114894e-05,
+      "loss": 0.5075,
+      "step": 8631
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4516986520124618e-05,
+      "loss": 0.5116,
+      "step": 8632
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515746831894093e-05,
+      "loss": 0.4956,
+      "step": 8633
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514507056481286e-05,
+      "loss": 0.4674,
+      "step": 8634
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513267193910141e-05,
+      "loss": 0.4917,
+      "step": 8635
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4512027244204588e-05,
+      "loss": 0.4882,
+      "step": 8636
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510787207388575e-05,
+      "loss": 0.4989,
+      "step": 8637
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509547083486033e-05,
+      "loss": 0.4892,
+      "step": 8638
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4508306872520912e-05,
+      "loss": 0.4955,
+      "step": 8639
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507066574517148e-05,
+      "loss": 0.4725,
+      "step": 8640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505826189498692e-05,
+      "loss": 0.5263,
+      "step": 8641
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4504585717489494e-05,
+      "loss": 0.469,
+      "step": 8642
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503345158513494e-05,
+      "loss": 0.2603,
+      "step": 8643
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450210451259465e-05,
+      "loss": 0.4909,
+      "step": 8644
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450086377975691e-05,
+      "loss": 0.486,
+      "step": 8645
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4499622960024233e-05,
+      "loss": 0.5017,
+      "step": 8646
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4498382053420568e-05,
+      "loss": 0.5163,
+      "step": 8647
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4497141059969878e-05,
+      "loss": 0.5068,
+      "step": 8648
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4495899979696117e-05,
+      "loss": 0.4927,
+      "step": 8649
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449465881262325e-05,
+      "loss": 0.4893,
+      "step": 8650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4493417558775237e-05,
+      "loss": 0.523,
+      "step": 8651
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4492176218176044e-05,
+      "loss": 0.4686,
+      "step": 8652
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490934790849633e-05,
+      "loss": 0.4907,
+      "step": 8653
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4489693276819976e-05,
+      "loss": 0.4976,
+      "step": 8654
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488451676111037e-05,
+      "loss": 0.4892,
+      "step": 8655
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4487209988746791e-05,
+      "loss": 0.4873,
+      "step": 8656
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485968214751206e-05,
+      "loss": 0.4888,
+      "step": 8657
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4484726354148261e-05,
+      "loss": 0.4823,
+      "step": 8658
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4483484406961932e-05,
+      "loss": 0.5034,
+      "step": 8659
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4482242373216191e-05,
+      "loss": 0.4891,
+      "step": 8660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448100025293502e-05,
+      "loss": 0.4771,
+      "step": 8661
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479758046142398e-05,
+      "loss": 0.4909,
+      "step": 8662
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4478515752862312e-05,
+      "loss": 0.4532,
+      "step": 8663
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447727337311874e-05,
+      "loss": 0.5021,
+      "step": 8664
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4476030906935672e-05,
+      "loss": 0.5,
+      "step": 8665
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4474788354337095e-05,
+      "loss": 0.518,
+      "step": 8666
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473545715346998e-05,
+      "loss": 0.5094,
+      "step": 8667
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472302989989368e-05,
+      "loss": 0.4864,
+      "step": 8668
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44710601782882e-05,
+      "loss": 0.5082,
+      "step": 8669
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469817280267491e-05,
+      "loss": 0.4712,
+      "step": 8670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4468574295951236e-05,
+      "loss": 0.4812,
+      "step": 8671
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467331225363425e-05,
+      "loss": 0.4818,
+      "step": 8672
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4466088068528068e-05,
+      "loss": 0.4815,
+      "step": 8673
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4464844825469155e-05,
+      "loss": 0.4979,
+      "step": 8674
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4463601496210699e-05,
+      "loss": 0.5254,
+      "step": 8675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4462358080776698e-05,
+      "loss": 0.4888,
+      "step": 8676
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461114579191156e-05,
+      "loss": 0.4909,
+      "step": 8677
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4459870991478082e-05,
+      "loss": 0.4957,
+      "step": 8678
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458627317661492e-05,
+      "loss": 0.5092,
+      "step": 8679
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 0.4784,
+      "step": 8680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4456139711813779e-05,
+      "loss": 0.5113,
+      "step": 8681
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445489577983069e-05,
+      "loss": 0.4963,
+      "step": 8682
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453651761840135e-05,
+      "loss": 0.5061,
+      "step": 8683
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4452407657866123e-05,
+      "loss": 0.4847,
+      "step": 8684
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451163467932681e-05,
+      "loss": 0.5,
+      "step": 8685
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4449919192063825e-05,
+      "loss": 0.517,
+      "step": 8686
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4448674830283582e-05,
+      "loss": 0.515,
+      "step": 8687
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444743038261597e-05,
+      "loss": 0.5216,
+      "step": 8688
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444618584908502e-05,
+      "loss": 0.4975,
+      "step": 8689
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444494122971476e-05,
+      "loss": 0.4851,
+      "step": 8690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443696524529211e-05,
+      "loss": 0.5004,
+      "step": 8691
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444245173355241e-05,
+      "loss": 0.5239,
+      "step": 8692
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4441206856808388e-05,
+      "loss": 0.4979,
+      "step": 8693
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439961894321181e-05,
+      "loss": 0.4795,
+      "step": 8694
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443871684611482e-05,
+      "loss": 0.477,
+      "step": 8695
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437471712213347e-05,
+      "loss": 0.5057,
+      "step": 8696
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44362264926408e-05,
+      "loss": 0.4869,
+      "step": 8697
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4434981187421218e-05,
+      "loss": 0.4877,
+      "step": 8698
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4433735796578638e-05,
+      "loss": 0.4835,
+      "step": 8699
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432490320137116e-05,
+      "loss": 0.5342,
+      "step": 8700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431244758120685e-05,
+      "loss": 0.4815,
+      "step": 8701
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429999110553405e-05,
+      "loss": 0.2392,
+      "step": 8702
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4428753377459313e-05,
+      "loss": 0.4979,
+      "step": 8703
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4427507558862468e-05,
+      "loss": 0.4887,
+      "step": 8704
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426261654786915e-05,
+      "loss": 0.5281,
+      "step": 8705
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425015665256711e-05,
+      "loss": 0.4922,
+      "step": 8706
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423769590295917e-05,
+      "loss": 0.5158,
+      "step": 8707
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4422523429928582e-05,
+      "loss": 0.5189,
+      "step": 8708
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4421277184178765e-05,
+      "loss": 0.4783,
+      "step": 8709
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442003085307053e-05,
+      "loss": 0.4997,
+      "step": 8710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441878443662794e-05,
+      "loss": 0.5193,
+      "step": 8711
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4417537934875055e-05,
+      "loss": 0.4834,
+      "step": 8712
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4416291347835941e-05,
+      "loss": 0.4837,
+      "step": 8713
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4415044675534666e-05,
+      "loss": 0.5173,
+      "step": 8714
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44137979179953e-05,
+      "loss": 0.4966,
+      "step": 8715
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4412551075241912e-05,
+      "loss": 0.4767,
+      "step": 8716
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441130414729857e-05,
+      "loss": 0.5104,
+      "step": 8717
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4410057134189355e-05,
+      "loss": 0.4582,
+      "step": 8718
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4408810035938338e-05,
+      "loss": 0.4979,
+      "step": 8719
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407562852569594e-05,
+      "loss": 0.4697,
+      "step": 8720
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4406315584107206e-05,
+      "loss": 0.4759,
+      "step": 8721
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440506823057525e-05,
+      "loss": 0.494,
+      "step": 8722
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440382079199781e-05,
+      "loss": 0.5147,
+      "step": 8723
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402573268398969e-05,
+      "loss": 0.4721,
+      "step": 8724
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401325659802812e-05,
+      "loss": 0.4778,
+      "step": 8725
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400077966233426e-05,
+      "loss": 0.4864,
+      "step": 8726
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.43988301877149e-05,
+      "loss": 0.512,
+      "step": 8727
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4397582324271322e-05,
+      "loss": 0.4665,
+      "step": 8728
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4396334375926784e-05,
+      "loss": 0.466,
+      "step": 8729
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439508634270538e-05,
+      "loss": 0.5109,
+      "step": 8730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4393838224631203e-05,
+      "loss": 0.471,
+      "step": 8731
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4392590021728353e-05,
+      "loss": 0.4869,
+      "step": 8732
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4391341734020925e-05,
+      "loss": 0.5296,
+      "step": 8733
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390093361533022e-05,
+      "loss": 0.483,
+      "step": 8734
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388844904288743e-05,
+      "loss": 0.492,
+      "step": 8735
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.438759636231219e-05,
+      "loss": 0.4815,
+      "step": 8736
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.438634773562747e-05,
+      "loss": 0.4821,
+      "step": 8737
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4385099024258691e-05,
+      "loss": 0.53,
+      "step": 8738
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4383850228229956e-05,
+      "loss": 0.5029,
+      "step": 8739
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4382601347565377e-05,
+      "loss": 0.5067,
+      "step": 8740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4381352382289064e-05,
+      "loss": 0.4983,
+      "step": 8741
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4380103332425137e-05,
+      "loss": 0.4727,
+      "step": 8742
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43788541979977e-05,
+      "loss": 0.4994,
+      "step": 8743
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4377604979030877e-05,
+      "loss": 0.4885,
+      "step": 8744
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437635567554878e-05,
+      "loss": 0.4922,
+      "step": 8745
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375106287575533e-05,
+      "loss": 0.4995,
+      "step": 8746
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373856815135255e-05,
+      "loss": 0.4795,
+      "step": 8747
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4372607258252068e-05,
+      "loss": 0.2124,
+      "step": 8748
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43713576169501e-05,
+      "loss": 0.5011,
+      "step": 8749
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370107891253468e-05,
+      "loss": 0.5044,
+      "step": 8750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436885808118631e-05,
+      "loss": 0.4625,
+      "step": 8751
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436760818677275e-05,
+      "loss": 0.4771,
+      "step": 8752
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366358208036922e-05,
+      "loss": 0.4623,
+      "step": 8753
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4365108145002953e-05,
+      "loss": 0.4953,
+      "step": 8754
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436385799769498e-05,
+      "loss": 0.491,
+      "step": 8755
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362607766137138e-05,
+      "loss": 0.5058,
+      "step": 8756
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361357450353568e-05,
+      "loss": 0.4804,
+      "step": 8757
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360107050368403e-05,
+      "loss": 0.4713,
+      "step": 8758
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4358856566205793e-05,
+      "loss": 0.5046,
+      "step": 8759
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357605997889867e-05,
+      "loss": 0.5068,
+      "step": 8760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4356355345444778e-05,
+      "loss": 0.489,
+      "step": 8761
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4355104608894667e-05,
+      "loss": 0.4853,
+      "step": 8762
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4353853788263687e-05,
+      "loss": 0.493,
+      "step": 8763
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4352602883575984e-05,
+      "loss": 0.4811,
+      "step": 8764
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351351894855705e-05,
+      "loss": 0.5224,
+      "step": 8765
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4350100822127e-05,
+      "loss": 0.4971,
+      "step": 8766
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348849665414033e-05,
+      "loss": 0.5042,
+      "step": 8767
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4347598424740952e-05,
+      "loss": 0.4798,
+      "step": 8768
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434634710013191e-05,
+      "loss": 0.4952,
+      "step": 8769
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345095691611076e-05,
+      "loss": 0.4821,
+      "step": 8770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4343844199202602e-05,
+      "loss": 0.4827,
+      "step": 8771
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434259262293065e-05,
+      "loss": 0.505,
+      "step": 8772
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4341340962819385e-05,
+      "loss": 0.4906,
+      "step": 8773
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4340089218892977e-05,
+      "loss": 0.4937,
+      "step": 8774
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.4953,
+      "step": 8775
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337585479691379e-05,
+      "loss": 0.5037,
+      "step": 8776
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4336333484464527e-05,
+      "loss": 0.5214,
+      "step": 8777
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335081405519206e-05,
+      "loss": 0.513,
+      "step": 8778
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4333829242879585e-05,
+      "loss": 0.5382,
+      "step": 8779
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332576996569838e-05,
+      "loss": 0.4868,
+      "step": 8780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4331324666614144e-05,
+      "loss": 0.506,
+      "step": 8781
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.433007225303668e-05,
+      "loss": 0.4962,
+      "step": 8782
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4328819755861622e-05,
+      "loss": 0.5015,
+      "step": 8783
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4327567175113157e-05,
+      "loss": 0.1898,
+      "step": 8784
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326314510815464e-05,
+      "loss": 0.4946,
+      "step": 8785
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432506176299273e-05,
+      "loss": 0.5241,
+      "step": 8786
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323808931669137e-05,
+      "loss": 0.4973,
+      "step": 8787
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4322556016868872e-05,
+      "loss": 0.5102,
+      "step": 8788
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4321303018616129e-05,
+      "loss": 0.5005,
+      "step": 8789
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4320049936935097e-05,
+      "loss": 0.5144,
+      "step": 8790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318796771849966e-05,
+      "loss": 0.4738,
+      "step": 8791
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317543523384928e-05,
+      "loss": 0.4924,
+      "step": 8792
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316290191564188e-05,
+      "loss": 0.5013,
+      "step": 8793
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4315036776411937e-05,
+      "loss": 0.5184,
+      "step": 8794
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313783277952371e-05,
+      "loss": 0.5114,
+      "step": 8795
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312529696209693e-05,
+      "loss": 0.4964,
+      "step": 8796
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431127603120811e-05,
+      "loss": 0.5047,
+      "step": 8797
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310022282971816e-05,
+      "loss": 0.528,
+      "step": 8798
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4308768451525024e-05,
+      "loss": 0.4927,
+      "step": 8799
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307514536891938e-05,
+      "loss": 0.4744,
+      "step": 8800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4306260539096765e-05,
+      "loss": 0.4802,
+      "step": 8801
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4305006458163719e-05,
+      "loss": 0.4792,
+      "step": 8802
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4303752294117006e-05,
+      "loss": 0.4665,
+      "step": 8803
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4302498046980844e-05,
+      "loss": 0.4793,
+      "step": 8804
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301243716779447e-05,
+      "loss": 0.4848,
+      "step": 8805
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429998930353703e-05,
+      "loss": 0.4984,
+      "step": 8806
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4298734807277812e-05,
+      "loss": 0.4986,
+      "step": 8807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4297480228026013e-05,
+      "loss": 0.4877,
+      "step": 8808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4296225565805854e-05,
+      "loss": 0.4943,
+      "step": 8809
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4294970820641556e-05,
+      "loss": 0.5198,
+      "step": 8810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293715992557344e-05,
+      "loss": 0.4749,
+      "step": 8811
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429246108157745e-05,
+      "loss": 0.4878,
+      "step": 8812
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429120608772609e-05,
+      "loss": 0.5144,
+      "step": 8813
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4289951011027505e-05,
+      "loss": 0.457,
+      "step": 8814
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288695851505919e-05,
+      "loss": 0.4974,
+      "step": 8815
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4287440609185568e-05,
+      "loss": 0.4577,
+      "step": 8816
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4286185284090683e-05,
+      "loss": 0.4691,
+      "step": 8817
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4284929876245501e-05,
+      "loss": 0.5156,
+      "step": 8818
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4283674385674258e-05,
+      "loss": 0.5259,
+      "step": 8819
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282418812401197e-05,
+      "loss": 0.4767,
+      "step": 8820
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4281163156450553e-05,
+      "loss": 0.4918,
+      "step": 8821
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4279907417846572e-05,
+      "loss": 0.4945,
+      "step": 8822
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4278651596613494e-05,
+      "loss": 0.4629,
+      "step": 8823
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427739569277557e-05,
+      "loss": 0.5053,
+      "step": 8824
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276139706357042e-05,
+      "loss": 0.4944,
+      "step": 8825
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4274883637382162e-05,
+      "loss": 0.4805,
+      "step": 8826
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4273627485875175e-05,
+      "loss": 0.4828,
+      "step": 8827
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4272371251860337e-05,
+      "loss": 0.469,
+      "step": 8828
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4271114935361897e-05,
+      "loss": 0.4965,
+      "step": 8829
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269858536404116e-05,
+      "loss": 0.4973,
+      "step": 8830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268602055011242e-05,
+      "loss": 0.4672,
+      "step": 8831
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4267345491207543e-05,
+      "loss": 0.4993,
+      "step": 8832
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4266088845017272e-05,
+      "loss": 0.4513,
+      "step": 8833
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264832116464691e-05,
+      "loss": 0.5001,
+      "step": 8834
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263575305574065e-05,
+      "loss": 0.4893,
+      "step": 8835
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426231841236965e-05,
+      "loss": 0.4919,
+      "step": 8836
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261061436875725e-05,
+      "loss": 0.496,
+      "step": 8837
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4259804379116551e-05,
+      "loss": 0.541,
+      "step": 8838
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258547239116396e-05,
+      "loss": 0.4922,
+      "step": 8839
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425729001689953e-05,
+      "loss": 0.4941,
+      "step": 8840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425603271249023e-05,
+      "loss": 0.5098,
+      "step": 8841
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4254775325912765e-05,
+      "loss": 0.4606,
+      "step": 8842
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4253517857191415e-05,
+      "loss": 0.5075,
+      "step": 8843
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252260306350451e-05,
+      "loss": 0.4862,
+      "step": 8844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4251002673414158e-05,
+      "loss": 0.4842,
+      "step": 8845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4249744958406813e-05,
+      "loss": 0.4894,
+      "step": 8846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4248487161352697e-05,
+      "loss": 0.4852,
+      "step": 8847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4247229282276095e-05,
+      "loss": 0.5132,
+      "step": 8848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4245971321201293e-05,
+      "loss": 0.4784,
+      "step": 8849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244713278152575e-05,
+      "loss": 0.4868,
+      "step": 8850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4243455153154231e-05,
+      "loss": 0.4376,
+      "step": 8851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424219694623055e-05,
+      "loss": 0.4819,
+      "step": 8852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240938657405823e-05,
+      "loss": 0.5238,
+      "step": 8853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4239680286704341e-05,
+      "loss": 0.5161,
+      "step": 8854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238421834150403e-05,
+      "loss": 0.4704,
+      "step": 8855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237163299768301e-05,
+      "loss": 0.482,
+      "step": 8856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235904683582336e-05,
+      "loss": 0.4823,
+      "step": 8857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4234645985616805e-05,
+      "loss": 0.5054,
+      "step": 8858
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4233387205896008e-05,
+      "loss": 0.4659,
+      "step": 8859
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4232128344444251e-05,
+      "loss": 0.4634,
+      "step": 8860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4230869401285834e-05,
+      "loss": 0.4911,
+      "step": 8861
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229610376445065e-05,
+      "loss": 0.491,
+      "step": 8862
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228351269946248e-05,
+      "loss": 0.5021,
+      "step": 8863
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4227092081813696e-05,
+      "loss": 0.5069,
+      "step": 8864
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225832812071716e-05,
+      "loss": 0.4798,
+      "step": 8865
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224573460744619e-05,
+      "loss": 0.5144,
+      "step": 8866
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4223314027856723e-05,
+      "loss": 0.4855,
+      "step": 8867
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422205451343234e-05,
+      "loss": 0.4674,
+      "step": 8868
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4220794917495787e-05,
+      "loss": 0.4841,
+      "step": 8869
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219535240071378e-05,
+      "loss": 0.4925,
+      "step": 8870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421827548118344e-05,
+      "loss": 0.4886,
+      "step": 8871
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421701564085629e-05,
+      "loss": 0.4886,
+      "step": 8872
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215755719114252e-05,
+      "loss": 0.4645,
+      "step": 8873
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4214495715981648e-05,
+      "loss": 0.4902,
+      "step": 8874
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213235631482809e-05,
+      "loss": 0.477,
+      "step": 8875
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421197546564206e-05,
+      "loss": 0.187,
+      "step": 8876
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210715218483726e-05,
+      "loss": 0.21,
+      "step": 8877
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4209454890032144e-05,
+      "loss": 0.4935,
+      "step": 8878
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4208194480311646e-05,
+      "loss": 0.5183,
+      "step": 8879
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206933989346559e-05,
+      "loss": 0.523,
+      "step": 8880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205673417161222e-05,
+      "loss": 0.5176,
+      "step": 8881
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204412763779975e-05,
+      "loss": 0.4869,
+      "step": 8882
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4203152029227157e-05,
+      "loss": 0.5161,
+      "step": 8883
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4201891213527102e-05,
+      "loss": 0.5016,
+      "step": 8884
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4200630316704153e-05,
+      "loss": 0.5099,
+      "step": 8885
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4199369338782659e-05,
+      "loss": 0.5337,
+      "step": 8886
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198108279786962e-05,
+      "loss": 0.5078,
+      "step": 8887
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4196847139741401e-05,
+      "loss": 0.485,
+      "step": 8888
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4195585918670338e-05,
+      "loss": 0.504,
+      "step": 8889
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419432461659811e-05,
+      "loss": 0.5217,
+      "step": 8890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4193063233549076e-05,
+      "loss": 0.1944,
+      "step": 8891
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191801769547582e-05,
+      "loss": 0.5026,
+      "step": 8892
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419054022461799e-05,
+      "loss": 0.5005,
+      "step": 8893
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4189278598784648e-05,
+      "loss": 0.513,
+      "step": 8894
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418801689207192e-05,
+      "loss": 0.4749,
+      "step": 8895
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4186755104504157e-05,
+      "loss": 0.5005,
+      "step": 8896
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185493236105727e-05,
+      "loss": 0.4921,
+      "step": 8897
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4184231286900991e-05,
+      "loss": 0.4891,
+      "step": 8898
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4182969256914305e-05,
+      "loss": 0.5056,
+      "step": 8899
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181707146170046e-05,
+      "loss": 0.4817,
+      "step": 8900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180444954692569e-05,
+      "loss": 0.499,
+      "step": 8901
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4179182682506252e-05,
+      "loss": 0.4652,
+      "step": 8902
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177920329635457e-05,
+      "loss": 0.4936,
+      "step": 8903
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417665789610456e-05,
+      "loss": 0.4785,
+      "step": 8904
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175395381937933e-05,
+      "loss": 0.4987,
+      "step": 8905
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174132787159949e-05,
+      "loss": 0.48,
+      "step": 8906
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172870111794984e-05,
+      "loss": 0.4726,
+      "step": 8907
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4171607355867419e-05,
+      "loss": 0.5229,
+      "step": 8908
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4170344519401629e-05,
+      "loss": 0.4656,
+      "step": 8909
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169081602421996e-05,
+      "loss": 0.4908,
+      "step": 8910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4167818604952906e-05,
+      "loss": 0.4715,
+      "step": 8911
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166555527018737e-05,
+      "loss": 0.4777,
+      "step": 8912
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4165292368643876e-05,
+      "loss": 0.4704,
+      "step": 8913
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4164029129852712e-05,
+      "loss": 0.4819,
+      "step": 8914
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416276581066963e-05,
+      "loss": 0.4761,
+      "step": 8915
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4161502411119025e-05,
+      "loss": 0.507,
+      "step": 8916
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4160238931225286e-05,
+      "loss": 0.4954,
+      "step": 8917
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158975371012802e-05,
+      "loss": 0.5114,
+      "step": 8918
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4157711730505974e-05,
+      "loss": 0.486,
+      "step": 8919
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156448009729195e-05,
+      "loss": 0.4969,
+      "step": 8920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4155184208706864e-05,
+      "loss": 0.5184,
+      "step": 8921
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4153920327463378e-05,
+      "loss": 0.5058,
+      "step": 8922
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152656366023144e-05,
+      "loss": 0.5149,
+      "step": 8923
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4151392324410554e-05,
+      "loss": 0.516,
+      "step": 8924
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415012820265002e-05,
+      "loss": 0.497,
+      "step": 8925
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4148864000765942e-05,
+      "loss": 0.4756,
+      "step": 8926
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147599718782737e-05,
+      "loss": 0.4736,
+      "step": 8927
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146335356724803e-05,
+      "loss": 0.5172,
+      "step": 8928
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4145070914616555e-05,
+      "loss": 0.5096,
+      "step": 8929
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4143806392482403e-05,
+      "loss": 0.4901,
+      "step": 8930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4142541790346762e-05,
+      "loss": 0.4687,
+      "step": 8931
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4141277108234045e-05,
+      "loss": 0.5068,
+      "step": 8932
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140012346168669e-05,
+      "loss": 0.4958,
+      "step": 8933
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413874750417505e-05,
+      "loss": 0.4634,
+      "step": 8934
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4137482582277611e-05,
+      "loss": 0.4835,
+      "step": 8935
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4136217580500771e-05,
+      "loss": 0.466,
+      "step": 8936
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134952498868953e-05,
+      "loss": 0.506,
+      "step": 8937
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413368733740658e-05,
+      "loss": 0.4843,
+      "step": 8938
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4132422096138081e-05,
+      "loss": 0.5026,
+      "step": 8939
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131156775087877e-05,
+      "loss": 0.5049,
+      "step": 8940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41298913742804e-05,
+      "loss": 0.4951,
+      "step": 8941
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4128625893740084e-05,
+      "loss": 0.4785,
+      "step": 8942
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4127360333491354e-05,
+      "loss": 0.4629,
+      "step": 8943
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126094693558645e-05,
+      "loss": 0.4492,
+      "step": 8944
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124828973966393e-05,
+      "loss": 0.4599,
+      "step": 8945
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4123563174739036e-05,
+      "loss": 0.4969,
+      "step": 8946
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4122297295901011e-05,
+      "loss": 0.5019,
+      "step": 8947
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4121031337476753e-05,
+      "loss": 0.4806,
+      "step": 8948
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411976529949071e-05,
+      "loss": 0.5131,
+      "step": 8949
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118499181967322e-05,
+      "loss": 0.4624,
+      "step": 8950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117232984931026e-05,
+      "loss": 0.4821,
+      "step": 8951
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4115966708406282e-05,
+      "loss": 0.488,
+      "step": 8952
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4114700352417523e-05,
+      "loss": 0.4869,
+      "step": 8953
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4113433916989205e-05,
+      "loss": 0.4992,
+      "step": 8954
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112167402145773e-05,
+      "loss": 0.4937,
+      "step": 8955
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110900807911688e-05,
+      "loss": 0.491,
+      "step": 8956
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4109634134311395e-05,
+      "loss": 0.5264,
+      "step": 8957
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410836738136935e-05,
+      "loss": 0.4912,
+      "step": 8958
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107100549110009e-05,
+      "loss": 0.5036,
+      "step": 8959
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105833637557835e-05,
+      "loss": 0.4899,
+      "step": 8960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4104566646737283e-05,
+      "loss": 0.4806,
+      "step": 8961
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410329957667281e-05,
+      "loss": 0.528,
+      "step": 8962
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410203242738889e-05,
+      "loss": 0.4888,
+      "step": 8963
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4100765198909975e-05,
+      "loss": 0.492,
+      "step": 8964
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4099497891260538e-05,
+      "loss": 0.4909,
+      "step": 8965
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4098230504465041e-05,
+      "loss": 0.4703,
+      "step": 8966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4096963038547956e-05,
+      "loss": 0.4681,
+      "step": 8967
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095695493533753e-05,
+      "loss": 0.5133,
+      "step": 8968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094427869446898e-05,
+      "loss": 0.4981,
+      "step": 8969
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4093160166311873e-05,
+      "loss": 0.4861,
+      "step": 8970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091892384153148e-05,
+      "loss": 0.4687,
+      "step": 8971
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4090624522995197e-05,
+      "loss": 0.5033,
+      "step": 8972
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089356582862501e-05,
+      "loss": 0.494,
+      "step": 8973
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4088088563779536e-05,
+      "loss": 0.4646,
+      "step": 8974
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408682046577079e-05,
+      "loss": 0.4986,
+      "step": 8975
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4085552288860737e-05,
+      "loss": 0.1988,
+      "step": 8976
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4084284033073865e-05,
+      "loss": 0.4768,
+      "step": 8977
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083015698434657e-05,
+      "loss": 0.5187,
+      "step": 8978
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4081747284967602e-05,
+      "loss": 0.4823,
+      "step": 8979
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080478792697188e-05,
+      "loss": 0.4709,
+      "step": 8980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079210221647903e-05,
+      "loss": 0.514,
+      "step": 8981
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077941571844243e-05,
+      "loss": 0.5017,
+      "step": 8982
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4076672843310695e-05,
+      "loss": 0.5002,
+      "step": 8983
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4075404036071755e-05,
+      "loss": 0.5131,
+      "step": 8984
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4074135150151922e-05,
+      "loss": 0.477,
+      "step": 8985
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407286618557569e-05,
+      "loss": 0.4898,
+      "step": 8986
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407159714236756e-05,
+      "loss": 0.4967,
+      "step": 8987
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4070328020552033e-05,
+      "loss": 0.4821,
+      "step": 8988
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406905882015361e-05,
+      "loss": 0.4951,
+      "step": 8989
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067789541196795e-05,
+      "loss": 0.4822,
+      "step": 8990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4066520183706091e-05,
+      "loss": 0.5131,
+      "step": 8991
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4065250747706007e-05,
+      "loss": 0.4791,
+      "step": 8992
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406398123322105e-05,
+      "loss": 0.4693,
+      "step": 8993
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406271164027573e-05,
+      "loss": 0.4684,
+      "step": 8994
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4061441968894559e-05,
+      "loss": 0.4897,
+      "step": 8995
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4060172219102046e-05,
+      "loss": 0.4935,
+      "step": 8996
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405890239092271e-05,
+      "loss": 0.4936,
+      "step": 8997
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4057632484381063e-05,
+      "loss": 0.4921,
+      "step": 8998
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4056362499501623e-05,
+      "loss": 0.5069,
+      "step": 8999
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405509243630891e-05,
+      "loss": 0.4853,
+      "step": 9000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4053822294827443e-05,
+      "loss": 0.4734,
+      "step": 9001
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405255207508175e-05,
+      "loss": 0.499,
+      "step": 9002
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4051281777096339e-05,
+      "loss": 0.4672,
+      "step": 9003
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4050011400895752e-05,
+      "loss": 0.4566,
+      "step": 9004
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4048740946504506e-05,
+      "loss": 0.4975,
+      "step": 9005
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047470413947126e-05,
+      "loss": 0.5126,
+      "step": 9006
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046199803248149e-05,
+      "loss": 0.4771,
+      "step": 9007
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40449291144321e-05,
+      "loss": 0.4677,
+      "step": 9008
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043658347523516e-05,
+      "loss": 0.486,
+      "step": 9009
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4042387502546925e-05,
+      "loss": 0.4915,
+      "step": 9010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041116579526868e-05,
+      "loss": 0.5197,
+      "step": 9011
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4039845578487875e-05,
+      "loss": 0.5113,
+      "step": 9012
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4038574499454496e-05,
+      "loss": 0.4895,
+      "step": 9013
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4037303342451257e-05,
+      "loss": 0.4858,
+      "step": 9014
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4036032107502708e-05,
+      "loss": 0.5204,
+      "step": 9015
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403476079463339e-05,
+      "loss": 0.5069,
+      "step": 9016
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4033489403867849e-05,
+      "loss": 0.4962,
+      "step": 9017
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032217935230623e-05,
+      "loss": 0.4867,
+      "step": 9018
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403094638874627e-05,
+      "loss": 0.4805,
+      "step": 9019
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4029674764439331e-05,
+      "loss": 0.4939,
+      "step": 9020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4028403062334362e-05,
+      "loss": 0.5165,
+      "step": 9021
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027131282455908e-05,
+      "loss": 0.4901,
+      "step": 9022
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402585942482853e-05,
+      "loss": 0.5297,
+      "step": 9023
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4024587489476778e-05,
+      "loss": 0.4916,
+      "step": 9024
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4023315476425207e-05,
+      "loss": 0.4886,
+      "step": 9025
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022043385698384e-05,
+      "loss": 0.4908,
+      "step": 9026
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020771217320857e-05,
+      "loss": 0.5227,
+      "step": 9027
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019498971317192e-05,
+      "loss": 0.4754,
+      "step": 9028
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018226647711948e-05,
+      "loss": 0.4949,
+      "step": 9029
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.5073,
+      "step": 9030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4015681767794995e-05,
+      "loss": 0.4934,
+      "step": 9031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4014409211532417e-05,
+      "loss": 0.4985,
+      "step": 9032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4013136577766525e-05,
+      "loss": 0.5122,
+      "step": 9033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401186386652189e-05,
+      "loss": 0.4878,
+      "step": 9034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4010591077823086e-05,
+      "loss": 0.4864,
+      "step": 9035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4009318211694683e-05,
+      "loss": 0.4779,
+      "step": 9036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008045268161256e-05,
+      "loss": 0.4817,
+      "step": 9037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4006772247247384e-05,
+      "loss": 0.4965,
+      "step": 9038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005499148977641e-05,
+      "loss": 0.4771,
+      "step": 9039
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004225973376604e-05,
+      "loss": 0.4855,
+      "step": 9040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4002952720468856e-05,
+      "loss": 0.5055,
+      "step": 9041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4001679390278983e-05,
+      "loss": 0.4855,
+      "step": 9042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000405982831559e-05,
+      "loss": 0.4765,
+      "step": 9043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3999132498151176e-05,
+      "loss": 0.5269,
+      "step": 9044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3997858936262415e-05,
+      "loss": 0.5176,
+      "step": 9045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3996585297189871e-05,
+      "loss": 0.4883,
+      "step": 9046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995311580958124e-05,
+      "loss": 0.5003,
+      "step": 9047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3994037787591773e-05,
+      "loss": 0.4765,
+      "step": 9048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992763917115406e-05,
+      "loss": 0.5223,
+      "step": 9049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991489969553615e-05,
+      "loss": 0.4777,
+      "step": 9050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3990215944931e-05,
+      "loss": 0.4831,
+      "step": 9051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3988941843272153e-05,
+      "loss": 0.5086,
+      "step": 9052
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398766766460168e-05,
+      "loss": 0.5299,
+      "step": 9053
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986393408944168e-05,
+      "loss": 0.5027,
+      "step": 9054
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985119076324228e-05,
+      "loss": 0.4914,
+      "step": 9055
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3983844666766459e-05,
+      "loss": 0.4687,
+      "step": 9056
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398257018029547e-05,
+      "loss": 0.473,
+      "step": 9057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981295616935857e-05,
+      "loss": 0.5065,
+      "step": 9058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980020976712234e-05,
+      "loss": 0.5052,
+      "step": 9059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397874625964921e-05,
+      "loss": 0.4948,
+      "step": 9060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3977471465771393e-05,
+      "loss": 0.5258,
+      "step": 9061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976196595103395e-05,
+      "loss": 0.5088,
+      "step": 9062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3974921647669826e-05,
+      "loss": 0.4897,
+      "step": 9063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3973646623495305e-05,
+      "loss": 0.517,
+      "step": 9064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972371522604449e-05,
+      "loss": 0.5109,
+      "step": 9065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3971096345021868e-05,
+      "loss": 0.4601,
+      "step": 9066
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969821090772188e-05,
+      "loss": 0.5073,
+      "step": 9067
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968545759880028e-05,
+      "loss": 0.4842,
+      "step": 9068
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967270352370006e-05,
+      "loss": 0.5008,
+      "step": 9069
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396599486826675e-05,
+      "loss": 0.492,
+      "step": 9070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3964719307594882e-05,
+      "loss": 0.5159,
+      "step": 9071
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3963443670379034e-05,
+      "loss": 0.5109,
+      "step": 9072
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962167956643825e-05,
+      "loss": 0.4724,
+      "step": 9073
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3960892166413888e-05,
+      "loss": 0.5014,
+      "step": 9074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3959616299713857e-05,
+      "loss": 0.463,
+      "step": 9075
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3958340356568361e-05,
+      "loss": 0.5114,
+      "step": 9076
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3957064337002031e-05,
+      "loss": 0.4826,
+      "step": 9077
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395578824103951e-05,
+      "loss": 0.5092,
+      "step": 9078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3954512068705425e-05,
+      "loss": 0.4831,
+      "step": 9079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953235820024427e-05,
+      "loss": 0.4786,
+      "step": 9080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395195949502114e-05,
+      "loss": 0.5121,
+      "step": 9081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3950683093720218e-05,
+      "loss": 0.4994,
+      "step": 9082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3949406616146298e-05,
+      "loss": 0.4634,
+      "step": 9083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948130062324023e-05,
+      "loss": 0.4729,
+      "step": 9084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394685343227804e-05,
+      "loss": 0.4582,
+      "step": 9085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3945576726032997e-05,
+      "loss": 0.4901,
+      "step": 9086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3944299943613541e-05,
+      "loss": 0.4883,
+      "step": 9087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943023085044323e-05,
+      "loss": 0.4998,
+      "step": 9088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941746150349995e-05,
+      "loss": 0.5232,
+      "step": 9089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3940469139555209e-05,
+      "loss": 0.468,
+      "step": 9090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3939192052684618e-05,
+      "loss": 0.4803,
+      "step": 9091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937914889762875e-05,
+      "loss": 0.4641,
+      "step": 9092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936637650814645e-05,
+      "loss": 0.5157,
+      "step": 9093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3935360335864582e-05,
+      "loss": 0.5105,
+      "step": 9094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393408294493735e-05,
+      "loss": 0.4935,
+      "step": 9095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3932805478057603e-05,
+      "loss": 0.4978,
+      "step": 9096
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931527935250013e-05,
+      "loss": 0.4944,
+      "step": 9097
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3930250316539237e-05,
+      "loss": 0.4737,
+      "step": 9098
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928972621949945e-05,
+      "loss": 0.4917,
+      "step": 9099
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927694851506805e-05,
+      "loss": 0.4945,
+      "step": 9100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3926417005234487e-05,
+      "loss": 0.4938,
+      "step": 9101
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925139083157658e-05,
+      "loss": 0.5109,
+      "step": 9102
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3923861085300989e-05,
+      "loss": 0.4872,
+      "step": 9103
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3922583011689158e-05,
+      "loss": 0.483,
+      "step": 9104
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921304862346838e-05,
+      "loss": 0.4904,
+      "step": 9105
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3920026637298704e-05,
+      "loss": 0.4892,
+      "step": 9106
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918748336569437e-05,
+      "loss": 0.512,
+      "step": 9107
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391746996018371e-05,
+      "loss": 0.498,
+      "step": 9108
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391619150816621e-05,
+      "loss": 0.4772,
+      "step": 9109
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3914912980541617e-05,
+      "loss": 0.5124,
+      "step": 9110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3913634377334612e-05,
+      "loss": 0.4897,
+      "step": 9111
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912355698569889e-05,
+      "loss": 0.4939,
+      "step": 9112
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3911076944272122e-05,
+      "loss": 0.4724,
+      "step": 9113
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3909798114466004e-05,
+      "loss": 0.5026,
+      "step": 9114
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908519209176227e-05,
+      "loss": 0.4765,
+      "step": 9115
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3907240228427485e-05,
+      "loss": 0.4927,
+      "step": 9116
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905961172244462e-05,
+      "loss": 0.1943,
+      "step": 9117
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3904682040651855e-05,
+      "loss": 0.4864,
+      "step": 9118
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.390340283367436e-05,
+      "loss": 0.4888,
+      "step": 9119
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902123551336674e-05,
+      "loss": 0.4722,
+      "step": 9120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900844193663497e-05,
+      "loss": 0.4861,
+      "step": 9121
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3899564760679522e-05,
+      "loss": 0.4892,
+      "step": 9122
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898285252409456e-05,
+      "loss": 0.4843,
+      "step": 9123
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3897005668878003e-05,
+      "loss": 0.4583,
+      "step": 9124
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895726010109861e-05,
+      "loss": 0.5374,
+      "step": 9125
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3894446276129738e-05,
+      "loss": 0.4934,
+      "step": 9126
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3893166466962342e-05,
+      "loss": 0.513,
+      "step": 9127
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3891886582632384e-05,
+      "loss": 0.4475,
+      "step": 9128
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3890606623164567e-05,
+      "loss": 0.4786,
+      "step": 9129
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3889326588583605e-05,
+      "loss": 0.19,
+      "step": 9130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3888046478914215e-05,
+      "loss": 0.4799,
+      "step": 9131
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886766294181105e-05,
+      "loss": 0.4629,
+      "step": 9132
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3885486034408992e-05,
+      "loss": 0.4646,
+      "step": 9133
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3884205699622598e-05,
+      "loss": 0.5109,
+      "step": 9134
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3882925289846637e-05,
+      "loss": 0.4972,
+      "step": 9135
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3881644805105827e-05,
+      "loss": 0.4827,
+      "step": 9136
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880364245424896e-05,
+      "loss": 0.4839,
+      "step": 9137
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387908361082856e-05,
+      "loss": 0.4965,
+      "step": 9138
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3877802901341549e-05,
+      "loss": 0.4983,
+      "step": 9139
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876522116988583e-05,
+      "loss": 0.4974,
+      "step": 9140
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875241257794394e-05,
+      "loss": 0.4975,
+      "step": 9141
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873960323783706e-05,
+      "loss": 0.4978,
+      "step": 9142
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872679314981256e-05,
+      "loss": 0.499,
+      "step": 9143
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3871398231411768e-05,
+      "loss": 0.4963,
+      "step": 9144
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870117073099983e-05,
+      "loss": 0.5,
+      "step": 9145
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3868835840070627e-05,
+      "loss": 0.4931,
+      "step": 9146
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867554532348442e-05,
+      "loss": 0.4853,
+      "step": 9147
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3866273149958158e-05,
+      "loss": 0.5038,
+      "step": 9148
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3864991692924524e-05,
+      "loss": 0.4693,
+      "step": 9149
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863710161272273e-05,
+      "loss": 0.4832,
+      "step": 9150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3862428555026148e-05,
+      "loss": 0.4643,
+      "step": 9151
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3861146874210891e-05,
+      "loss": 0.4929,
+      "step": 9152
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3859865118851252e-05,
+      "loss": 0.504,
+      "step": 9153
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858583288971975e-05,
+      "loss": 0.467,
+      "step": 9154
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857301384597796e-05,
+      "loss": 0.501,
+      "step": 9155
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385601940575348e-05,
+      "loss": 0.4758,
+      "step": 9156
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385473735246377e-05,
+      "loss": 0.4714,
+      "step": 9157
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3853455224753419e-05,
+      "loss": 0.4965,
+      "step": 9158
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852173022647173e-05,
+      "loss": 0.4658,
+      "step": 9159
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850890746169796e-05,
+      "loss": 0.4899,
+      "step": 9160
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849608395346042e-05,
+      "loss": 0.4801,
+      "step": 9161
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3848325970200665e-05,
+      "loss": 0.4788,
+      "step": 9162
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3847043470758426e-05,
+      "loss": 0.492,
+      "step": 9163
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384576089704409e-05,
+      "loss": 0.5184,
+      "step": 9164
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844478249082407e-05,
+      "loss": 0.5099,
+      "step": 9165
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384319552689815e-05,
+      "loss": 0.5039,
+      "step": 9166
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841912730516081e-05,
+      "loss": 0.4946,
+      "step": 9167
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840629859960967e-05,
+      "loss": 0.5191,
+      "step": 9168
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3839346915257574e-05,
+      "loss": 0.4988,
+      "step": 9169
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383806389643067e-05,
+      "loss": 0.4998,
+      "step": 9170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3836780803505025e-05,
+      "loss": 0.4776,
+      "step": 9171
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3835497636505415e-05,
+      "loss": 0.4513,
+      "step": 9172
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383421439545661e-05,
+      "loss": 0.475,
+      "step": 9173
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832931080383383e-05,
+      "loss": 0.4847,
+      "step": 9174
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3831647691310514e-05,
+      "loss": 0.4877,
+      "step": 9175
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3830364228262778e-05,
+      "loss": 0.4894,
+      "step": 9176
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3829080691264953e-05,
+      "loss": 0.5363,
+      "step": 9177
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827797080341822e-05,
+      "loss": 0.487,
+      "step": 9178
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826513395518167e-05,
+      "loss": 0.4995,
+      "step": 9179
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3825229636818767e-05,
+      "loss": 0.5404,
+      "step": 9180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382394580426841e-05,
+      "loss": 0.4734,
+      "step": 9181
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382266189789188e-05,
+      "loss": 0.504,
+      "step": 9182
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821377917713969e-05,
+      "loss": 0.4683,
+      "step": 9183
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382009386375946e-05,
+      "loss": 0.4863,
+      "step": 9184
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818809736053147e-05,
+      "loss": 0.4782,
+      "step": 9185
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381752553461982e-05,
+      "loss": 0.4756,
+      "step": 9186
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3816241259484275e-05,
+      "loss": 0.4435,
+      "step": 9187
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3814956910671302e-05,
+      "loss": 0.4799,
+      "step": 9188
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38136724882057e-05,
+      "loss": 0.4945,
+      "step": 9189
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3812387992112264e-05,
+      "loss": 0.4616,
+      "step": 9190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3811103422415799e-05,
+      "loss": 0.4758,
+      "step": 9191
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809818779141097e-05,
+      "loss": 0.4884,
+      "step": 9192
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3808534062312964e-05,
+      "loss": 0.4694,
+      "step": 9193
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3807249271956204e-05,
+      "loss": 0.4939,
+      "step": 9194
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380596440809562e-05,
+      "loss": 0.4903,
+      "step": 9195
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804679470756019e-05,
+      "loss": 0.48,
+      "step": 9196
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803394459962207e-05,
+      "loss": 0.4693,
+      "step": 9197
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3802109375738995e-05,
+      "loss": 0.5072,
+      "step": 9198
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3800824218111188e-05,
+      "loss": 0.4851,
+      "step": 9199
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.37995389871036e-05,
+      "loss": 0.4806,
+      "step": 9200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3798253682741049e-05,
+      "loss": 0.4957,
+      "step": 9201
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796968305048344e-05,
+      "loss": 0.5048,
+      "step": 9202
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795682854050303e-05,
+      "loss": 0.4685,
+      "step": 9203
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3794397329771742e-05,
+      "loss": 0.4789,
+      "step": 9204
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379311173223748e-05,
+      "loss": 0.5,
+      "step": 9205
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379182606147234e-05,
+      "loss": 0.5198,
+      "step": 9206
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3790540317501138e-05,
+      "loss": 0.4948,
+      "step": 9207
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3789254500348704e-05,
+      "loss": 0.466,
+      "step": 9208
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3787968610039854e-05,
+      "loss": 0.4775,
+      "step": 9209
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3786682646599421e-05,
+      "loss": 0.4647,
+      "step": 9210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3785396610052227e-05,
+      "loss": 0.49,
+      "step": 9211
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3784110500423104e-05,
+      "loss": 0.4626,
+      "step": 9212
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378282431773688e-05,
+      "loss": 0.5179,
+      "step": 9213
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378153806201839e-05,
+      "loss": 0.4689,
+      "step": 9214
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378025173329246e-05,
+      "loss": 0.5033,
+      "step": 9215
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3778965331583933e-05,
+      "loss": 0.4879,
+      "step": 9216
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777678856917637e-05,
+      "loss": 0.4969,
+      "step": 9217
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776392309318411e-05,
+      "loss": 0.5225,
+      "step": 9218
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3775105688811098e-05,
+      "loss": 0.4968,
+      "step": 9219
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773818995420532e-05,
+      "loss": 0.4886,
+      "step": 9220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3772532229171555e-05,
+      "loss": 0.5057,
+      "step": 9221
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771245390089012e-05,
+      "loss": 0.4668,
+      "step": 9222
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376995847819775e-05,
+      "loss": 0.4954,
+      "step": 9223
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768671493522605e-05,
+      "loss": 0.4803,
+      "step": 9224
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3767384436088434e-05,
+      "loss": 0.5145,
+      "step": 9225
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766097305920076e-05,
+      "loss": 0.4688,
+      "step": 9226
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3764810103042391e-05,
+      "loss": 0.5014,
+      "step": 9227
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3763522827480223e-05,
+      "loss": 0.5093,
+      "step": 9228
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3762235479258425e-05,
+      "loss": 0.4952,
+      "step": 9229
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760948058401855e-05,
+      "loss": 0.4963,
+      "step": 9230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3759660564935363e-05,
+      "loss": 0.481,
+      "step": 9231
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758372998883809e-05,
+      "loss": 0.4824,
+      "step": 9232
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757085360272049e-05,
+      "loss": 0.4673,
+      "step": 9233
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3755797649124944e-05,
+      "loss": 0.5149,
+      "step": 9234
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754509865467359e-05,
+      "loss": 0.4863,
+      "step": 9235
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753222009324148e-05,
+      "loss": 0.4892,
+      "step": 9236
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3751934080720179e-05,
+      "loss": 0.5019,
+      "step": 9237
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375064607968032e-05,
+      "loss": 0.4928,
+      "step": 9238
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374935800622943e-05,
+      "loss": 0.4541,
+      "step": 9239
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748069860392381e-05,
+      "loss": 0.4895,
+      "step": 9240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746781642194045e-05,
+      "loss": 0.5247,
+      "step": 9241
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3745493351659289e-05,
+      "loss": 0.5237,
+      "step": 9242
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3744204988812989e-05,
+      "loss": 0.4495,
+      "step": 9243
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3742916553680012e-05,
+      "loss": 0.4715,
+      "step": 9244
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741628046285239e-05,
+      "loss": 0.5037,
+      "step": 9245
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740339466653545e-05,
+      "loss": 0.49,
+      "step": 9246
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3739050814809805e-05,
+      "loss": 0.4844,
+      "step": 9247
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37377620907789e-05,
+      "loss": 0.503,
+      "step": 9248
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3736473294585708e-05,
+      "loss": 0.4681,
+      "step": 9249
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3735184426255117e-05,
+      "loss": 0.1879,
+      "step": 9250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3733895485812005e-05,
+      "loss": 0.4861,
+      "step": 9251
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732606473281258e-05,
+      "loss": 0.4681,
+      "step": 9252
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731317388687761e-05,
+      "loss": 0.4704,
+      "step": 9253
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3730028232056405e-05,
+      "loss": 0.5198,
+      "step": 9254
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728739003412075e-05,
+      "loss": 0.4896,
+      "step": 9255
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727449702779661e-05,
+      "loss": 0.4786,
+      "step": 9256
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3726160330184058e-05,
+      "loss": 0.4944,
+      "step": 9257
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372487088565016e-05,
+      "loss": 0.4871,
+      "step": 9258
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3723581369202852e-05,
+      "loss": 0.4829,
+      "step": 9259
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372229178086704e-05,
+      "loss": 0.4893,
+      "step": 9260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721002120667617e-05,
+      "loss": 0.5143,
+      "step": 9261
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371971238862948e-05,
+      "loss": 0.4851,
+      "step": 9262
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3718422584777533e-05,
+      "loss": 0.5029,
+      "step": 9263
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3717132709136673e-05,
+      "loss": 0.4955,
+      "step": 9264
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371584276173181e-05,
+      "loss": 0.4976,
+      "step": 9265
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714552742587838e-05,
+      "loss": 0.4975,
+      "step": 9266
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3713262651729671e-05,
+      "loss": 0.4763,
+      "step": 9267
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3711972489182208e-05,
+      "loss": 0.4563,
+      "step": 9268
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710682254970366e-05,
+      "loss": 0.5049,
+      "step": 9269
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3709391949119047e-05,
+      "loss": 0.4782,
+      "step": 9270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3708101571653167e-05,
+      "loss": 0.4808,
+      "step": 9271
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3706811122597634e-05,
+      "loss": 0.4981,
+      "step": 9272
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3705520601977368e-05,
+      "loss": 0.4918,
+      "step": 9273
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3704230009817277e-05,
+      "loss": 0.5005,
+      "step": 9274
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702939346142286e-05,
+      "loss": 0.5022,
+      "step": 9275
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3701648610977303e-05,
+      "loss": 0.4803,
+      "step": 9276
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700357804347253e-05,
+      "loss": 0.4805,
+      "step": 9277
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3699066926277056e-05,
+      "loss": 0.5067,
+      "step": 9278
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697775976791637e-05,
+      "loss": 0.4987,
+      "step": 9279
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696484955915917e-05,
+      "loss": 0.4432,
+      "step": 9280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3695193863674813e-05,
+      "loss": 0.4692,
+      "step": 9281
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3693902700093265e-05,
+      "loss": 0.5011,
+      "step": 9282
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369261146519619e-05,
+      "loss": 0.4945,
+      "step": 9283
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3691320159008525e-05,
+      "loss": 0.5009,
+      "step": 9284
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369002878155519e-05,
+      "loss": 0.5202,
+      "step": 9285
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368873733286113e-05,
+      "loss": 0.5222,
+      "step": 9286
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3687445812951267e-05,
+      "loss": 0.4901,
+      "step": 9287
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368615422185054e-05,
+      "loss": 0.505,
+      "step": 9288
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3684862559583885e-05,
+      "loss": 0.5293,
+      "step": 9289
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368357082617624e-05,
+      "loss": 0.5092,
+      "step": 9290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682279021652541e-05,
+      "loss": 0.4887,
+      "step": 9291
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3680987146037726e-05,
+      "loss": 0.4846,
+      "step": 9292
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679695199356741e-05,
+      "loss": 0.5039,
+      "step": 9293
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3678403181634529e-05,
+      "loss": 0.4847,
+      "step": 9294
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367711109289603e-05,
+      "loss": 0.4895,
+      "step": 9295
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675818933166192e-05,
+      "loss": 0.5017,
+      "step": 9296
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674526702469958e-05,
+      "loss": 0.5009,
+      "step": 9297
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673234400832287e-05,
+      "loss": 0.477,
+      "step": 9298
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3671942028278116e-05,
+      "loss": 0.4893,
+      "step": 9299
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36706495848324e-05,
+      "loss": 0.5135,
+      "step": 9300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3669357070520093e-05,
+      "loss": 0.4849,
+      "step": 9301
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366806448536615e-05,
+      "loss": 0.487,
+      "step": 9302
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3666771829395522e-05,
+      "loss": 0.507,
+      "step": 9303
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3665479102633169e-05,
+      "loss": 0.4553,
+      "step": 9304
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3664186305104046e-05,
+      "loss": 0.4899,
+      "step": 9305
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366289343683311e-05,
+      "loss": 0.5003,
+      "step": 9306
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3661600497845329e-05,
+      "loss": 0.5117,
+      "step": 9307
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3660307488165656e-05,
+      "loss": 0.4615,
+      "step": 9308
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659014407819062e-05,
+      "loss": 0.4917,
+      "step": 9309
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657721256830507e-05,
+      "loss": 0.4987,
+      "step": 9310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3656428035224958e-05,
+      "loss": 0.4853,
+      "step": 9311
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3655134743027381e-05,
+      "loss": 0.5087,
+      "step": 9312
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653841380262748e-05,
+      "loss": 0.4749,
+      "step": 9313
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3652547946956027e-05,
+      "loss": 0.4951,
+      "step": 9314
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365125444313219e-05,
+      "loss": 0.4797,
+      "step": 9315
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3649960868816207e-05,
+      "loss": 0.479,
+      "step": 9316
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3648667224033057e-05,
+      "loss": 0.4824,
+      "step": 9317
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647373508807711e-05,
+      "loss": 0.4778,
+      "step": 9318
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3646079723165148e-05,
+      "loss": 0.4686,
+      "step": 9319
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3644785867130343e-05,
+      "loss": 0.4878,
+      "step": 9320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643491940728284e-05,
+      "loss": 0.4908,
+      "step": 9321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3642197943983942e-05,
+      "loss": 0.4812,
+      "step": 9322
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640903876922304e-05,
+      "loss": 0.4766,
+      "step": 9323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3639609739568358e-05,
+      "loss": 0.4703,
+      "step": 9324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363831553194708e-05,
+      "loss": 0.4831,
+      "step": 9325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363702125408346e-05,
+      "loss": 0.4562,
+      "step": 9326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635726906002485e-05,
+      "loss": 0.1844,
+      "step": 9327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3634432487729147e-05,
+      "loss": 0.4805,
+      "step": 9328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633137999288435e-05,
+      "loss": 0.4871,
+      "step": 9329
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3631843440705338e-05,
+      "loss": 0.509,
+      "step": 9330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630548812004856e-05,
+      "loss": 0.4904,
+      "step": 9331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629254113211977e-05,
+      "loss": 0.491,
+      "step": 9332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3627959344351695e-05,
+      "loss": 0.4679,
+      "step": 9333
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626664505449014e-05,
+      "loss": 0.5089,
+      "step": 9334
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362536959652893e-05,
+      "loss": 0.4865,
+      "step": 9335
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624074617616443e-05,
+      "loss": 0.4734,
+      "step": 9336
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622779568736551e-05,
+      "loss": 0.4994,
+      "step": 9337
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3621484449914262e-05,
+      "loss": 0.5067,
+      "step": 9338
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620189261174575e-05,
+      "loss": 0.5165,
+      "step": 9339
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36188940025425e-05,
+      "loss": 0.4866,
+      "step": 9340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3617598674043035e-05,
+      "loss": 0.4771,
+      "step": 9341
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36163032757012e-05,
+      "loss": 0.5123,
+      "step": 9342
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3615007807541996e-05,
+      "loss": 0.4935,
+      "step": 9343
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3613712269590434e-05,
+      "loss": 0.5034,
+      "step": 9344
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3612416661871532e-05,
+      "loss": 0.4677,
+      "step": 9345
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3611120984410296e-05,
+      "loss": 0.4842,
+      "step": 9346
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609825237231746e-05,
+      "loss": 0.4821,
+      "step": 9347
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3608529420360894e-05,
+      "loss": 0.4804,
+      "step": 9348
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3607233533822762e-05,
+      "loss": 0.1852,
+      "step": 9349
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3605937577642366e-05,
+      "loss": 0.4912,
+      "step": 9350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604641551844724e-05,
+      "loss": 0.4287,
+      "step": 9351
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360334545645486e-05,
+      "loss": 0.5395,
+      "step": 9352
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3602049291497798e-05,
+      "loss": 0.5171,
+      "step": 9353
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3600753056998561e-05,
+      "loss": 0.4904,
+      "step": 9354
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599456752982172e-05,
+      "loss": 0.4721,
+      "step": 9355
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3598160379473661e-05,
+      "loss": 0.5151,
+      "step": 9356
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596863936498057e-05,
+      "loss": 0.5121,
+      "step": 9357
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3595567424080386e-05,
+      "loss": 0.5075,
+      "step": 9358
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359427084224568e-05,
+      "loss": 0.4516,
+      "step": 9359
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3592974191018973e-05,
+      "loss": 0.5055,
+      "step": 9360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3591677470425297e-05,
+      "loss": 0.4848,
+      "step": 9361
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3590380680489688e-05,
+      "loss": 0.5133,
+      "step": 9362
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3589083821237177e-05,
+      "loss": 0.468,
+      "step": 9363
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358778689269281e-05,
+      "loss": 0.4945,
+      "step": 9364
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3586489894881618e-05,
+      "loss": 0.4864,
+      "step": 9365
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3585192827828648e-05,
+      "loss": 0.4747,
+      "step": 9366
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3583895691558935e-05,
+      "loss": 0.4705,
+      "step": 9367
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358259848609753e-05,
+      "loss": 0.4872,
+      "step": 9368
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3581301211469469e-05,
+      "loss": 0.5174,
+      "step": 9369
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580003867699801e-05,
+      "loss": 0.4881,
+      "step": 9370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3578706454813571e-05,
+      "loss": 0.4974,
+      "step": 9371
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3577408972835832e-05,
+      "loss": 0.4822,
+      "step": 9372
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357611142179163e-05,
+      "loss": 0.4846,
+      "step": 9373
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574813801706017e-05,
+      "loss": 0.4791,
+      "step": 9374
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573516112604042e-05,
+      "loss": 0.4834,
+      "step": 9375
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3572218354510764e-05,
+      "loss": 0.4571,
+      "step": 9376
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3570920527451235e-05,
+      "loss": 0.4413,
+      "step": 9377
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3569622631450509e-05,
+      "loss": 0.5276,
+      "step": 9378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3568324666533647e-05,
+      "loss": 0.509,
+      "step": 9379
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356702663272571e-05,
+      "loss": 0.4874,
+      "step": 9380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565728530051752e-05,
+      "loss": 0.4696,
+      "step": 9381
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3564430358536838e-05,
+      "loss": 0.4917,
+      "step": 9382
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3563132118206031e-05,
+      "loss": 0.5209,
+      "step": 9383
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561833809084392e-05,
+      "loss": 0.4774,
+      "step": 9384
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560535431196991e-05,
+      "loss": 0.4997,
+      "step": 9385
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3559236984568893e-05,
+      "loss": 0.4808,
+      "step": 9386
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3557938469225167e-05,
+      "loss": 0.5015,
+      "step": 9387
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355663988519088e-05,
+      "loss": 0.5066,
+      "step": 9388
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3555341232491106e-05,
+      "loss": 0.482,
+      "step": 9389
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554042511150913e-05,
+      "loss": 0.522,
+      "step": 9390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355274372119538e-05,
+      "loss": 0.4717,
+      "step": 9391
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551444862649576e-05,
+      "loss": 0.4853,
+      "step": 9392
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355014593553858e-05,
+      "loss": 0.495,
+      "step": 9393
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548846939887473e-05,
+      "loss": 0.4937,
+      "step": 9394
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3547547875721327e-05,
+      "loss": 0.4822,
+      "step": 9395
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3546248743065227e-05,
+      "loss": 0.5094,
+      "step": 9396
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544949541944252e-05,
+      "loss": 0.4899,
+      "step": 9397
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543650272383485e-05,
+      "loss": 0.4855,
+      "step": 9398
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3542350934408013e-05,
+      "loss": 0.4752,
+      "step": 9399
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541051528042915e-05,
+      "loss": 0.4944,
+      "step": 9400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3539752053313284e-05,
+      "loss": 0.4984,
+      "step": 9401
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3538452510244207e-05,
+      "loss": 0.5005,
+      "step": 9402
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353715289886077e-05,
+      "loss": 0.5002,
+      "step": 9403
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535853219188064e-05,
+      "loss": 0.4682,
+      "step": 9404
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3534553471251185e-05,
+      "loss": 0.4974,
+      "step": 9405
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3533253655075225e-05,
+      "loss": 0.4449,
+      "step": 9406
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3531953770685275e-05,
+      "loss": 0.5077,
+      "step": 9407
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3530653818106435e-05,
+      "loss": 0.495,
+      "step": 9408
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529353797363802e-05,
+      "loss": 0.4647,
+      "step": 9409
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528053708482474e-05,
+      "loss": 0.4881,
+      "step": 9410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526753551487543e-05,
+      "loss": 0.495,
+      "step": 9411
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3525453326404125e-05,
+      "loss": 0.5198,
+      "step": 9412
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3524153033257312e-05,
+      "loss": 0.5028,
+      "step": 9413
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522852672072211e-05,
+      "loss": 0.5008,
+      "step": 9414
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521552242873926e-05,
+      "loss": 0.5056,
+      "step": 9415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3520251745687567e-05,
+      "loss": 0.5119,
+      "step": 9416
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3518951180538235e-05,
+      "loss": 0.5048,
+      "step": 9417
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3517650547451044e-05,
+      "loss": 0.4797,
+      "step": 9418
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3516349846451107e-05,
+      "loss": 0.4773,
+      "step": 9419
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3515049077563531e-05,
+      "loss": 0.5038,
+      "step": 9420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3513748240813429e-05,
+      "loss": 0.4743,
+      "step": 9421
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512447336225915e-05,
+      "loss": 0.5017,
+      "step": 9422
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3511146363826109e-05,
+      "loss": 0.4972,
+      "step": 9423
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3509845323639126e-05,
+      "loss": 0.4846,
+      "step": 9424
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350854421569008e-05,
+      "loss": 0.492,
+      "step": 9425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3507243040004095e-05,
+      "loss": 0.4856,
+      "step": 9426
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3505941796606292e-05,
+      "loss": 0.507,
+      "step": 9427
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504640485521793e-05,
+      "loss": 0.4908,
+      "step": 9428
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3503339106775716e-05,
+      "loss": 0.4851,
+      "step": 9429
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502037660393195e-05,
+      "loss": 0.4684,
+      "step": 9430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3500736146399351e-05,
+      "loss": 0.4978,
+      "step": 9431
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499434564819309e-05,
+      "loss": 0.5059,
+      "step": 9432
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3498132915678201e-05,
+      "loss": 0.4987,
+      "step": 9433
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496831199001157e-05,
+      "loss": 0.5092,
+      "step": 9434
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349552941481331e-05,
+      "loss": 0.4833,
+      "step": 9435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3494227563139786e-05,
+      "loss": 0.5072,
+      "step": 9436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492925644005725e-05,
+      "loss": 0.4919,
+      "step": 9437
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349162365743626e-05,
+      "loss": 0.4746,
+      "step": 9438
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349032160345653e-05,
+      "loss": 0.51,
+      "step": 9439
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348901948209167e-05,
+      "loss": 0.4931,
+      "step": 9440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3487717293366818e-05,
+      "loss": 0.509,
+      "step": 9441
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3486415037307117e-05,
+      "loss": 0.5217,
+      "step": 9442
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3485112713937712e-05,
+      "loss": 0.4495,
+      "step": 9443
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3483810323283736e-05,
+      "loss": 0.4977,
+      "step": 9444
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3482507865370342e-05,
+      "loss": 0.1901,
+      "step": 9445
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3481205340222673e-05,
+      "loss": 0.5001,
+      "step": 9446
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3479902747865879e-05,
+      "loss": 0.4877,
+      "step": 9447
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3478600088325103e-05,
+      "loss": 0.4584,
+      "step": 9448
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3477297361625495e-05,
+      "loss": 0.5025,
+      "step": 9449
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475994567792208e-05,
+      "loss": 0.5147,
+      "step": 9450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3474691706850396e-05,
+      "loss": 0.5109,
+      "step": 9451
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3473388778825207e-05,
+      "loss": 0.4835,
+      "step": 9452
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34720857837418e-05,
+      "loss": 0.4816,
+      "step": 9453
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347078272162533e-05,
+      "loss": 0.5251,
+      "step": 9454
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469479592500954e-05,
+      "loss": 0.4857,
+      "step": 9455
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3468176396393828e-05,
+      "loss": 0.5182,
+      "step": 9456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466873133329118e-05,
+      "loss": 0.5087,
+      "step": 9457
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3465569803331979e-05,
+      "loss": 0.5104,
+      "step": 9458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3464266406427575e-05,
+      "loss": 0.5173,
+      "step": 9459
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462962942641075e-05,
+      "loss": 0.496,
+      "step": 9460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3461659411997637e-05,
+      "loss": 0.5106,
+      "step": 9461
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3460355814522433e-05,
+      "loss": 0.4708,
+      "step": 9462
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3459052150240624e-05,
+      "loss": 0.4853,
+      "step": 9463
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3457748419177383e-05,
+      "loss": 0.4847,
+      "step": 9464
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345644462135788e-05,
+      "loss": 0.5035,
+      "step": 9465
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345514075680729e-05,
+      "loss": 0.494,
+      "step": 9466
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453836825550777e-05,
+      "loss": 0.4785,
+      "step": 9467
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452532827613526e-05,
+      "loss": 0.4837,
+      "step": 9468
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34512287630207e-05,
+      "loss": 0.1917,
+      "step": 9469
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3449924631797485e-05,
+      "loss": 0.4909,
+      "step": 9470
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3448620433969057e-05,
+      "loss": 0.4869,
+      "step": 9471
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447316169560593e-05,
+      "loss": 0.4747,
+      "step": 9472
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344601183859728e-05,
+      "loss": 0.5037,
+      "step": 9473
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3444707441104285e-05,
+      "loss": 0.4974,
+      "step": 9474
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443402977106807e-05,
+      "loss": 0.4855,
+      "step": 9475
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442098446630023e-05,
+      "loss": 0.5038,
+      "step": 9476
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344079384969912e-05,
+      "loss": 0.5022,
+      "step": 9477
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3439489186339283e-05,
+      "loss": 0.4948,
+      "step": 9478
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3438184456575703e-05,
+      "loss": 0.5064,
+      "step": 9479
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436879660433567e-05,
+      "loss": 0.5081,
+      "step": 9480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3435574797938067e-05,
+      "loss": 0.4794,
+      "step": 9481
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3434269869114396e-05,
+      "loss": 0.5306,
+      "step": 9482
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3432964873987748e-05,
+      "loss": 0.4873,
+      "step": 9483
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431659812583313e-05,
+      "loss": 0.4798,
+      "step": 9484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343035468492629e-05,
+      "loss": 0.4813,
+      "step": 9485
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3429049491041876e-05,
+      "loss": 0.5025,
+      "step": 9486
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342774423095527e-05,
+      "loss": 0.488,
+      "step": 9487
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3426438904691672e-05,
+      "loss": 0.508,
+      "step": 9488
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425133512276284e-05,
+      "loss": 0.5206,
+      "step": 9489
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34238280537343e-05,
+      "loss": 0.4808,
+      "step": 9490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422522529090935e-05,
+      "loss": 0.47,
+      "step": 9491
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3421216938371388e-05,
+      "loss": 0.5066,
+      "step": 9492
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3419911281600865e-05,
+      "loss": 0.4977,
+      "step": 9493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3418605558804577e-05,
+      "loss": 0.5241,
+      "step": 9494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417299770007728e-05,
+      "loss": 0.4521,
+      "step": 9495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3415993915235528e-05,
+      "loss": 0.4938,
+      "step": 9496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3414687994513191e-05,
+      "loss": 0.5051,
+      "step": 9497
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341338200786593e-05,
+      "loss": 0.5049,
+      "step": 9498
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3412075955318955e-05,
+      "loss": 0.5187,
+      "step": 9499
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410769836897485e-05,
+      "loss": 0.5126,
+      "step": 9500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3409463652626732e-05,
+      "loss": 0.502,
+      "step": 9501
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340815740253192e-05,
+      "loss": 0.4691,
+      "step": 9502
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3406851086638261e-05,
+      "loss": 0.5148,
+      "step": 9503
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3405544704970981e-05,
+      "loss": 0.4877,
+      "step": 9504
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3404238257555296e-05,
+      "loss": 0.4659,
+      "step": 9505
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402931744416432e-05,
+      "loss": 0.5124,
+      "step": 9506
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401625165579612e-05,
+      "loss": 0.4791,
+      "step": 9507
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3400318521070061e-05,
+      "loss": 0.5084,
+      "step": 9508
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3399011810913004e-05,
+      "loss": 0.4748,
+      "step": 9509
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397705035133672e-05,
+      "loss": 0.4997,
+      "step": 9510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3396398193757291e-05,
+      "loss": 0.5089,
+      "step": 9511
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3395091286809095e-05,
+      "loss": 0.4902,
+      "step": 9512
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3393784314314311e-05,
+      "loss": 0.4927,
+      "step": 9513
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392477276298176e-05,
+      "loss": 0.4978,
+      "step": 9514
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391170172785918e-05,
+      "loss": 0.4813,
+      "step": 9515
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389863003802779e-05,
+      "loss": 0.4793,
+      "step": 9516
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3388555769373995e-05,
+      "loss": 0.4957,
+      "step": 9517
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3387248469524799e-05,
+      "loss": 0.4567,
+      "step": 9518
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338594110428043e-05,
+      "loss": 0.5005,
+      "step": 9519
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3384633673666132e-05,
+      "loss": 0.5198,
+      "step": 9520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338332617770715e-05,
+      "loss": 0.4897,
+      "step": 9521
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382018616428716e-05,
+      "loss": 0.4853,
+      "step": 9522
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3380710989856086e-05,
+      "loss": 0.4925,
+      "step": 9523
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33794032980145e-05,
+      "loss": 0.4663,
+      "step": 9524
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3378095540929204e-05,
+      "loss": 0.5051,
+      "step": 9525
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3376787718625444e-05,
+      "loss": 0.5079,
+      "step": 9526
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3375479831128476e-05,
+      "loss": 0.4885,
+      "step": 9527
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374171878463542e-05,
+      "loss": 0.4859,
+      "step": 9528
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3372863860655903e-05,
+      "loss": 0.5262,
+      "step": 9529
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33715557777308e-05,
+      "loss": 0.4971,
+      "step": 9530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33702476297135e-05,
+      "loss": 0.506,
+      "step": 9531
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336893941662925e-05,
+      "loss": 0.4973,
+      "step": 9532
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336763113850331e-05,
+      "loss": 0.4786,
+      "step": 9533
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3366322795360937e-05,
+      "loss": 0.4888,
+      "step": 9534
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3365014387227393e-05,
+      "loss": 0.4655,
+      "step": 9535
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3363705914127937e-05,
+      "loss": 0.476,
+      "step": 9536
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3362397376087825e-05,
+      "loss": 0.4913,
+      "step": 9537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336108877313233e-05,
+      "loss": 0.4964,
+      "step": 9538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3359780105286706e-05,
+      "loss": 0.5042,
+      "step": 9539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3358471372576229e-05,
+      "loss": 0.484,
+      "step": 9540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3357162575026154e-05,
+      "loss": 0.4648,
+      "step": 9541
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355853712661762e-05,
+      "loss": 0.5282,
+      "step": 9542
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335454478550831e-05,
+      "loss": 0.4895,
+      "step": 9543
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3353235793591077e-05,
+      "loss": 0.5022,
+      "step": 9544
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3351926736935331e-05,
+      "loss": 0.5041,
+      "step": 9545
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350617615566348e-05,
+      "loss": 0.5097,
+      "step": 9546
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349308429509397e-05,
+      "loss": 0.5274,
+      "step": 9547
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3347999178789758e-05,
+      "loss": 0.48,
+      "step": 9548
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3346689863432708e-05,
+      "loss": 0.1907,
+      "step": 9549
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3345380483463524e-05,
+      "loss": 0.5289,
+      "step": 9550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3344071038907483e-05,
+      "loss": 0.4796,
+      "step": 9551
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342761529789868e-05,
+      "loss": 0.4943,
+      "step": 9552
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334145195613596e-05,
+      "loss": 0.4869,
+      "step": 9553
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3340142317971042e-05,
+      "loss": 0.5081,
+      "step": 9554
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3338832615320399e-05,
+      "loss": 0.5058,
+      "step": 9555
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3337522848209313e-05,
+      "loss": 0.5241,
+      "step": 9556
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336213016663078e-05,
+      "loss": 0.5262,
+      "step": 9557
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3334903120706972e-05,
+      "loss": 0.4878,
+      "step": 9558
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3333593160366295e-05,
+      "loss": 0.5091,
+      "step": 9559
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333228313566633e-05,
+      "loss": 0.4904,
+      "step": 9560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333097304663237e-05,
+      "loss": 0.5045,
+      "step": 9561
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329662893289708e-05,
+      "loss": 0.4963,
+      "step": 9562
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332835267566364e-05,
+      "loss": 0.4929,
+      "step": 9563
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3327042393779459e-05,
+      "loss": 0.4797,
+      "step": 9564
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3325732047662467e-05,
+      "loss": 0.4826,
+      "step": 9565
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3324421637337954e-05,
+      "loss": 0.5178,
+      "step": 9566
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3323111162831226e-05,
+      "loss": 0.4704,
+      "step": 9567
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3321800624167578e-05,
+      "loss": 0.5008,
+      "step": 9568
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3320490021372318e-05,
+      "loss": 0.4636,
+      "step": 9569
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331917935447074e-05,
+      "loss": 0.4856,
+      "step": 9570
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317868623488156e-05,
+      "loss": 0.4938,
+      "step": 9571
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3316557828449869e-05,
+      "loss": 0.4952,
+      "step": 9572
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315246969381186e-05,
+      "loss": 0.4718,
+      "step": 9573
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3313936046307411e-05,
+      "loss": 0.5067,
+      "step": 9574
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312625059253858e-05,
+      "loss": 0.4736,
+      "step": 9575
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3311314008245833e-05,
+      "loss": 0.5235,
+      "step": 9576
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3310002893308653e-05,
+      "loss": 0.4836,
+      "step": 9577
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3308691714467626e-05,
+      "loss": 0.4676,
+      "step": 9578
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307380471748065e-05,
+      "loss": 0.4607,
+      "step": 9579
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330606916517529e-05,
+      "loss": 0.4955,
+      "step": 9580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3304757794774617e-05,
+      "loss": 0.469,
+      "step": 9581
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3303446360571358e-05,
+      "loss": 0.49,
+      "step": 9582
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302134862590836e-05,
+      "loss": 0.5042,
+      "step": 9583
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300823300858374e-05,
+      "loss": 0.4837,
+      "step": 9584
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299511675399288e-05,
+      "loss": 0.4788,
+      "step": 9585
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3298199986238902e-05,
+      "loss": 0.4843,
+      "step": 9586
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3296888233402543e-05,
+      "loss": 0.5046,
+      "step": 9587
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3295576416915534e-05,
+      "loss": 0.5147,
+      "step": 9588
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3294264536803196e-05,
+      "loss": 0.5111,
+      "step": 9589
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292952593090869e-05,
+      "loss": 0.1828,
+      "step": 9590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3291640585803869e-05,
+      "loss": 0.5024,
+      "step": 9591
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290328514967535e-05,
+      "loss": 0.4956,
+      "step": 9592
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3289016380607192e-05,
+      "loss": 0.4832,
+      "step": 9593
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3287704182748176e-05,
+      "loss": 0.4659,
+      "step": 9594
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3286391921415817e-05,
+      "loss": 0.4933,
+      "step": 9595
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3285079596635454e-05,
+      "loss": 0.4825,
+      "step": 9596
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3283767208432422e-05,
+      "loss": 0.4993,
+      "step": 9597
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3282454756832061e-05,
+      "loss": 0.4976,
+      "step": 9598
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3281142241859703e-05,
+      "loss": 0.4997,
+      "step": 9599
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279829663540693e-05,
+      "loss": 0.522,
+      "step": 9600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3278517021900372e-05,
+      "loss": 0.4861,
+      "step": 9601
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3277204316964076e-05,
+      "loss": 0.5074,
+      "step": 9602
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275891548757159e-05,
+      "loss": 0.4891,
+      "step": 9603
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3274578717304952e-05,
+      "loss": 0.5076,
+      "step": 9604
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3273265822632814e-05,
+      "loss": 0.4951,
+      "step": 9605
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3271952864766085e-05,
+      "loss": 0.4508,
+      "step": 9606
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270639843730115e-05,
+      "loss": 0.4491,
+      "step": 9607
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269326759550252e-05,
+      "loss": 0.4656,
+      "step": 9608
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3268013612251851e-05,
+      "loss": 0.4811,
+      "step": 9609
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266700401860262e-05,
+      "loss": 0.4916,
+      "step": 9610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3265387128400833e-05,
+      "loss": 0.4902,
+      "step": 9611
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264073791898926e-05,
+      "loss": 0.4968,
+      "step": 9612
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262760392379895e-05,
+      "loss": 0.4928,
+      "step": 9613
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3261446929869092e-05,
+      "loss": 0.4925,
+      "step": 9614
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260133404391881e-05,
+      "loss": 0.507,
+      "step": 9615
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3258819815973617e-05,
+      "loss": 0.4743,
+      "step": 9616
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3257506164639667e-05,
+      "loss": 0.479,
+      "step": 9617
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256192450415384e-05,
+      "loss": 0.4721,
+      "step": 9618
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3254878673326135e-05,
+      "loss": 0.4909,
+      "step": 9619
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253564833397283e-05,
+      "loss": 0.5047,
+      "step": 9620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32522509306542e-05,
+      "loss": 0.4547,
+      "step": 9621
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250936965122243e-05,
+      "loss": 0.5,
+      "step": 9622
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3249622936826782e-05,
+      "loss": 0.5013,
+      "step": 9623
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3248308845793193e-05,
+      "loss": 0.4827,
+      "step": 9624
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3246994692046837e-05,
+      "loss": 0.4707,
+      "step": 9625
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3245680475613089e-05,
+      "loss": 0.4954,
+      "step": 9626
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244366196517324e-05,
+      "loss": 0.4525,
+      "step": 9627
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3243051854784916e-05,
+      "loss": 0.4812,
+      "step": 9628
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241737450441234e-05,
+      "loss": 0.5246,
+      "step": 9629
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324042298351166e-05,
+      "loss": 0.4929,
+      "step": 9630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323910845402157e-05,
+      "loss": 0.4884,
+      "step": 9631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237793861996345e-05,
+      "loss": 0.4876,
+      "step": 9632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236479207461358e-05,
+      "loss": 0.4579,
+      "step": 9633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3235164490441997e-05,
+      "loss": 0.5128,
+      "step": 9634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3233849710963638e-05,
+      "loss": 0.467,
+      "step": 9635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3232534869051673e-05,
+      "loss": 0.4983,
+      "step": 9636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3231219964731479e-05,
+      "loss": 0.503,
+      "step": 9637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3229904998028445e-05,
+      "loss": 0.4795,
+      "step": 9638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322858996896796e-05,
+      "loss": 0.4548,
+      "step": 9639
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3227274877575409e-05,
+      "loss": 0.4878,
+      "step": 9640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322595972387618e-05,
+      "loss": 0.4635,
+      "step": 9641
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3224644507895672e-05,
+      "loss": 0.4727,
+      "step": 9642
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322332922965927e-05,
+      "loss": 0.4797,
+      "step": 9643
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3222013889192366e-05,
+      "loss": 0.485,
+      "step": 9644
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3220698486520358e-05,
+      "loss": 0.515,
+      "step": 9645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219383021668639e-05,
+      "loss": 0.6376,
+      "step": 9646
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3218067494662609e-05,
+      "loss": 0.5065,
+      "step": 9647
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216751905527665e-05,
+      "loss": 0.4728,
+      "step": 9648
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3215436254289202e-05,
+      "loss": 0.4711,
+      "step": 9649
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3214120540972626e-05,
+      "loss": 0.4796,
+      "step": 9650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3212804765603338e-05,
+      "loss": 0.4579,
+      "step": 9651
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211488928206733e-05,
+      "loss": 0.4878,
+      "step": 9652
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3210173028808226e-05,
+      "loss": 0.4817,
+      "step": 9653
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3208857067433213e-05,
+      "loss": 0.4898,
+      "step": 9654
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207541044107109e-05,
+      "loss": 0.4671,
+      "step": 9655
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206224958855311e-05,
+      "loss": 0.5047,
+      "step": 9656
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204908811703238e-05,
+      "loss": 0.5164,
+      "step": 9657
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203592602676293e-05,
+      "loss": 0.4571,
+      "step": 9658
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320227633179989e-05,
+      "loss": 0.4894,
+      "step": 9659
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3200959999099441e-05,
+      "loss": 0.4864,
+      "step": 9660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199643604600358e-05,
+      "loss": 0.5153,
+      "step": 9661
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198327148328059e-05,
+      "loss": 0.5352,
+      "step": 9662
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3197010630307956e-05,
+      "loss": 0.5346,
+      "step": 9663
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3195694050565472e-05,
+      "loss": 0.4738,
+      "step": 9664
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3194377409126016e-05,
+      "loss": 0.4975,
+      "step": 9665
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193060706015016e-05,
+      "loss": 0.5076,
+      "step": 9666
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191743941257884e-05,
+      "loss": 0.4815,
+      "step": 9667
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3190427114880052e-05,
+      "loss": 0.4776,
+      "step": 9668
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3189110226906936e-05,
+      "loss": 0.4636,
+      "step": 9669
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3187793277363963e-05,
+      "loss": 0.4972,
+      "step": 9670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3186476266276556e-05,
+      "loss": 0.4777,
+      "step": 9671
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3185159193670148e-05,
+      "loss": 0.5055,
+      "step": 9672
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183842059570156e-05,
+      "loss": 0.5139,
+      "step": 9673
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3182524864002018e-05,
+      "loss": 0.4652,
+      "step": 9674
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.318120760699116e-05,
+      "loss": 0.4781,
+      "step": 9675
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3179890288563015e-05,
+      "loss": 0.4989,
+      "step": 9676
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3178572908743015e-05,
+      "loss": 0.4618,
+      "step": 9677
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3177255467556593e-05,
+      "loss": 0.4825,
+      "step": 9678
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3175937965029184e-05,
+      "loss": 0.5263,
+      "step": 9679
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317462040118623e-05,
+      "loss": 0.4846,
+      "step": 9680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3173302776053159e-05,
+      "loss": 0.5262,
+      "step": 9681
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317198508965541e-05,
+      "loss": 0.5193,
+      "step": 9682
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317066734201843e-05,
+      "loss": 0.5071,
+      "step": 9683
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3169349533167655e-05,
+      "loss": 0.4661,
+      "step": 9684
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3168031663128526e-05,
+      "loss": 0.4705,
+      "step": 9685
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316671373192649e-05,
+      "loss": 0.4852,
+      "step": 9686
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316539573958699e-05,
+      "loss": 0.504,
+      "step": 9687
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316407768613547e-05,
+      "loss": 0.4992,
+      "step": 9688
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162759571597377e-05,
+      "loss": 0.5054,
+      "step": 9689
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3161441395998158e-05,
+      "loss": 0.4979,
+      "step": 9690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3160123159363266e-05,
+      "loss": 0.4991,
+      "step": 9691
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158804861718148e-05,
+      "loss": 0.4824,
+      "step": 9692
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3157486503088255e-05,
+      "loss": 0.4677,
+      "step": 9693
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3156168083499041e-05,
+      "loss": 0.5222,
+      "step": 9694
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315484960297596e-05,
+      "loss": 0.493,
+      "step": 9695
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3153531061544467e-05,
+      "loss": 0.5007,
+      "step": 9696
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3152212459230019e-05,
+      "loss": 0.4635,
+      "step": 9697
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3150893796058068e-05,
+      "loss": 0.5168,
+      "step": 9698
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314957507205408e-05,
+      "loss": 0.4843,
+      "step": 9699
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3148256287243508e-05,
+      "loss": 0.4729,
+      "step": 9700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146937441651817e-05,
+      "loss": 0.4635,
+      "step": 9701
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145618535304468e-05,
+      "loss": 0.203,
+      "step": 9702
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3144299568226926e-05,
+      "loss": 0.5038,
+      "step": 9703
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314298054044465e-05,
+      "loss": 0.4621,
+      "step": 9704
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314166145198311e-05,
+      "loss": 0.4704,
+      "step": 9705
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3140342302867774e-05,
+      "loss": 0.5255,
+      "step": 9706
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3139023093124105e-05,
+      "loss": 0.5146,
+      "step": 9707
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3137703822777574e-05,
+      "loss": 0.4995,
+      "step": 9708
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136384491853655e-05,
+      "loss": 0.4732,
+      "step": 9709
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3135065100377816e-05,
+      "loss": 0.4747,
+      "step": 9710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133745648375526e-05,
+      "loss": 0.5058,
+      "step": 9711
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3132426135872263e-05,
+      "loss": 0.479,
+      "step": 9712
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31311065628935e-05,
+      "loss": 0.5138,
+      "step": 9713
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312978692946472e-05,
+      "loss": 0.475,
+      "step": 9714
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128467235611389e-05,
+      "loss": 0.4934,
+      "step": 9715
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3127147481358994e-05,
+      "loss": 0.5334,
+      "step": 9716
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125827666733009e-05,
+      "loss": 0.4901,
+      "step": 9717
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312450779175892e-05,
+      "loss": 0.5006,
+      "step": 9718
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3123187856462202e-05,
+      "loss": 0.47,
+      "step": 9719
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121867860868346e-05,
+      "loss": 0.461,
+      "step": 9720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312054780500283e-05,
+      "loss": 0.4598,
+      "step": 9721
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3119227688891142e-05,
+      "loss": 0.4904,
+      "step": 9722
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3117907512558767e-05,
+      "loss": 0.4893,
+      "step": 9723
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116587276031199e-05,
+      "loss": 0.4816,
+      "step": 9724
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115266979333917e-05,
+      "loss": 0.5147,
+      "step": 9725
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113946622492415e-05,
+      "loss": 0.4615,
+      "step": 9726
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3112626205532189e-05,
+      "loss": 0.4837,
+      "step": 9727
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111305728478724e-05,
+      "loss": 0.4792,
+      "step": 9728
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310998519135752e-05,
+      "loss": 0.4691,
+      "step": 9729
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108664594194063e-05,
+      "loss": 0.5216,
+      "step": 9730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3107343937013859e-05,
+      "loss": 0.4817,
+      "step": 9731
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3106023219842396e-05,
+      "loss": 0.4978,
+      "step": 9732
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3104702442705179e-05,
+      "loss": 0.4782,
+      "step": 9733
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3103381605627705e-05,
+      "loss": 0.5034,
+      "step": 9734
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102060708635475e-05,
+      "loss": 0.4823,
+      "step": 9735
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100739751753989e-05,
+      "loss": 0.468,
+      "step": 9736
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099418735008747e-05,
+      "loss": 0.5094,
+      "step": 9737
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309809765842526e-05,
+      "loss": 0.5103,
+      "step": 9738
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3096776522029029e-05,
+      "loss": 0.4634,
+      "step": 9739
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3095455325845559e-05,
+      "loss": 0.5127,
+      "step": 9740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3094134069900359e-05,
+      "loss": 0.4627,
+      "step": 9741
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309281275421894e-05,
+      "loss": 0.5066,
+      "step": 9742
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3091491378826809e-05,
+      "loss": 0.5058,
+      "step": 9743
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.4902,
+      "step": 9744
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088848449012454e-05,
+      "loss": 0.5171,
+      "step": 9745
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3087526894641256e-05,
+      "loss": 0.5165,
+      "step": 9746
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.30862052806614e-05,
+      "loss": 0.4672,
+      "step": 9747
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3084883607098392e-05,
+      "loss": 0.509,
+      "step": 9748
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308356187397776e-05,
+      "loss": 0.4869,
+      "step": 9749
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082240081325015e-05,
+      "loss": 0.5012,
+      "step": 9750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3080918229165675e-05,
+      "loss": 0.4751,
+      "step": 9751
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3079596317525266e-05,
+      "loss": 0.5174,
+      "step": 9752
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30782743464293e-05,
+      "loss": 0.4659,
+      "step": 9753
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.307695231590331e-05,
+      "loss": 0.4728,
+      "step": 9754
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3075630225972813e-05,
+      "loss": 0.5115,
+      "step": 9755
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3074308076663336e-05,
+      "loss": 0.4753,
+      "step": 9756
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3072985868000402e-05,
+      "loss": 0.5282,
+      "step": 9757
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3071663600009542e-05,
+      "loss": 0.4843,
+      "step": 9758
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3070341272716279e-05,
+      "loss": 0.4809,
+      "step": 9759
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069018886146146e-05,
+      "loss": 0.4734,
+      "step": 9760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3067696440324671e-05,
+      "loss": 0.4945,
+      "step": 9761
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306637393527739e-05,
+      "loss": 0.4987,
+      "step": 9762
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306505137102983e-05,
+      "loss": 0.4727,
+      "step": 9763
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3063728747607529e-05,
+      "loss": 0.4628,
+      "step": 9764
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306240606503602e-05,
+      "loss": 0.4474,
+      "step": 9765
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306108332334084e-05,
+      "loss": 0.4535,
+      "step": 9766
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305976052254752e-05,
+      "loss": 0.5019,
+      "step": 9767
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3058437662681611e-05,
+      "loss": 0.496,
+      "step": 9768
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057114743768642e-05,
+      "loss": 0.4923,
+      "step": 9769
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.305579176583416e-05,
+      "loss": 0.5001,
+      "step": 9770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3054468728903702e-05,
+      "loss": 0.4916,
+      "step": 9771
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3053145633002812e-05,
+      "loss": 0.4817,
+      "step": 9772
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3051822478157039e-05,
+      "loss": 0.4689,
+      "step": 9773
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3050499264391921e-05,
+      "loss": 0.5066,
+      "step": 9774
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049175991733006e-05,
+      "loss": 0.5079,
+      "step": 9775
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3047852660205842e-05,
+      "loss": 0.4903,
+      "step": 9776
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3046529269835983e-05,
+      "loss": 0.5128,
+      "step": 9777
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045205820648969e-05,
+      "loss": 0.483,
+      "step": 9778
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043882312670362e-05,
+      "loss": 0.4816,
+      "step": 9779
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3042558745925704e-05,
+      "loss": 0.5102,
+      "step": 9780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041235120440554e-05,
+      "loss": 0.4562,
+      "step": 9781
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039911436240462e-05,
+      "loss": 0.4771,
+      "step": 9782
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3038587693350991e-05,
+      "loss": 0.5167,
+      "step": 9783
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3037263891797688e-05,
+      "loss": 0.4847,
+      "step": 9784
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035940031606117e-05,
+      "loss": 0.5165,
+      "step": 9785
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3034616112801835e-05,
+      "loss": 0.5009,
+      "step": 9786
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033292135410403e-05,
+      "loss": 0.4744,
+      "step": 9787
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.303196809945738e-05,
+      "loss": 0.5237,
+      "step": 9788
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3030644004968327e-05,
+      "loss": 0.4805,
+      "step": 9789
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029319851968811e-05,
+      "loss": 0.5136,
+      "step": 9790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30279956404844e-05,
+      "loss": 0.4863,
+      "step": 9791
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302667137054065e-05,
+      "loss": 0.4932,
+      "step": 9792
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3025347042163132e-05,
+      "loss": 0.4928,
+      "step": 9793
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3024022655377417e-05,
+      "loss": 0.4917,
+      "step": 9794
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3022698210209069e-05,
+      "loss": 0.4996,
+      "step": 9795
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302137370668366e-05,
+      "loss": 0.4922,
+      "step": 9796
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3020049144826764e-05,
+      "loss": 0.4937,
+      "step": 9797
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3018724524663951e-05,
+      "loss": 0.4877,
+      "step": 9798
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301739984622079e-05,
+      "loss": 0.509,
+      "step": 9799
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016075109522865e-05,
+      "loss": 0.4809,
+      "step": 9800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014750314595744e-05,
+      "loss": 0.4707,
+      "step": 9801
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301342546146501e-05,
+      "loss": 0.4729,
+      "step": 9802
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3012100550156234e-05,
+      "loss": 0.4698,
+      "step": 9803
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3010775580694998e-05,
+      "loss": 0.4796,
+      "step": 9804
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3009450553106886e-05,
+      "loss": 0.4818,
+      "step": 9805
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3008125467417478e-05,
+      "loss": 0.4849,
+      "step": 9806
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300680032365235e-05,
+      "loss": 0.5034,
+      "step": 9807
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3005475121837095e-05,
+      "loss": 0.5062,
+      "step": 9808
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300414986199729e-05,
+      "loss": 0.5128,
+      "step": 9809
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3002824544158525e-05,
+      "loss": 0.4906,
+      "step": 9810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001499168346387e-05,
+      "loss": 0.5418,
+      "step": 9811
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000173734586461e-05,
+      "loss": 0.4836,
+      "step": 9812
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2998848242904341e-05,
+      "loss": 0.5057,
+      "step": 9813
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997522693325613e-05,
+      "loss": 0.4953,
+      "step": 9814
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996197085875867e-05,
+      "loss": 0.514,
+      "step": 9815
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2994871420580702e-05,
+      "loss": 0.4952,
+      "step": 9816
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299354569746571e-05,
+      "loss": 0.513,
+      "step": 9817
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2992219916556481e-05,
+      "loss": 0.4765,
+      "step": 9818
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2990894077878613e-05,
+      "loss": 0.4939,
+      "step": 9819
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2989568181457704e-05,
+      "loss": 0.4984,
+      "step": 9820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988242227319354e-05,
+      "loss": 0.5042,
+      "step": 9821
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986916215489161e-05,
+      "loss": 0.4869,
+      "step": 9822
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2985590145992721e-05,
+      "loss": 0.4758,
+      "step": 9823
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2984264018855641e-05,
+      "loss": 0.4861,
+      "step": 9824
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2982937834103523e-05,
+      "loss": 0.485,
+      "step": 9825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981611591761969e-05,
+      "loss": 0.5014,
+      "step": 9826
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2980285291856582e-05,
+      "loss": 0.4974,
+      "step": 9827
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.297895893441297e-05,
+      "loss": 0.4926,
+      "step": 9828
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2977632519456745e-05,
+      "loss": 0.5145,
+      "step": 9829
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2976306047013508e-05,
+      "loss": 0.1747,
+      "step": 9830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2974979517108867e-05,
+      "loss": 0.4475,
+      "step": 9831
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973652929768438e-05,
+      "loss": 0.4988,
+      "step": 9832
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2972326285017834e-05,
+      "loss": 0.4651,
+      "step": 9833
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2970999582882657e-05,
+      "loss": 0.4989,
+      "step": 9834
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2969672823388533e-05,
+      "loss": 0.4744,
+      "step": 9835
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296834600656107e-05,
+      "loss": 0.4875,
+      "step": 9836
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2967019132425885e-05,
+      "loss": 0.4976,
+      "step": 9837
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2965692201008593e-05,
+      "loss": 0.4828,
+      "step": 9838
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2964365212334817e-05,
+      "loss": 0.4854,
+      "step": 9839
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963038166430173e-05,
+      "loss": 0.4776,
+      "step": 9840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2961711063320282e-05,
+      "loss": 0.5125,
+      "step": 9841
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2960383903030764e-05,
+      "loss": 0.471,
+      "step": 9842
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959056685587245e-05,
+      "loss": 0.5006,
+      "step": 9843
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2957729411015347e-05,
+      "loss": 0.4698,
+      "step": 9844
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295640207934069e-05,
+      "loss": 0.4757,
+      "step": 9845
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295507469058891e-05,
+      "loss": 0.5062,
+      "step": 9846
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953747244785623e-05,
+      "loss": 0.5106,
+      "step": 9847
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2952419741956467e-05,
+      "loss": 0.4611,
+      "step": 9848
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2951092182127063e-05,
+      "loss": 0.5102,
+      "step": 9849
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2949764565323046e-05,
+      "loss": 0.4803,
+      "step": 9850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948436891570044e-05,
+      "loss": 0.5226,
+      "step": 9851
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2947109160893691e-05,
+      "loss": 0.5165,
+      "step": 9852
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2945781373319621e-05,
+      "loss": 0.4623,
+      "step": 9853
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294445352887347e-05,
+      "loss": 0.4944,
+      "step": 9854
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943125627580875e-05,
+      "loss": 0.5123,
+      "step": 9855
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2941797669467464e-05,
+      "loss": 0.46,
+      "step": 9856
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940469654558885e-05,
+      "loss": 0.5016,
+      "step": 9857
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293914158288077e-05,
+      "loss": 0.4972,
+      "step": 9858
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2937813454458766e-05,
+      "loss": 0.4995,
+      "step": 9859
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936485269318504e-05,
+      "loss": 0.4879,
+      "step": 9860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2935157027485638e-05,
+      "loss": 0.4928,
+      "step": 9861
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2933828728985804e-05,
+      "loss": 0.4592,
+      "step": 9862
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293250037384465e-05,
+      "loss": 0.4985,
+      "step": 9863
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293117196208782e-05,
+      "loss": 0.4967,
+      "step": 9864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2929843493740959e-05,
+      "loss": 0.4908,
+      "step": 9865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2928514968829717e-05,
+      "loss": 0.478,
+      "step": 9866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2927186387379741e-05,
+      "loss": 0.5184,
+      "step": 9867
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925857749416686e-05,
+      "loss": 0.4836,
+      "step": 9868
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29245290549662e-05,
+      "loss": 0.5076,
+      "step": 9869
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2923200304053932e-05,
+      "loss": 0.4983,
+      "step": 9870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2921871496705538e-05,
+      "loss": 0.4951,
+      "step": 9871
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920542632946672e-05,
+      "loss": 0.4863,
+      "step": 9872
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2919213712802991e-05,
+      "loss": 0.4845,
+      "step": 9873
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291788473630015e-05,
+      "loss": 0.4924,
+      "step": 9874
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916555703463805e-05,
+      "loss": 0.4582,
+      "step": 9875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915226614319621e-05,
+      "loss": 0.4772,
+      "step": 9876
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2913897468893249e-05,
+      "loss": 0.4807,
+      "step": 9877
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2912568267210355e-05,
+      "loss": 0.4679,
+      "step": 9878
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29112390092966e-05,
+      "loss": 0.5026,
+      "step": 9879
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2909909695177647e-05,
+      "loss": 0.4889,
+      "step": 9880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290858032487916e-05,
+      "loss": 0.4896,
+      "step": 9881
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2907250898426804e-05,
+      "loss": 0.4524,
+      "step": 9882
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905921415846245e-05,
+      "loss": 0.4966,
+      "step": 9883
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2904591877163153e-05,
+      "loss": 0.4843,
+      "step": 9884
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2903262282403195e-05,
+      "loss": 0.5017,
+      "step": 9885
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2901932631592038e-05,
+      "loss": 0.491,
+      "step": 9886
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2900602924755354e-05,
+      "loss": 0.4802,
+      "step": 9887
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899273161918818e-05,
+      "loss": 0.5053,
+      "step": 9888
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897943343108097e-05,
+      "loss": 0.5005,
+      "step": 9889
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896613468348869e-05,
+      "loss": 0.5049,
+      "step": 9890
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2895283537666808e-05,
+      "loss": 0.4516,
+      "step": 9891
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289395355108759e-05,
+      "loss": 0.4432,
+      "step": 9892
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2892623508636892e-05,
+      "loss": 0.5281,
+      "step": 9893
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289129341034039e-05,
+      "loss": 0.5143,
+      "step": 9894
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2889963256223767e-05,
+      "loss": 0.4579,
+      "step": 9895
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888633046312705e-05,
+      "loss": 0.5158,
+      "step": 9896
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2887302780632876e-05,
+      "loss": 0.4913,
+      "step": 9897
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2885972459209973e-05,
+      "loss": 0.4745,
+      "step": 9898
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2884642082069675e-05,
+      "loss": 0.4923,
+      "step": 9899
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883311649237665e-05,
+      "loss": 0.4871,
+      "step": 9900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2881981160739629e-05,
+      "loss": 0.4591,
+      "step": 9901
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2880650616601256e-05,
+      "loss": 0.479,
+      "step": 9902
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2879320016848236e-05,
+      "loss": 0.4872,
+      "step": 9903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2877989361506251e-05,
+      "loss": 0.5002,
+      "step": 9904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876658650600999e-05,
+      "loss": 0.4581,
+      "step": 9905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2875327884158165e-05,
+      "loss": 0.5256,
+      "step": 9906
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2873997062203444e-05,
+      "loss": 0.4826,
+      "step": 9907
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2872666184762528e-05,
+      "loss": 0.4801,
+      "step": 9908
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2871335251861112e-05,
+      "loss": 0.4893,
+      "step": 9909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.287000426352489e-05,
+      "loss": 0.4886,
+      "step": 9910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2868673219779563e-05,
+      "loss": 0.4737,
+      "step": 9911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.286734212065082e-05,
+      "loss": 0.4842,
+      "step": 9912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866010966164371e-05,
+      "loss": 0.4847,
+      "step": 9913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2864679756345905e-05,
+      "loss": 0.5066,
+      "step": 9914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2863348491221129e-05,
+      "loss": 0.4875,
+      "step": 9915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862017170815739e-05,
+      "loss": 0.4958,
+      "step": 9916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2860685795155448e-05,
+      "loss": 0.4987,
+      "step": 9917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2859354364265948e-05,
+      "loss": 0.4866,
+      "step": 9918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2858022878172951e-05,
+      "loss": 0.4867,
+      "step": 9919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856691336902163e-05,
+      "loss": 0.4819,
+      "step": 9920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285535974047929e-05,
+      "loss": 0.4857,
+      "step": 9921
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854028088930039e-05,
+      "loss": 0.4632,
+      "step": 9922
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2852696382280118e-05,
+      "loss": 0.4502,
+      "step": 9923
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2851364620555241e-05,
+      "loss": 0.4743,
+      "step": 9924
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2850032803781119e-05,
+      "loss": 0.5039,
+      "step": 9925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848700931983462e-05,
+      "loss": 0.4686,
+      "step": 9926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2847369005187984e-05,
+      "loss": 0.4802,
+      "step": 9927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2846037023420403e-05,
+      "loss": 0.4894,
+      "step": 9928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2844704986706429e-05,
+      "loss": 0.4837,
+      "step": 9929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2843372895071782e-05,
+      "loss": 0.464,
+      "step": 9930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284204074854218e-05,
+      "loss": 0.5121,
+      "step": 9931
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2840708547143342e-05,
+      "loss": 0.4664,
+      "step": 9932
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839376290900984e-05,
+      "loss": 0.4723,
+      "step": 9933
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2838043979840833e-05,
+      "loss": 0.4527,
+      "step": 9934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836711613988607e-05,
+      "loss": 0.4863,
+      "step": 9935
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283537919337003e-05,
+      "loss": 0.4696,
+      "step": 9936
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2834046718010824e-05,
+      "loss": 0.5223,
+      "step": 9937
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2832714187936718e-05,
+      "loss": 0.5072,
+      "step": 9938
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2831381603173434e-05,
+      "loss": 0.4406,
+      "step": 9939
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830048963746703e-05,
+      "loss": 0.473,
+      "step": 9940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2828716269682253e-05,
+      "loss": 0.4983,
+      "step": 9941
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2827383521005809e-05,
+      "loss": 0.4805,
+      "step": 9942
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2826050717743111e-05,
+      "loss": 0.4533,
+      "step": 9943
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824717859919877e-05,
+      "loss": 0.4833,
+      "step": 9944
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282338494756185e-05,
+      "loss": 0.4822,
+      "step": 9945
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822051980694761e-05,
+      "loss": 0.5204,
+      "step": 9946
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820718959344342e-05,
+      "loss": 0.4763,
+      "step": 9947
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819385883536332e-05,
+      "loss": 0.4952,
+      "step": 9948
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2818052753296463e-05,
+      "loss": 0.5187,
+      "step": 9949
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281671956865048e-05,
+      "loss": 0.4879,
+      "step": 9950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2815386329624118e-05,
+      "loss": 0.4488,
+      "step": 9951
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2814053036243115e-05,
+      "loss": 0.4886,
+      "step": 9952
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2812719688533214e-05,
+      "loss": 0.5201,
+      "step": 9953
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2811386286520154e-05,
+      "loss": 0.507,
+      "step": 9954
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810052830229684e-05,
+      "loss": 0.4704,
+      "step": 9955
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2808719319687544e-05,
+      "loss": 0.4713,
+      "step": 9956
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807385754919479e-05,
+      "loss": 0.5059,
+      "step": 9957
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2806052135951237e-05,
+      "loss": 0.4638,
+      "step": 9958
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804718462808566e-05,
+      "loss": 0.4824,
+      "step": 9959
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2803384735517209e-05,
+      "loss": 0.4776,
+      "step": 9960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2802050954102923e-05,
+      "loss": 0.4865,
+      "step": 9961
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2800717118591452e-05,
+      "loss": 0.5007,
+      "step": 9962
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279938322900855e-05,
+      "loss": 0.5123,
+      "step": 9963
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798049285379968e-05,
+      "loss": 0.4832,
+      "step": 9964
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2796715287731463e-05,
+      "loss": 0.5005,
+      "step": 9965
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2795381236088786e-05,
+      "loss": 0.5058,
+      "step": 9966
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2794047130477694e-05,
+      "loss": 0.5256,
+      "step": 9967
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2792712970923942e-05,
+      "loss": 0.4549,
+      "step": 9968
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791378757453288e-05,
+      "loss": 0.4545,
+      "step": 9969
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790044490091496e-05,
+      "loss": 0.487,
+      "step": 9970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2788710168864316e-05,
+      "loss": 0.4688,
+      "step": 9971
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2787375793797518e-05,
+      "loss": 0.4983,
+      "step": 9972
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2786041364916859e-05,
+      "loss": 0.4864,
+      "step": 9973
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2784706882248106e-05,
+      "loss": 0.4974,
+      "step": 9974
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2783372345817013e-05,
+      "loss": 0.5183,
+      "step": 9975
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278203775564936e-05,
+      "loss": 0.49,
+      "step": 9976
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780703111770898e-05,
+      "loss": 0.4585,
+      "step": 9977
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2779368414207403e-05,
+      "loss": 0.5172,
+      "step": 9978
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2778033662984638e-05,
+      "loss": 0.4638,
+      "step": 9979
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2776698858128383e-05,
+      "loss": 0.4946,
+      "step": 9980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2775363999664392e-05,
+      "loss": 0.4854,
+      "step": 9981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2774029087618448e-05,
+      "loss": 0.4912,
+      "step": 9982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2772694122016318e-05,
+      "loss": 0.4954,
+      "step": 9983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2771359102883775e-05,
+      "loss": 0.4638,
+      "step": 9984
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27700240302466e-05,
+      "loss": 0.49,
+      "step": 9985
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2768688904130558e-05,
+      "loss": 0.5015,
+      "step": 9986
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2767353724561431e-05,
+      "loss": 0.5252,
+      "step": 9987
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2766018491564997e-05,
+      "loss": 0.4598,
+      "step": 9988
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2764683205167031e-05,
+      "loss": 0.4735,
+      "step": 9989
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2763347865393317e-05,
+      "loss": 0.4724,
+      "step": 9990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2762012472269634e-05,
+      "loss": 0.4992,
+      "step": 9991
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2760677025821759e-05,
+      "loss": 0.4896,
+      "step": 9992
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759341526075479e-05,
+      "loss": 0.5101,
+      "step": 9993
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2758005973056577e-05,
+      "loss": 0.5067,
+      "step": 9994
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2756670366790837e-05,
+      "loss": 0.5011,
+      "step": 9995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2755334707304045e-05,
+      "loss": 0.4995,
+      "step": 9996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2753998994621987e-05,
+      "loss": 0.5176,
+      "step": 9997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275266322877045e-05,
+      "loss": 0.4939,
+      "step": 9998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2751327409775227e-05,
+      "loss": 0.5232,
+      "step": 9999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27499915376621e-05,
+      "loss": 0.4929,
+      "step": 10000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2748655612456867e-05,
+      "loss": 0.4818,
+      "step": 10001
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2747319634185314e-05,
+      "loss": 0.4677,
+      "step": 10002
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274598360287324e-05,
+      "loss": 0.4769,
+      "step": 10003
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2744647518546433e-05,
+      "loss": 0.4601,
+      "step": 10004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274331138123069e-05,
+      "loss": 0.4756,
+      "step": 10005
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274197519095181e-05,
+      "loss": 0.5033,
+      "step": 10006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2740638947735583e-05,
+      "loss": 0.4763,
+      "step": 10007
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273930265160781e-05,
+      "loss": 0.5145,
+      "step": 10008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737966302594293e-05,
+      "loss": 0.4498,
+      "step": 10009
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 0.4829,
+      "step": 10010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2735293446013222e-05,
+      "loss": 0.5147,
+      "step": 10011
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733956938497267e-05,
+      "loss": 0.4961,
+      "step": 10012
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732620378198775e-05,
+      "loss": 0.459,
+      "step": 10013
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2731283765143545e-05,
+      "loss": 0.4781,
+      "step": 10014
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2729947099357385e-05,
+      "loss": 0.4773,
+      "step": 10015
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2728610380866097e-05,
+      "loss": 0.4933,
+      "step": 10016
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2727273609695492e-05,
+      "loss": 0.4931,
+      "step": 10017
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2725936785871378e-05,
+      "loss": 0.4896,
+      "step": 10018
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2724599909419559e-05,
+      "loss": 0.5169,
+      "step": 10019
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2723262980365852e-05,
+      "loss": 0.4851,
+      "step": 10020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2721925998736063e-05,
+      "loss": 0.5145,
+      "step": 10021
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272058896455601e-05,
+      "loss": 0.5026,
+      "step": 10022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719251877851495e-05,
+      "loss": 0.4991,
+      "step": 10023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717914738648347e-05,
+      "loss": 0.4777,
+      "step": 10024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2716577546972371e-05,
+      "loss": 0.5013,
+      "step": 10025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2715240302849386e-05,
+      "loss": 0.4663,
+      "step": 10026
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2713903006305208e-05,
+      "loss": 0.4868,
+      "step": 10027
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2712565657365657e-05,
+      "loss": 0.4907,
+      "step": 10028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2711228256056553e-05,
+      "loss": 0.5086,
+      "step": 10029
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709890802403712e-05,
+      "loss": 0.4864,
+      "step": 10030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270855329643296e-05,
+      "loss": 0.4961,
+      "step": 10031
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2707215738170118e-05,
+      "loss": 0.4968,
+      "step": 10032
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2705878127641007e-05,
+      "loss": 0.4927,
+      "step": 10033
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270454046487145e-05,
+      "loss": 0.4869,
+      "step": 10034
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703202749887282e-05,
+      "loss": 0.481,
+      "step": 10035
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2701864982714318e-05,
+      "loss": 0.4668,
+      "step": 10036
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2700527163378392e-05,
+      "loss": 0.4975,
+      "step": 10037
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2699189291905328e-05,
+      "loss": 0.5196,
+      "step": 10038
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2697851368320958e-05,
+      "loss": 0.4863,
+      "step": 10039
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2696513392651111e-05,
+      "loss": 0.4696,
+      "step": 10040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269517536492162e-05,
+      "loss": 0.5101,
+      "step": 10041
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2693837285158317e-05,
+      "loss": 0.4878,
+      "step": 10042
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692499153387035e-05,
+      "loss": 0.5032,
+      "step": 10043
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2691160969633608e-05,
+      "loss": 0.4859,
+      "step": 10044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268982273392387e-05,
+      "loss": 0.484,
+      "step": 10045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2688484446283657e-05,
+      "loss": 0.5055,
+      "step": 10046
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2687146106738815e-05,
+      "loss": 0.5266,
+      "step": 10047
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268580771531517e-05,
+      "loss": 0.4812,
+      "step": 10048
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684469272038567e-05,
+      "loss": 0.4999,
+      "step": 10049
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268313077693485e-05,
+      "loss": 0.4375,
+      "step": 10050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2681792230029853e-05,
+      "loss": 0.4687,
+      "step": 10051
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268045363134942e-05,
+      "loss": 0.5041,
+      "step": 10052
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.26791149809194e-05,
+      "loss": 0.4887,
+      "step": 10053
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2677776278765634e-05,
+      "loss": 0.4704,
+      "step": 10054
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676437524913967e-05,
+      "loss": 0.48,
+      "step": 10055
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2675098719390243e-05,
+      "loss": 0.4508,
+      "step": 10056
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673759862220314e-05,
+      "loss": 0.4587,
+      "step": 10057
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2672420953430027e-05,
+      "loss": 0.4834,
+      "step": 10058
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2671081993045228e-05,
+      "loss": 0.502,
+      "step": 10059
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2669742981091771e-05,
+      "loss": 0.5103,
+      "step": 10060
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2668403917595507e-05,
+      "loss": 0.4827,
+      "step": 10061
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2667064802582287e-05,
+      "loss": 0.4939,
+      "step": 10062
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2665725636077963e-05,
+      "loss": 0.4887,
+      "step": 10063
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664386418108392e-05,
+      "loss": 0.4748,
+      "step": 10064
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2663047148699428e-05,
+      "loss": 0.488,
+      "step": 10065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2661707827876931e-05,
+      "loss": 0.4752,
+      "step": 10066
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2660368455666752e-05,
+      "loss": 0.4888,
+      "step": 10067
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2659029032094754e-05,
+      "loss": 0.4722,
+      "step": 10068
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2657689557186795e-05,
+      "loss": 0.4617,
+      "step": 10069
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656350030968734e-05,
+      "loss": 0.4768,
+      "step": 10070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2655010453466433e-05,
+      "loss": 0.4895,
+      "step": 10071
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653670824705753e-05,
+      "loss": 0.5064,
+      "step": 10072
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2652331144712563e-05,
+      "loss": 0.5097,
+      "step": 10073
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2650991413512718e-05,
+      "loss": 0.4813,
+      "step": 10074
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2649651631132089e-05,
+      "loss": 0.4672,
+      "step": 10075
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2648311797596544e-05,
+      "loss": 0.447,
+      "step": 10076
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2646971912931946e-05,
+      "loss": 0.4877,
+      "step": 10077
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2645631977164165e-05,
+      "loss": 0.4538,
+      "step": 10078
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264429199031907e-05,
+      "loss": 0.4652,
+      "step": 10079
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2642951952422533e-05,
+      "loss": 0.4729,
+      "step": 10080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641611863500423e-05,
+      "loss": 0.4623,
+      "step": 10081
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264027172357861e-05,
+      "loss": 0.4802,
+      "step": 10082
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263893153268297e-05,
+      "loss": 0.4825,
+      "step": 10083
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2637591290839377e-05,
+      "loss": 0.4787,
+      "step": 10084
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2636250998073707e-05,
+      "loss": 0.5254,
+      "step": 10085
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634910654411831e-05,
+      "loss": 0.4939,
+      "step": 10086
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2633570259879636e-05,
+      "loss": 0.4919,
+      "step": 10087
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2632229814502988e-05,
+      "loss": 0.4852,
+      "step": 10088
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2630889318307778e-05,
+      "loss": 0.4792,
+      "step": 10089
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629548771319873e-05,
+      "loss": 0.4753,
+      "step": 10090
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2628208173565164e-05,
+      "loss": 0.4818,
+      "step": 10091
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2626867525069533e-05,
+      "loss": 0.4791,
+      "step": 10092
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2625526825858856e-05,
+      "loss": 0.4479,
+      "step": 10093
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262418607595902e-05,
+      "loss": 0.513,
+      "step": 10094
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622845275395912e-05,
+      "loss": 0.5129,
+      "step": 10095
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621504424195417e-05,
+      "loss": 0.4904,
+      "step": 10096
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262016352238342e-05,
+      "loss": 0.508,
+      "step": 10097
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618822569985811e-05,
+      "loss": 0.4969,
+      "step": 10098
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2617481567028479e-05,
+      "loss": 0.4997,
+      "step": 10099
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616140513537311e-05,
+      "loss": 0.4925,
+      "step": 10100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26147994095382e-05,
+      "loss": 0.5003,
+      "step": 10101
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2613458255057037e-05,
+      "loss": 0.497,
+      "step": 10102
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2612117050119714e-05,
+      "loss": 0.4904,
+      "step": 10103
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2610775794752129e-05,
+      "loss": 0.5065,
+      "step": 10104
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609434488980168e-05,
+      "loss": 0.465,
+      "step": 10105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2608093132829735e-05,
+      "loss": 0.4731,
+      "step": 10106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2606751726326722e-05,
+      "loss": 0.5015,
+      "step": 10107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605410269497029e-05,
+      "loss": 0.4843,
+      "step": 10108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2604068762366555e-05,
+      "loss": 0.5047,
+      "step": 10109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2602727204961198e-05,
+      "loss": 0.5018,
+      "step": 10110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2601385597306858e-05,
+      "loss": 0.5018,
+      "step": 10111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2600043939429435e-05,
+      "loss": 0.503,
+      "step": 10112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2598702231354836e-05,
+      "loss": 0.5052,
+      "step": 10113
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259736047310896e-05,
+      "loss": 0.44,
+      "step": 10114
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2596018664717713e-05,
+      "loss": 0.4753,
+      "step": 10115
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2594676806207e-05,
+      "loss": 0.5042,
+      "step": 10116
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2593334897602732e-05,
+      "loss": 0.4868,
+      "step": 10117
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2591992938930808e-05,
+      "loss": 0.5066,
+      "step": 10118
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259065093021714e-05,
+      "loss": 0.4495,
+      "step": 10119
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258930887148764e-05,
+      "loss": 0.4857,
+      "step": 10120
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2587966762768215e-05,
+      "loss": 0.4788,
+      "step": 10121
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2586624604084778e-05,
+      "loss": 0.4927,
+      "step": 10122
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2585282395463236e-05,
+      "loss": 0.4905,
+      "step": 10123
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258394013692951e-05,
+      "loss": 0.4793,
+      "step": 10124
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2582597828509508e-05,
+      "loss": 0.4783,
+      "step": 10125
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2581255470229148e-05,
+      "loss": 0.502,
+      "step": 10126
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2579913062114344e-05,
+      "loss": 0.4665,
+      "step": 10127
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2578570604191013e-05,
+      "loss": 0.4882,
+      "step": 10128
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2577228096485076e-05,
+      "loss": 0.4876,
+      "step": 10129
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575885539022448e-05,
+      "loss": 0.4733,
+      "step": 10130
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2574542931829051e-05,
+      "loss": 0.4774,
+      "step": 10131
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2573200274930806e-05,
+      "loss": 0.4762,
+      "step": 10132
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2571857568353632e-05,
+      "loss": 0.5166,
+      "step": 10133
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2570514812123453e-05,
+      "loss": 0.4914,
+      "step": 10134
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2569172006266192e-05,
+      "loss": 0.4615,
+      "step": 10135
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567829150807778e-05,
+      "loss": 0.4577,
+      "step": 10136
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2566486245774132e-05,
+      "loss": 0.5128,
+      "step": 10137
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.256514329119118e-05,
+      "loss": 0.5281,
+      "step": 10138
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563800287084853e-05,
+      "loss": 0.492,
+      "step": 10139
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2562457233481077e-05,
+      "loss": 0.4779,
+      "step": 10140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2561114130405783e-05,
+      "loss": 0.5068,
+      "step": 10141
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.25597709778849e-05,
+      "loss": 0.5185,
+      "step": 10142
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2558427775944357e-05,
+      "loss": 0.4546,
+      "step": 10143
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557084524610093e-05,
+      "loss": 0.4821,
+      "step": 10144
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2555741223908033e-05,
+      "loss": 0.4633,
+      "step": 10145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2554397873864117e-05,
+      "loss": 0.4737,
+      "step": 10146
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2553054474504278e-05,
+      "loss": 0.475,
+      "step": 10147
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2551711025854453e-05,
+      "loss": 0.4968,
+      "step": 10148
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2550367527940575e-05,
+      "loss": 0.5029,
+      "step": 10149
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549023980788588e-05,
+      "loss": 0.4888,
+      "step": 10150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2547680384424428e-05,
+      "loss": 0.4768,
+      "step": 10151
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546336738874037e-05,
+      "loss": 0.4589,
+      "step": 10152
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254499304416335e-05,
+      "loss": 0.4733,
+      "step": 10153
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2543649300318315e-05,
+      "loss": 0.4943,
+      "step": 10154
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542305507364872e-05,
+      "loss": 0.507,
+      "step": 10155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540961665328961e-05,
+      "loss": 0.516,
+      "step": 10156
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2539617774236534e-05,
+      "loss": 0.4728,
+      "step": 10157
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538273834113533e-05,
+      "loss": 0.4911,
+      "step": 10158
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2536929844985904e-05,
+      "loss": 0.4665,
+      "step": 10159
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253558580687959e-05,
+      "loss": 0.4858,
+      "step": 10160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2534241719820552e-05,
+      "loss": 0.4798,
+      "step": 10161
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2532897583834728e-05,
+      "loss": 0.4695,
+      "step": 10162
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2531553398948075e-05,
+      "loss": 0.5196,
+      "step": 10163
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2530209165186533e-05,
+      "loss": 0.4626,
+      "step": 10164
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252886488257607e-05,
+      "loss": 0.467,
+      "step": 10165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2527520551142629e-05,
+      "loss": 0.4736,
+      "step": 10166
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2526176170912168e-05,
+      "loss": 0.4898,
+      "step": 10167
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2524831741910635e-05,
+      "loss": 0.4769,
+      "step": 10168
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2523487264163997e-05,
+      "loss": 0.4884,
+      "step": 10169
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522142737698202e-05,
+      "loss": 0.4758,
+      "step": 10170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2520798162539212e-05,
+      "loss": 0.4708,
+      "step": 10171
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2519453538712982e-05,
+      "loss": 0.4815,
+      "step": 10172
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251810886624548e-05,
+      "loss": 0.4738,
+      "step": 10173
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2516764145162657e-05,
+      "loss": 0.4558,
+      "step": 10174
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515419375490479e-05,
+      "loss": 0.4641,
+      "step": 10175
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2514074557254909e-05,
+      "loss": 0.4948,
+      "step": 10176
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512729690481908e-05,
+      "loss": 0.4957,
+      "step": 10177
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511384775197443e-05,
+      "loss": 0.464,
+      "step": 10178
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2510039811427477e-05,
+      "loss": 0.4782,
+      "step": 10179
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2508694799197981e-05,
+      "loss": 0.464,
+      "step": 10180
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2507349738534913e-05,
+      "loss": 0.5075,
+      "step": 10181
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250600462946425e-05,
+      "loss": 0.481,
+      "step": 10182
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2504659472011954e-05,
+      "loss": 0.4795,
+      "step": 10183
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2503314266204005e-05,
+      "loss": 0.4726,
+      "step": 10184
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2501969012066366e-05,
+      "loss": 0.4915,
+      "step": 10185
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2500623709625008e-05,
+      "loss": 0.4834,
+      "step": 10186
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499278358905906e-05,
+      "loss": 0.5092,
+      "step": 10187
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2497932959935038e-05,
+      "loss": 0.4793,
+      "step": 10188
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2496587512738372e-05,
+      "loss": 0.4889,
+      "step": 10189
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2495242017341887e-05,
+      "loss": 0.4896,
+      "step": 10190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493896473771557e-05,
+      "loss": 0.4746,
+      "step": 10191
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492550882053365e-05,
+      "loss": 0.4475,
+      "step": 10192
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2491205242213283e-05,
+      "loss": 0.5073,
+      "step": 10193
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248985955427729e-05,
+      "loss": 0.5222,
+      "step": 10194
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2488513818271372e-05,
+      "loss": 0.5026,
+      "step": 10195
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248716803422151e-05,
+      "loss": 0.4922,
+      "step": 10196
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485822202153681e-05,
+      "loss": 0.504,
+      "step": 10197
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2484476322093868e-05,
+      "loss": 0.4833,
+      "step": 10198
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483130394068062e-05,
+      "loss": 0.4832,
+      "step": 10199
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248178441810224e-05,
+      "loss": 0.4652,
+      "step": 10200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2480438394222392e-05,
+      "loss": 0.4758,
+      "step": 10201
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2479092322454503e-05,
+      "loss": 0.5074,
+      "step": 10202
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477746202824563e-05,
+      "loss": 0.506,
+      "step": 10203
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476400035358558e-05,
+      "loss": 0.4834,
+      "step": 10204
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475053820082476e-05,
+      "loss": 0.4906,
+      "step": 10205
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2473707557022312e-05,
+      "loss": 0.4603,
+      "step": 10206
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2472361246204057e-05,
+      "loss": 0.485,
+      "step": 10207
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2471014887653698e-05,
+      "loss": 0.5099,
+      "step": 10208
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2469668481397233e-05,
+      "loss": 0.4861,
+      "step": 10209
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2468322027460653e-05,
+      "loss": 0.5001,
+      "step": 10210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2466975525869957e-05,
+      "loss": 0.4639,
+      "step": 10211
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2465628976651138e-05,
+      "loss": 0.5082,
+      "step": 10212
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2464282379830193e-05,
+      "loss": 0.5371,
+      "step": 10213
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2462935735433119e-05,
+      "loss": 0.4498,
+      "step": 10214
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246158904348592e-05,
+      "loss": 0.498,
+      "step": 10215
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2460242304014585e-05,
+      "loss": 0.4964,
+      "step": 10216
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458895517045127e-05,
+      "loss": 0.5261,
+      "step": 10217
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2457548682603542e-05,
+      "loss": 0.4723,
+      "step": 10218
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2456201800715828e-05,
+      "loss": 0.4833,
+      "step": 10219
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.4648,
+      "step": 10220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453507894706041e-05,
+      "loss": 0.486,
+      "step": 10221
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2452160870635979e-05,
+      "loss": 0.5007,
+      "step": 10222
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2450813799223804e-05,
+      "loss": 0.4832,
+      "step": 10223
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2449466680495535e-05,
+      "loss": 0.4859,
+      "step": 10224
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448119514477172e-05,
+      "loss": 0.4901,
+      "step": 10225
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2446772301194727e-05,
+      "loss": 0.4881,
+      "step": 10226
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2445425040674208e-05,
+      "loss": 0.4993,
+      "step": 10227
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444077732941628e-05,
+      "loss": 0.4876,
+      "step": 10228
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2442730378022994e-05,
+      "loss": 0.4768,
+      "step": 10229
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441382975944325e-05,
+      "loss": 0.4897,
+      "step": 10230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2440035526731626e-05,
+      "loss": 0.4805,
+      "step": 10231
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243868803041092e-05,
+      "loss": 0.4751,
+      "step": 10232
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2437340487008217e-05,
+      "loss": 0.4711,
+      "step": 10233
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435992896549532e-05,
+      "loss": 0.4733,
+      "step": 10234
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2434645259060884e-05,
+      "loss": 0.4951,
+      "step": 10235
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2433297574568293e-05,
+      "loss": 0.4749,
+      "step": 10236
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2431949843097776e-05,
+      "loss": 0.5169,
+      "step": 10237
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2430602064675348e-05,
+      "loss": 0.481,
+      "step": 10238
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2429254239327037e-05,
+      "loss": 0.4995,
+      "step": 10239
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2427906367078861e-05,
+      "loss": 0.4859,
+      "step": 10240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2426558447956844e-05,
+      "loss": 0.4995,
+      "step": 10241
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2425210481987004e-05,
+      "loss": 0.4797,
+      "step": 10242
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423862469195376e-05,
+      "loss": 0.4894,
+      "step": 10243
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2422514409607972e-05,
+      "loss": 0.5005,
+      "step": 10244
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2421166303250827e-05,
+      "loss": 0.4881,
+      "step": 10245
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2419818150149965e-05,
+      "loss": 0.5301,
+      "step": 10246
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2418469950331415e-05,
+      "loss": 0.478,
+      "step": 10247
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417121703821205e-05,
+      "loss": 0.4834,
+      "step": 10248
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2415773410645363e-05,
+      "loss": 0.5097,
+      "step": 10249
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414425070829921e-05,
+      "loss": 0.4798,
+      "step": 10250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2413076684400913e-05,
+      "loss": 0.4735,
+      "step": 10251
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2411728251384364e-05,
+      "loss": 0.4855,
+      "step": 10252
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410379771806316e-05,
+      "loss": 0.459,
+      "step": 10253
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409031245692798e-05,
+      "loss": 0.4982,
+      "step": 10254
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2407682673069848e-05,
+      "loss": 0.4928,
+      "step": 10255
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406334053963498e-05,
+      "loss": 0.4695,
+      "step": 10256
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2404985388399787e-05,
+      "loss": 0.4738,
+      "step": 10257
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2403636676404751e-05,
+      "loss": 0.4857,
+      "step": 10258
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402287918004434e-05,
+      "loss": 0.4694,
+      "step": 10259
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240093911322487e-05,
+      "loss": 0.4524,
+      "step": 10260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.23995902620921e-05,
+      "loss": 0.4515,
+      "step": 10261
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239824136463217e-05,
+      "loss": 0.5245,
+      "step": 10262
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2396892420871113e-05,
+      "loss": 0.4571,
+      "step": 10263
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2395543430834982e-05,
+      "loss": 0.5004,
+      "step": 10264
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394194394549815e-05,
+      "loss": 0.4982,
+      "step": 10265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239284531204166e-05,
+      "loss": 0.4931,
+      "step": 10266
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.239149618333656e-05,
+      "loss": 0.4811,
+      "step": 10267
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2390147008460562e-05,
+      "loss": 0.4793,
+      "step": 10268
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2388797787439715e-05,
+      "loss": 0.5166,
+      "step": 10269
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238744852030007e-05,
+      "loss": 0.483,
+      "step": 10270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238609920706767e-05,
+      "loss": 0.486,
+      "step": 10271
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238474984776857e-05,
+      "loss": 0.4849,
+      "step": 10272
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238340044242882e-05,
+      "loss": 0.4907,
+      "step": 10273
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238205099107447e-05,
+      "loss": 0.4938,
+      "step": 10274
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2380701493731576e-05,
+      "loss": 0.5041,
+      "step": 10275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2379351950426188e-05,
+      "loss": 0.4992,
+      "step": 10276
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2378002361184366e-05,
+      "loss": 0.4882,
+      "step": 10277
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376652726032163e-05,
+      "loss": 0.4563,
+      "step": 10278
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237530304499563e-05,
+      "loss": 0.4857,
+      "step": 10279
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373953318100833e-05,
+      "loss": 0.4924,
+      "step": 10280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2372603545373829e-05,
+      "loss": 0.5039,
+      "step": 10281
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237125372684067e-05,
+      "loss": 0.4935,
+      "step": 10282
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2369903862527421e-05,
+      "loss": 0.4891,
+      "step": 10283
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368553952460145e-05,
+      "loss": 0.4816,
+      "step": 10284
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2367203996664901e-05,
+      "loss": 0.4595,
+      "step": 10285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2365853995167747e-05,
+      "loss": 0.508,
+      "step": 10286
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2364503947994755e-05,
+      "loss": 0.4791,
+      "step": 10287
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2363153855171985e-05,
+      "loss": 0.5005,
+      "step": 10288
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2361803716725506e-05,
+      "loss": 0.4616,
+      "step": 10289
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2360453532681376e-05,
+      "loss": 0.4626,
+      "step": 10290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235910330306567e-05,
+      "loss": 0.4732,
+      "step": 10291
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235775302790445e-05,
+      "loss": 0.4754,
+      "step": 10292
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2356402707223794e-05,
+      "loss": 0.5009,
+      "step": 10293
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2355052341049758e-05,
+      "loss": 0.4707,
+      "step": 10294
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2353701929408425e-05,
+      "loss": 0.5083,
+      "step": 10295
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2352351472325861e-05,
+      "loss": 0.4843,
+      "step": 10296
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2351000969828138e-05,
+      "loss": 0.4512,
+      "step": 10297
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234965042194133e-05,
+      "loss": 0.4944,
+      "step": 10298
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2348299828691512e-05,
+      "loss": 0.496,
+      "step": 10299
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346949190104757e-05,
+      "loss": 0.475,
+      "step": 10300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2345598506207142e-05,
+      "loss": 0.4851,
+      "step": 10301
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2344247777024746e-05,
+      "loss": 0.474,
+      "step": 10302
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342897002583639e-05,
+      "loss": 0.494,
+      "step": 10303
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2341546182909907e-05,
+      "loss": 0.5337,
+      "step": 10304
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2340195318029623e-05,
+      "loss": 0.4992,
+      "step": 10305
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338844407968877e-05,
+      "loss": 0.5097,
+      "step": 10306
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337493452753741e-05,
+      "loss": 0.5144,
+      "step": 10307
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.23361424524103e-05,
+      "loss": 0.4911,
+      "step": 10308
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2334791406964636e-05,
+      "loss": 0.4886,
+      "step": 10309
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2333440316442834e-05,
+      "loss": 0.4983,
+      "step": 10310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332089180870979e-05,
+      "loss": 0.4859,
+      "step": 10311
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2330738000275153e-05,
+      "loss": 0.4668,
+      "step": 10312
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2329386774681447e-05,
+      "loss": 0.4864,
+      "step": 10313
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2328035504115945e-05,
+      "loss": 0.4628,
+      "step": 10314
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2326684188604733e-05,
+      "loss": 0.4863,
+      "step": 10315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325332828173906e-05,
+      "loss": 0.4678,
+      "step": 10316
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232398142284955e-05,
+      "loss": 0.5008,
+      "step": 10317
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322629972657756e-05,
+      "loss": 0.4961,
+      "step": 10318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2321278477624615e-05,
+      "loss": 0.4764,
+      "step": 10319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319926937776217e-05,
+      "loss": 0.493,
+      "step": 10320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231857535313866e-05,
+      "loss": 0.4927,
+      "step": 10321
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2317223723738036e-05,
+      "loss": 0.4834,
+      "step": 10322
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2315872049600439e-05,
+      "loss": 0.4924,
+      "step": 10323
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2314520330751966e-05,
+      "loss": 0.4762,
+      "step": 10324
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2313168567218718e-05,
+      "loss": 0.4748,
+      "step": 10325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2311816759026783e-05,
+      "loss": 0.5053,
+      "step": 10326
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2310464906202264e-05,
+      "loss": 0.4851,
+      "step": 10327
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2309113008771263e-05,
+      "loss": 0.4853,
+      "step": 10328
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2307761066759878e-05,
+      "loss": 0.4856,
+      "step": 10329
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2306409080194209e-05,
+      "loss": 0.4641,
+      "step": 10330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2305057049100358e-05,
+      "loss": 0.4688,
+      "step": 10331
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303704973504427e-05,
+      "loss": 0.499,
+      "step": 10332
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302352853432521e-05,
+      "loss": 0.4965,
+      "step": 10333
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2301000688910745e-05,
+      "loss": 0.4897,
+      "step": 10334
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299648479965205e-05,
+      "loss": 0.4767,
+      "step": 10335
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2298296226622003e-05,
+      "loss": 0.4961,
+      "step": 10336
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229694392890725e-05,
+      "loss": 0.484,
+      "step": 10337
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2295591586847052e-05,
+      "loss": 0.5098,
+      "step": 10338
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2294239200467516e-05,
+      "loss": 0.4766,
+      "step": 10339
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2292886769794756e-05,
+      "loss": 0.4733,
+      "step": 10340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291534294854882e-05,
+      "loss": 0.5007,
+      "step": 10341
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2290181775673997e-05,
+      "loss": 0.4809,
+      "step": 10342
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2288829212278226e-05,
+      "loss": 0.4661,
+      "step": 10343
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287476604693672e-05,
+      "loss": 0.5128,
+      "step": 10344
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2286123952946454e-05,
+      "loss": 0.4711,
+      "step": 10345
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2284771257062684e-05,
+      "loss": 0.4572,
+      "step": 10346
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283418517068477e-05,
+      "loss": 0.4968,
+      "step": 10347
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2282065732989954e-05,
+      "loss": 0.4573,
+      "step": 10348
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280712904853228e-05,
+      "loss": 0.4901,
+      "step": 10349
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2279360032684415e-05,
+      "loss": 0.4966,
+      "step": 10350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2278007116509639e-05,
+      "loss": 0.4769,
+      "step": 10351
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227665415635502e-05,
+      "loss": 0.5073,
+      "step": 10352
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227530115224667e-05,
+      "loss": 0.4682,
+      "step": 10353
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2273948104210722e-05,
+      "loss": 0.497,
+      "step": 10354
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2272595012273292e-05,
+      "loss": 0.4584,
+      "step": 10355
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271241876460507e-05,
+      "loss": 0.4786,
+      "step": 10356
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269888696798484e-05,
+      "loss": 0.4976,
+      "step": 10357
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2268535473313356e-05,
+      "loss": 0.4675,
+      "step": 10358
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267182206031243e-05,
+      "loss": 0.4747,
+      "step": 10359
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265828894978273e-05,
+      "loss": 0.503,
+      "step": 10360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2264475540180574e-05,
+      "loss": 0.4721,
+      "step": 10361
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2263122141664276e-05,
+      "loss": 0.4711,
+      "step": 10362
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2261768699455506e-05,
+      "loss": 0.4921,
+      "step": 10363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2260415213580394e-05,
+      "loss": 0.4859,
+      "step": 10364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225906168406507e-05,
+      "loss": 0.5054,
+      "step": 10365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225770811093567e-05,
+      "loss": 0.4757,
+      "step": 10366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2256354494218321e-05,
+      "loss": 0.5095,
+      "step": 10367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2255000833939155e-05,
+      "loss": 0.4603,
+      "step": 10368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2253647130124313e-05,
+      "loss": 0.5017,
+      "step": 10369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252293382799927e-05,
+      "loss": 0.4461,
+      "step": 10370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225093959199213e-05,
+      "loss": 0.4667,
+      "step": 10371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2249585757727061e-05,
+      "loss": 0.4975,
+      "step": 10372
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2248231880030861e-05,
+      "loss": 0.4877,
+      "step": 10373
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2246877958929662e-05,
+      "loss": 0.505,
+      "step": 10374
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2245523994449607e-05,
+      "loss": 0.4719,
+      "step": 10375
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2244169986616834e-05,
+      "loss": 0.4826,
+      "step": 10376
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2242815935457487e-05,
+      "loss": 0.5108,
+      "step": 10377
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2241461840997703e-05,
+      "loss": 0.4524,
+      "step": 10378
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240107703263628e-05,
+      "loss": 0.4934,
+      "step": 10379
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2238753522281407e-05,
+      "loss": 0.4854,
+      "step": 10380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237399298077179e-05,
+      "loss": 0.5081,
+      "step": 10381
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2236045030677093e-05,
+      "loss": 0.4933,
+      "step": 10382
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2234690720107292e-05,
+      "loss": 0.5061,
+      "step": 10383
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2233336366393924e-05,
+      "loss": 0.5092,
+      "step": 10384
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223198196956314e-05,
+      "loss": 0.4904,
+      "step": 10385
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2230627529641084e-05,
+      "loss": 0.4741,
+      "step": 10386
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2229273046653905e-05,
+      "loss": 0.469,
+      "step": 10387
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2227918520627757e-05,
+      "loss": 0.4996,
+      "step": 10388
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2226563951588786e-05,
+      "loss": 0.4967,
+      "step": 10389
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.4738,
+      "step": 10390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222385468457699e-05,
+      "loss": 0.4581,
+      "step": 10391
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222249998665647e-05,
+      "loss": 0.4817,
+      "step": 10392
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221145245827744e-05,
+      "loss": 0.4999,
+      "step": 10393
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221979046211696e-05,
+      "loss": 0.5022,
+      "step": 10394
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2218435635550278e-05,
+      "loss": 0.4597,
+      "step": 10395
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221708076615386e-05,
+      "loss": 0.4724,
+      "step": 10396
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215725853953853e-05,
+      "loss": 0.5109,
+      "step": 10397
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2214370898976422e-05,
+      "loss": 0.493,
+      "step": 10398
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2213015901247722e-05,
+      "loss": 0.4711,
+      "step": 10399
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211660860793922e-05,
+      "loss": 0.4842,
+      "step": 10400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221030577764117e-05,
+      "loss": 0.4877,
+      "step": 10401
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2208950651815634e-05,
+      "loss": 0.4811,
+      "step": 10402
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2207595483343479e-05,
+      "loss": 0.51,
+      "step": 10403
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2206240272250866e-05,
+      "loss": 0.4838,
+      "step": 10404
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2204885018563955e-05,
+      "loss": 0.4668,
+      "step": 10405
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203529722308918e-05,
+      "loss": 0.4827,
+      "step": 10406
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202174383511916e-05,
+      "loss": 0.4921,
+      "step": 10407
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2200819002199118e-05,
+      "loss": 0.528,
+      "step": 10408
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199463578396688e-05,
+      "loss": 0.4839,
+      "step": 10409
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198108112130797e-05,
+      "loss": 0.4934,
+      "step": 10410
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196752603427615e-05,
+      "loss": 0.4734,
+      "step": 10411
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2195397052313309e-05,
+      "loss": 0.4643,
+      "step": 10412
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219404145881405e-05,
+      "loss": 0.4944,
+      "step": 10413
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2192685822956012e-05,
+      "loss": 0.5198,
+      "step": 10414
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191330144765368e-05,
+      "loss": 0.4779,
+      "step": 10415
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218997442426828e-05,
+      "loss": 0.4862,
+      "step": 10416
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2188618661490939e-05,
+      "loss": 0.4961,
+      "step": 10417
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218726285645951e-05,
+      "loss": 0.5185,
+      "step": 10418
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2185907009200169e-05,
+      "loss": 0.4695,
+      "step": 10419
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218455111973909e-05,
+      "loss": 0.4554,
+      "step": 10420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183195188102458e-05,
+      "loss": 0.5177,
+      "step": 10421
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2181839214316447e-05,
+      "loss": 0.4598,
+      "step": 10422
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2180483198407232e-05,
+      "loss": 0.4635,
+      "step": 10423
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2179127140400997e-05,
+      "loss": 0.4859,
+      "step": 10424
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2177771040323923e-05,
+      "loss": 0.4737,
+      "step": 10425
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217641489820219e-05,
+      "loss": 0.4872,
+      "step": 10426
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2175058714061976e-05,
+      "loss": 0.1793,
+      "step": 10427
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2173702487929469e-05,
+      "loss": 0.4835,
+      "step": 10428
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2172346219830853e-05,
+      "loss": 0.4905,
+      "step": 10429
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170989909792313e-05,
+      "loss": 0.496,
+      "step": 10430
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2169633557840025e-05,
+      "loss": 0.4899,
+      "step": 10431
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2168277164000189e-05,
+      "loss": 0.4895,
+      "step": 10432
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216692072829898e-05,
+      "loss": 0.4912,
+      "step": 10433
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165564250762594e-05,
+      "loss": 0.4957,
+      "step": 10434
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164207731417213e-05,
+      "loss": 0.5067,
+      "step": 10435
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162851170289034e-05,
+      "loss": 0.4644,
+      "step": 10436
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216149456740424e-05,
+      "loss": 0.1943,
+      "step": 10437
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2160137922789023e-05,
+      "loss": 0.4969,
+      "step": 10438
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2158781236469582e-05,
+      "loss": 0.4711,
+      "step": 10439
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.21574245084721e-05,
+      "loss": 0.5067,
+      "step": 10440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156067738822777e-05,
+      "loss": 0.494,
+      "step": 10441
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2154710927547801e-05,
+      "loss": 0.4836,
+      "step": 10442
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2153354074673373e-05,
+      "loss": 0.5142,
+      "step": 10443
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2151997180225687e-05,
+      "loss": 0.5036,
+      "step": 10444
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2150640244230939e-05,
+      "loss": 0.4867,
+      "step": 10445
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2149283266715326e-05,
+      "loss": 0.4936,
+      "step": 10446
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147926247705043e-05,
+      "loss": 0.492,
+      "step": 10447
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2146569187226301e-05,
+      "loss": 0.4902,
+      "step": 10448
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2145212085305284e-05,
+      "loss": 0.5068,
+      "step": 10449
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2143854941968203e-05,
+      "loss": 0.483,
+      "step": 10450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2142497757241256e-05,
+      "loss": 0.4832,
+      "step": 10451
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2141140531150643e-05,
+      "loss": 0.4812,
+      "step": 10452
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2139783263722573e-05,
+      "loss": 0.4853,
+      "step": 10453
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138425954983245e-05,
+      "loss": 0.5249,
+      "step": 10454
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2137068604958866e-05,
+      "loss": 0.4805,
+      "step": 10455
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2135711213675638e-05,
+      "loss": 0.4776,
+      "step": 10456
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2134353781159771e-05,
+      "loss": 0.4772,
+      "step": 10457
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213299630743747e-05,
+      "loss": 0.4949,
+      "step": 10458
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2131638792534944e-05,
+      "loss": 0.4723,
+      "step": 10459
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21302812364784e-05,
+      "loss": 0.4866,
+      "step": 10460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2128923639294047e-05,
+      "loss": 0.4789,
+      "step": 10461
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127566001008097e-05,
+      "loss": 0.4899,
+      "step": 10462
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126208321646761e-05,
+      "loss": 0.4736,
+      "step": 10463
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212485060123625e-05,
+      "loss": 0.4744,
+      "step": 10464
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123492839802777e-05,
+      "loss": 0.5166,
+      "step": 10465
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2122135037372553e-05,
+      "loss": 0.4977,
+      "step": 10466
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2120777193971795e-05,
+      "loss": 0.4665,
+      "step": 10467
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2119419309626714e-05,
+      "loss": 0.4882,
+      "step": 10468
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2118061384363533e-05,
+      "loss": 0.4874,
+      "step": 10469
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2116703418208462e-05,
+      "loss": 0.498,
+      "step": 10470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2115345411187722e-05,
+      "loss": 0.4568,
+      "step": 10471
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2113987363327526e-05,
+      "loss": 0.4871,
+      "step": 10472
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21126292746541e-05,
+      "loss": 0.4884,
+      "step": 10473
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111271145193659e-05,
+      "loss": 0.5251,
+      "step": 10474
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2109912974972424e-05,
+      "loss": 0.4955,
+      "step": 10475
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2108554764016614e-05,
+      "loss": 0.4779,
+      "step": 10476
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2107196512352455e-05,
+      "loss": 0.4482,
+      "step": 10477
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210583822000617e-05,
+      "loss": 0.4656,
+      "step": 10478
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104479887003976e-05,
+      "loss": 0.4509,
+      "step": 10479
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103121513372109e-05,
+      "loss": 0.4894,
+      "step": 10480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2101763099136782e-05,
+      "loss": 0.5035,
+      "step": 10481
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210040464432423e-05,
+      "loss": 0.4731,
+      "step": 10482
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209904614896067e-05,
+      "loss": 0.4986,
+      "step": 10483
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097687613072344e-05,
+      "loss": 0.5329,
+      "step": 10484
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2096329036685469e-05,
+      "loss": 0.4904,
+      "step": 10485
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2094970419826274e-05,
+      "loss": 0.4723,
+      "step": 10486
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2093611762520991e-05,
+      "loss": 0.4837,
+      "step": 10487
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092253064795856e-05,
+      "loss": 0.4745,
+      "step": 10488
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2090894326677092e-05,
+      "loss": 0.4913,
+      "step": 10489
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2089535548190935e-05,
+      "loss": 0.4702,
+      "step": 10490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088176729363618e-05,
+      "loss": 0.4876,
+      "step": 10491
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2086817870221376e-05,
+      "loss": 0.4596,
+      "step": 10492
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2085458970790442e-05,
+      "loss": 0.4835,
+      "step": 10493
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208410003109705e-05,
+      "loss": 0.4929,
+      "step": 10494
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082741051167438e-05,
+      "loss": 0.4901,
+      "step": 10495
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2081382031027843e-05,
+      "loss": 0.461,
+      "step": 10496
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20800229707045e-05,
+      "loss": 0.5087,
+      "step": 10497
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2078663870223653e-05,
+      "loss": 0.4935,
+      "step": 10498
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2077304729611536e-05,
+      "loss": 0.4809,
+      "step": 10499
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2075945548894388e-05,
+      "loss": 0.5029,
+      "step": 10500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074586328098454e-05,
+      "loss": 0.4745,
+      "step": 10501
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073227067249973e-05,
+      "loss": 0.4959,
+      "step": 10502
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2071867766375189e-05,
+      "loss": 0.4892,
+      "step": 10503
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2070508425500344e-05,
+      "loss": 0.4668,
+      "step": 10504
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206914904465168e-05,
+      "loss": 0.5081,
+      "step": 10505
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2067789623855446e-05,
+      "loss": 0.4808,
+      "step": 10506
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2066430163137884e-05,
+      "loss": 0.4818,
+      "step": 10507
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206507066252524e-05,
+      "loss": 0.4981,
+      "step": 10508
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2063711122043759e-05,
+      "loss": 0.4879,
+      "step": 10509
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2062351541719697e-05,
+      "loss": 0.4486,
+      "step": 10510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2060991921579294e-05,
+      "loss": 0.4738,
+      "step": 10511
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059632261648804e-05,
+      "loss": 0.5107,
+      "step": 10512
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058272561954472e-05,
+      "loss": 0.478,
+      "step": 10513
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2056912822522552e-05,
+      "loss": 0.4851,
+      "step": 10514
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.20555530433793e-05,
+      "loss": 0.5235,
+      "step": 10515
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2054193224550958e-05,
+      "loss": 0.4572,
+      "step": 10516
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2052833366063787e-05,
+      "loss": 0.538,
+      "step": 10517
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2051473467944039e-05,
+      "loss": 0.4895,
+      "step": 10518
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2050113530217968e-05,
+      "loss": 0.4504,
+      "step": 10519
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2048753552911827e-05,
+      "loss": 0.4802,
+      "step": 10520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2047393536051877e-05,
+      "loss": 0.4805,
+      "step": 10521
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2046033479664372e-05,
+      "loss": 0.4764,
+      "step": 10522
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204467338377557e-05,
+      "loss": 0.4877,
+      "step": 10523
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2043313248411728e-05,
+      "loss": 0.4752,
+      "step": 10524
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2041953073599109e-05,
+      "loss": 0.4681,
+      "step": 10525
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204059285936397e-05,
+      "loss": 0.5129,
+      "step": 10526
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2039232605732569e-05,
+      "loss": 0.4753,
+      "step": 10527
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2037872312731173e-05,
+      "loss": 0.4899,
+      "step": 10528
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2036511980386041e-05,
+      "loss": 0.4741,
+      "step": 10529
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2035151608723438e-05,
+      "loss": 0.4755,
+      "step": 10530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2033791197769622e-05,
+      "loss": 0.4748,
+      "step": 10531
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2032430747550868e-05,
+      "loss": 0.4924,
+      "step": 10532
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2031070258093431e-05,
+      "loss": 0.1956,
+      "step": 10533
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2029709729423587e-05,
+      "loss": 0.4742,
+      "step": 10534
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2028349161567588e-05,
+      "loss": 0.1935,
+      "step": 10535
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2026988554551717e-05,
+      "loss": 0.4635,
+      "step": 10536
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2025627908402235e-05,
+      "loss": 0.4696,
+      "step": 10537
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202426722314541e-05,
+      "loss": 0.4392,
+      "step": 10538
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2022906498807514e-05,
+      "loss": 0.4622,
+      "step": 10539
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021545735414816e-05,
+      "loss": 0.5007,
+      "step": 10540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202018493299359e-05,
+      "loss": 0.4915,
+      "step": 10541
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018824091570103e-05,
+      "loss": 0.4851,
+      "step": 10542
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2017463211170635e-05,
+      "loss": 0.4617,
+      "step": 10543
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2016102291821455e-05,
+      "loss": 0.4728,
+      "step": 10544
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014741333548836e-05,
+      "loss": 0.5034,
+      "step": 10545
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2013380336379055e-05,
+      "loss": 0.4856,
+      "step": 10546
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.201201930033839e-05,
+      "loss": 0.533,
+      "step": 10547
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2010658225453111e-05,
+      "loss": 0.4977,
+      "step": 10548
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2009297111749504e-05,
+      "loss": 0.4797,
+      "step": 10549
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200793595925384e-05,
+      "loss": 0.4923,
+      "step": 10550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006574767992403e-05,
+      "loss": 0.5008,
+      "step": 10551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2005213537991467e-05,
+      "loss": 0.4548,
+      "step": 10552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2003852269277316e-05,
+      "loss": 0.4946,
+      "step": 10553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200249096187623e-05,
+      "loss": 0.4754,
+      "step": 10554
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001129615814495e-05,
+      "loss": 0.4961,
+      "step": 10555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1999768231118385e-05,
+      "loss": 0.5053,
+      "step": 10556
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1998406807814188e-05,
+      "loss": 0.508,
+      "step": 10557
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997045345928192e-05,
+      "loss": 0.5094,
+      "step": 10558
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1995683845486675e-05,
+      "loss": 0.5002,
+      "step": 10559
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994322306515926e-05,
+      "loss": 0.1906,
+      "step": 10560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1992960729042227e-05,
+      "loss": 0.4944,
+      "step": 10561
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1991599113091871e-05,
+      "loss": 0.4912,
+      "step": 10562
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1990237458691143e-05,
+      "loss": 0.4693,
+      "step": 10563
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1988875765866333e-05,
+      "loss": 0.4675,
+      "step": 10564
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987514034643725e-05,
+      "loss": 0.5204,
+      "step": 10565
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1986152265049617e-05,
+      "loss": 0.49,
+      "step": 10566
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1984790457110293e-05,
+      "loss": 0.4531,
+      "step": 10567
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1983428610852048e-05,
+      "loss": 0.5028,
+      "step": 10568
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198206672630117e-05,
+      "loss": 0.4879,
+      "step": 10569
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980704803483963e-05,
+      "loss": 0.4887,
+      "step": 10570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1979342842426706e-05,
+      "loss": 0.4603,
+      "step": 10571
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977980843155701e-05,
+      "loss": 0.5036,
+      "step": 10572
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1976618805697244e-05,
+      "loss": 0.4678,
+      "step": 10573
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197525673007763e-05,
+      "loss": 0.4962,
+      "step": 10574
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973894616323152e-05,
+      "loss": 0.1837,
+      "step": 10575
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1972532464460112e-05,
+      "loss": 0.4689,
+      "step": 10576
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971170274514802e-05,
+      "loss": 0.5145,
+      "step": 10577
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196980804651353e-05,
+      "loss": 0.4547,
+      "step": 10578
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1968445780482592e-05,
+      "loss": 0.4735,
+      "step": 10579
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967083476448282e-05,
+      "loss": 0.4857,
+      "step": 10580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196572113443691e-05,
+      "loss": 0.4668,
+      "step": 10581
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1964358754474769e-05,
+      "loss": 0.4803,
+      "step": 10582
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196299633658817e-05,
+      "loss": 0.4901,
+      "step": 10583
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1961633880803412e-05,
+      "loss": 0.5027,
+      "step": 10584
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1960271387146799e-05,
+      "loss": 0.5037,
+      "step": 10585
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958908855644635e-05,
+      "loss": 0.4753,
+      "step": 10586
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1957546286323228e-05,
+      "loss": 0.451,
+      "step": 10587
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956183679208879e-05,
+      "loss": 0.4879,
+      "step": 10588
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1954821034327904e-05,
+      "loss": 0.4973,
+      "step": 10589
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1953458351706601e-05,
+      "loss": 0.4887,
+      "step": 10590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1952095631371283e-05,
+      "loss": 0.461,
+      "step": 10591
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950732873348258e-05,
+      "loss": 0.4871,
+      "step": 10592
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194937007766384e-05,
+      "loss": 0.5011,
+      "step": 10593
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1948007244344334e-05,
+      "loss": 0.4799,
+      "step": 10594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946644373416049e-05,
+      "loss": 0.4698,
+      "step": 10595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1945281464905302e-05,
+      "loss": 0.4929,
+      "step": 10596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1943918518838409e-05,
+      "loss": 0.4571,
+      "step": 10597
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942555535241676e-05,
+      "loss": 0.5168,
+      "step": 10598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1941192514141421e-05,
+      "loss": 0.494,
+      "step": 10599
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193982945556396e-05,
+      "loss": 0.4863,
+      "step": 10600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1938466359535606e-05,
+      "loss": 0.4516,
+      "step": 10601
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1937103226082673e-05,
+      "loss": 0.4688,
+      "step": 10602
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1935740055231482e-05,
+      "loss": 0.5097,
+      "step": 10603
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934376847008353e-05,
+      "loss": 0.4746,
+      "step": 10604
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1933013601439596e-05,
+      "loss": 0.442,
+      "step": 10605
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193165031855154e-05,
+      "loss": 0.5147,
+      "step": 10606
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1930286998370498e-05,
+      "loss": 0.4841,
+      "step": 10607
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1928923640922795e-05,
+      "loss": 0.4715,
+      "step": 10608
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192756024623475e-05,
+      "loss": 0.4856,
+      "step": 10609
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1926196814332685e-05,
+      "loss": 0.5156,
+      "step": 10610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1924833345242921e-05,
+      "loss": 0.4779,
+      "step": 10611
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1923469838991788e-05,
+      "loss": 0.477,
+      "step": 10612
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1922106295605602e-05,
+      "loss": 0.4405,
+      "step": 10613
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1920742715110695e-05,
+      "loss": 0.4742,
+      "step": 10614
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919379097533388e-05,
+      "loss": 0.4871,
+      "step": 10615
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191801544290001e-05,
+      "loss": 0.5101,
+      "step": 10616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916651751236885e-05,
+      "loss": 0.4799,
+      "step": 10617
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1915288022570346e-05,
+      "loss": 0.5015,
+      "step": 10618
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191392425692672e-05,
+      "loss": 0.4812,
+      "step": 10619
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1912560454332329e-05,
+      "loss": 0.4629,
+      "step": 10620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1911196614813511e-05,
+      "loss": 0.4851,
+      "step": 10621
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1909832738396598e-05,
+      "loss": 0.4744,
+      "step": 10622
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1908468825107917e-05,
+      "loss": 0.4862,
+      "step": 10623
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1907104874973795e-05,
+      "loss": 0.4807,
+      "step": 10624
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1905740888020576e-05,
+      "loss": 0.5046,
+      "step": 10625
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1904376864274588e-05,
+      "loss": 0.1842,
+      "step": 10626
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903012803762161e-05,
+      "loss": 0.48,
+      "step": 10627
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1901648706509637e-05,
+      "loss": 0.5051,
+      "step": 10628
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190028457254335e-05,
+      "loss": 0.4827,
+      "step": 10629
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898920401889636e-05,
+      "loss": 0.5018,
+      "step": 10630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1897556194574829e-05,
+      "loss": 0.4786,
+      "step": 10631
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1896191950625268e-05,
+      "loss": 0.4934,
+      "step": 10632
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1894827670067297e-05,
+      "loss": 0.4867,
+      "step": 10633
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1893463352927248e-05,
+      "loss": 0.4671,
+      "step": 10634
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1892098999231463e-05,
+      "loss": 0.4667,
+      "step": 10635
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890734609006285e-05,
+      "loss": 0.4801,
+      "step": 10636
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1889370182278053e-05,
+      "loss": 0.4939,
+      "step": 10637
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188800571907311e-05,
+      "loss": 0.4591,
+      "step": 10638
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886641219417798e-05,
+      "loss": 0.4932,
+      "step": 10639
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188527668333846e-05,
+      "loss": 0.4924,
+      "step": 10640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1883912110861442e-05,
+      "loss": 0.4605,
+      "step": 10641
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1882547502013086e-05,
+      "loss": 0.4732,
+      "step": 10642
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881182856819742e-05,
+      "loss": 0.4961,
+      "step": 10643
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1879818175307753e-05,
+      "loss": 0.4802,
+      "step": 10644
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1878453457503465e-05,
+      "loss": 0.48,
+      "step": 10645
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877088703433226e-05,
+      "loss": 0.5127,
+      "step": 10646
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1875723913123384e-05,
+      "loss": 0.4734,
+      "step": 10647
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874359086600293e-05,
+      "loss": 0.4906,
+      "step": 10648
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.18729942238903e-05,
+      "loss": 0.4904,
+      "step": 10649
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187162932501975e-05,
+      "loss": 0.4342,
+      "step": 10650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1870264390015001e-05,
+      "loss": 0.1726,
+      "step": 10651
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1868899418902402e-05,
+      "loss": 0.4772,
+      "step": 10652
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867534411708307e-05,
+      "loss": 0.4964,
+      "step": 10653
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1866169368459066e-05,
+      "loss": 0.4699,
+      "step": 10654
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864804289181036e-05,
+      "loss": 0.4513,
+      "step": 10655
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1863439173900573e-05,
+      "loss": 0.4805,
+      "step": 10656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862074022644025e-05,
+      "loss": 0.4839,
+      "step": 10657
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1860708835437756e-05,
+      "loss": 0.4591,
+      "step": 10658
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1859343612308118e-05,
+      "loss": 0.4704,
+      "step": 10659
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857978353281474e-05,
+      "loss": 0.502,
+      "step": 10660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856613058384172e-05,
+      "loss": 0.5136,
+      "step": 10661
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185524772764258e-05,
+      "loss": 0.4868,
+      "step": 10662
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1853882361083055e-05,
+      "loss": 0.4895,
+      "step": 10663
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852516958731957e-05,
+      "loss": 0.4947,
+      "step": 10664
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1851151520615645e-05,
+      "loss": 0.4852,
+      "step": 10665
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184978604676048e-05,
+      "loss": 0.4922,
+      "step": 10666
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1848420537192831e-05,
+      "loss": 0.468,
+      "step": 10667
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184705499193905e-05,
+      "loss": 0.4917,
+      "step": 10668
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1845689411025512e-05,
+      "loss": 0.4746,
+      "step": 10669
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1844323794478572e-05,
+      "loss": 0.4601,
+      "step": 10670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1842958142324603e-05,
+      "loss": 0.4805,
+      "step": 10671
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184159245458996e-05,
+      "loss": 0.4916,
+      "step": 10672
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1840226731301021e-05,
+      "loss": 0.4783,
+      "step": 10673
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838860972484146e-05,
+      "loss": 0.5098,
+      "step": 10674
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837495178165706e-05,
+      "loss": 0.4799,
+      "step": 10675
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1836129348372065e-05,
+      "loss": 0.5008,
+      "step": 10676
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834763483129598e-05,
+      "loss": 0.4887,
+      "step": 10677
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.183339758246467e-05,
+      "loss": 0.4774,
+      "step": 10678
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1832031646403654e-05,
+      "loss": 0.5142,
+      "step": 10679
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830665674972921e-05,
+      "loss": 0.4727,
+      "step": 10680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1829299668198843e-05,
+      "loss": 0.4905,
+      "step": 10681
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.182793362610779e-05,
+      "loss": 0.4752,
+      "step": 10682
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1826567548726138e-05,
+      "loss": 0.4717,
+      "step": 10683
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1825201436080258e-05,
+      "loss": 0.4837,
+      "step": 10684
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.182383528819653e-05,
+      "loss": 0.4731,
+      "step": 10685
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1822469105101325e-05,
+      "loss": 0.5086,
+      "step": 10686
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821102886821017e-05,
+      "loss": 0.5114,
+      "step": 10687
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1819736633381986e-05,
+      "loss": 0.477,
+      "step": 10688
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181837034481061e-05,
+      "loss": 0.4869,
+      "step": 10689
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817004021133263e-05,
+      "loss": 0.481,
+      "step": 10690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1815637662376326e-05,
+      "loss": 0.5017,
+      "step": 10691
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1814271268566182e-05,
+      "loss": 0.511,
+      "step": 10692
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812904839729205e-05,
+      "loss": 0.4827,
+      "step": 10693
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1811538375891778e-05,
+      "loss": 0.4754,
+      "step": 10694
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181017187708028e-05,
+      "loss": 0.4643,
+      "step": 10695
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1808805343321102e-05,
+      "loss": 0.4707,
+      "step": 10696
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807438774640614e-05,
+      "loss": 0.4891,
+      "step": 10697
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806072171065206e-05,
+      "loss": 0.4602,
+      "step": 10698
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180470553262126e-05,
+      "loss": 0.4776,
+      "step": 10699
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1803338859335169e-05,
+      "loss": 0.5,
+      "step": 10700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801972151233306e-05,
+      "loss": 0.4787,
+      "step": 10701
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1800605408342064e-05,
+      "loss": 0.5006,
+      "step": 10702
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1799238630687827e-05,
+      "loss": 0.468,
+      "step": 10703
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1797871818296987e-05,
+      "loss": 0.486,
+      "step": 10704
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1796504971195923e-05,
+      "loss": 0.4946,
+      "step": 10705
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795138089411033e-05,
+      "loss": 0.5004,
+      "step": 10706
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1793771172968705e-05,
+      "loss": 0.4932,
+      "step": 10707
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1792404221895324e-05,
+      "loss": 0.4686,
+      "step": 10708
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1791037236217283e-05,
+      "loss": 0.4961,
+      "step": 10709
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789670215960975e-05,
+      "loss": 0.4848,
+      "step": 10710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1788303161152792e-05,
+      "loss": 0.4643,
+      "step": 10711
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1786936071819125e-05,
+      "loss": 0.486,
+      "step": 10712
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785568947986368e-05,
+      "loss": 0.492,
+      "step": 10713
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1784201789680916e-05,
+      "loss": 0.4726,
+      "step": 10714
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1782834596929164e-05,
+      "loss": 0.4669,
+      "step": 10715
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1781467369757504e-05,
+      "loss": 0.4856,
+      "step": 10716
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780100108192336e-05,
+      "loss": 0.4954,
+      "step": 10717
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778732812260056e-05,
+      "loss": 0.4855,
+      "step": 10718
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777365481987062e-05,
+      "loss": 0.4497,
+      "step": 10719
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775998117399747e-05,
+      "loss": 0.4638,
+      "step": 10720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1774630718524516e-05,
+      "loss": 0.4678,
+      "step": 10721
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1773263285387767e-05,
+      "loss": 0.4734,
+      "step": 10722
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1771895818015897e-05,
+      "loss": 0.4804,
+      "step": 10723
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1770528316435305e-05,
+      "loss": 0.515,
+      "step": 10724
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769160780672405e-05,
+      "loss": 0.5,
+      "step": 10725
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1767793210753581e-05,
+      "loss": 0.4933,
+      "step": 10726
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1766425606705251e-05,
+      "loss": 0.4708,
+      "step": 10727
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176505796855381e-05,
+      "loss": 0.5265,
+      "step": 10728
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1763690296325666e-05,
+      "loss": 0.5161,
+      "step": 10729
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176232259004722e-05,
+      "loss": 0.4709,
+      "step": 10730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176095484974488e-05,
+      "loss": 0.5091,
+      "step": 10731
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175958707544505e-05,
+      "loss": 0.4796,
+      "step": 10732
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1758219267174136e-05,
+      "loss": 0.4856,
+      "step": 10733
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175685142495855e-05,
+      "loss": 0.463,
+      "step": 10734
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1755483548824693e-05,
+      "loss": 0.5118,
+      "step": 10735
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754115638798982e-05,
+      "loss": 0.4575,
+      "step": 10736
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175274769490782e-05,
+      "loss": 0.5139,
+      "step": 10737
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175137971717762e-05,
+      "loss": 0.4747,
+      "step": 10738
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1750011705634787e-05,
+      "loss": 0.4725,
+      "step": 10739
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1748643660305743e-05,
+      "loss": 0.4664,
+      "step": 10740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747275581216888e-05,
+      "loss": 0.4735,
+      "step": 10741
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1745907468394642e-05,
+      "loss": 0.5016,
+      "step": 10742
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1744539321865416e-05,
+      "loss": 0.5034,
+      "step": 10743
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1743171141655625e-05,
+      "loss": 0.4788,
+      "step": 10744
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741802927791681e-05,
+      "loss": 0.484,
+      "step": 10745
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1740434680299999e-05,
+      "loss": 0.4692,
+      "step": 10746
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1739066399206997e-05,
+      "loss": 0.5035,
+      "step": 10747
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1737698084539093e-05,
+      "loss": 0.4902,
+      "step": 10748
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736329736322703e-05,
+      "loss": 0.5102,
+      "step": 10749
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1734961354584238e-05,
+      "loss": 0.5282,
+      "step": 10750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733592939350125e-05,
+      "loss": 0.463,
+      "step": 10751
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1732224490646779e-05,
+      "loss": 0.4834,
+      "step": 10752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1730856008500619e-05,
+      "loss": 0.4863,
+      "step": 10753
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172948749293807e-05,
+      "loss": 0.4986,
+      "step": 10754
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172811894398555e-05,
+      "loss": 0.4848,
+      "step": 10755
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1726750361669476e-05,
+      "loss": 0.4748,
+      "step": 10756
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1725381746016278e-05,
+      "loss": 0.4828,
+      "step": 10757
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1724013097052373e-05,
+      "loss": 0.4878,
+      "step": 10758
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172264441480419e-05,
+      "loss": 0.4679,
+      "step": 10759
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721275699298147e-05,
+      "loss": 0.4825,
+      "step": 10760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1719906950560673e-05,
+      "loss": 0.4783,
+      "step": 10761
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1718538168618192e-05,
+      "loss": 0.4964,
+      "step": 10762
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1717169353497132e-05,
+      "loss": 0.46,
+      "step": 10763
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1715800505223918e-05,
+      "loss": 0.4766,
+      "step": 10764
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1714431623824975e-05,
+      "loss": 0.455,
+      "step": 10765
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1713062709326736e-05,
+      "loss": 0.4689,
+      "step": 10766
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171169376175563e-05,
+      "loss": 0.482,
+      "step": 10767
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1710324781138078e-05,
+      "loss": 0.4849,
+      "step": 10768
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1708955767500515e-05,
+      "loss": 0.5169,
+      "step": 10769
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707586720869375e-05,
+      "loss": 0.4738,
+      "step": 10770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1706217641271087e-05,
+      "loss": 0.5188,
+      "step": 10771
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170484852873208e-05,
+      "loss": 0.4856,
+      "step": 10772
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703479383278788e-05,
+      "loss": 0.508,
+      "step": 10773
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170211020493765e-05,
+      "loss": 0.4776,
+      "step": 10774
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700740993735087e-05,
+      "loss": 0.5005,
+      "step": 10775
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1699371749697543e-05,
+      "loss": 0.5093,
+      "step": 10776
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698002472851451e-05,
+      "loss": 0.5148,
+      "step": 10777
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1696633163223248e-05,
+      "loss": 0.4746,
+      "step": 10778
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1695263820839368e-05,
+      "loss": 0.4808,
+      "step": 10779
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1693894445726245e-05,
+      "loss": 0.4828,
+      "step": 10780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692525037910325e-05,
+      "loss": 0.4781,
+      "step": 10781
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691155597418041e-05,
+      "loss": 0.4909,
+      "step": 10782
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1689786124275828e-05,
+      "loss": 0.5134,
+      "step": 10783
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688416618510133e-05,
+      "loss": 0.5021,
+      "step": 10784
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168704708014739e-05,
+      "loss": 0.4829,
+      "step": 10785
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1685677509214044e-05,
+      "loss": 0.4627,
+      "step": 10786
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1684307905736532e-05,
+      "loss": 0.4698,
+      "step": 10787
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1682938269741301e-05,
+      "loss": 0.519,
+      "step": 10788
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681568601254792e-05,
+      "loss": 0.4878,
+      "step": 10789
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1680198900303446e-05,
+      "loss": 0.4557,
+      "step": 10790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1678829166913705e-05,
+      "loss": 0.5139,
+      "step": 10791
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1677459401112021e-05,
+      "loss": 0.4941,
+      "step": 10792
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676089602924836e-05,
+      "loss": 0.4688,
+      "step": 10793
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674719772378591e-05,
+      "loss": 0.4823,
+      "step": 10794
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1673349909499736e-05,
+      "loss": 0.4629,
+      "step": 10795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671980014314716e-05,
+      "loss": 0.4915,
+      "step": 10796
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1670610086849985e-05,
+      "loss": 0.4677,
+      "step": 10797
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.166924012713198e-05,
+      "loss": 0.5148,
+      "step": 10798
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1667870135187163e-05,
+      "loss": 0.4727,
+      "step": 10799
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1666500111041971e-05,
+      "loss": 0.5008,
+      "step": 10800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1665130054722863e-05,
+      "loss": 0.4941,
+      "step": 10801
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1663759966256285e-05,
+      "loss": 0.4959,
+      "step": 10802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.166238984566869e-05,
+      "loss": 0.4774,
+      "step": 10803
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1661019692986532e-05,
+      "loss": 0.4887,
+      "step": 10804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165964950823626e-05,
+      "loss": 0.4888,
+      "step": 10805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1658279291444326e-05,
+      "loss": 0.4831,
+      "step": 10806
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165690904263719e-05,
+      "loss": 0.4851,
+      "step": 10807
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16555387618413e-05,
+      "loss": 0.4647,
+      "step": 10808
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654168449083115e-05,
+      "loss": 0.4811,
+      "step": 10809
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165279810438909e-05,
+      "loss": 0.4969,
+      "step": 10810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1651427727785681e-05,
+      "loss": 0.5047,
+      "step": 10811
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650057319299342e-05,
+      "loss": 0.4659,
+      "step": 10812
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1648686878956532e-05,
+      "loss": 0.4768,
+      "step": 10813
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1647316406783715e-05,
+      "loss": 0.4759,
+      "step": 10814
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164594590280734e-05,
+      "loss": 0.5056,
+      "step": 10815
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1644575367053874e-05,
+      "loss": 0.1674,
+      "step": 10816
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643204799549773e-05,
+      "loss": 0.5214,
+      "step": 10817
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1641834200321499e-05,
+      "loss": 0.5122,
+      "step": 10818
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640463569395512e-05,
+      "loss": 0.4641,
+      "step": 10819
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1639092906798276e-05,
+      "loss": 0.4549,
+      "step": 10820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163772221255625e-05,
+      "loss": 0.4717,
+      "step": 10821
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16363514866959e-05,
+      "loss": 0.5014,
+      "step": 10822
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1634980729243688e-05,
+      "loss": 0.5003,
+      "step": 10823
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1633609940226078e-05,
+      "loss": 0.4854,
+      "step": 10824
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632239119669537e-05,
+      "loss": 0.5103,
+      "step": 10825
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163086826760053e-05,
+      "loss": 0.4545,
+      "step": 10826
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629497384045522e-05,
+      "loss": 0.5079,
+      "step": 10827
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628126469030978e-05,
+      "loss": 0.4612,
+      "step": 10828
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1626755522583369e-05,
+      "loss": 0.4983,
+      "step": 10829
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1625384544729162e-05,
+      "loss": 0.4528,
+      "step": 10830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162401353549482e-05,
+      "loss": 0.4874,
+      "step": 10831
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1622642494906819e-05,
+      "loss": 0.4574,
+      "step": 10832
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621271422991627e-05,
+      "loss": 0.4509,
+      "step": 10833
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1619900319775714e-05,
+      "loss": 0.4948,
+      "step": 10834
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161852918528555e-05,
+      "loss": 0.4919,
+      "step": 10835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1617158019547606e-05,
+      "loss": 0.48,
+      "step": 10836
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615786822588357e-05,
+      "loss": 0.4886,
+      "step": 10837
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1614415594434271e-05,
+      "loss": 0.4776,
+      "step": 10838
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1613044335111825e-05,
+      "loss": 0.483,
+      "step": 10839
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1611673044647492e-05,
+      "loss": 0.4702,
+      "step": 10840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.161030172306775e-05,
+      "loss": 0.4812,
+      "step": 10841
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1608930370399065e-05,
+      "loss": 0.4678,
+      "step": 10842
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1607558986667922e-05,
+      "loss": 0.5013,
+      "step": 10843
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1606187571900791e-05,
+      "loss": 0.4792,
+      "step": 10844
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1604816126124154e-05,
+      "loss": 0.4784,
+      "step": 10845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603444649364485e-05,
+      "loss": 0.5092,
+      "step": 10846
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602073141648264e-05,
+      "loss": 0.4719,
+      "step": 10847
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1600701603001967e-05,
+      "loss": 0.4468,
+      "step": 10848
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1599330033452078e-05,
+      "loss": 0.4779,
+      "step": 10849
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597958433025072e-05,
+      "loss": 0.508,
+      "step": 10850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1596586801747434e-05,
+      "loss": 0.4831,
+      "step": 10851
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1595215139645641e-05,
+      "loss": 0.4906,
+      "step": 10852
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1593843446746178e-05,
+      "loss": 0.4907,
+      "step": 10853
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1592471723075522e-05,
+      "loss": 0.466,
+      "step": 10854
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1591099968660164e-05,
+      "loss": 0.4589,
+      "step": 10855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1589728183526583e-05,
+      "loss": 0.4959,
+      "step": 10856
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588356367701262e-05,
+      "loss": 0.492,
+      "step": 10857
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1586984521210686e-05,
+      "loss": 0.4924,
+      "step": 10858
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585612644081342e-05,
+      "loss": 0.4808,
+      "step": 10859
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1584240736339716e-05,
+      "loss": 0.4825,
+      "step": 10860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582868798012287e-05,
+      "loss": 0.4758,
+      "step": 10861
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1581496829125556e-05,
+      "loss": 0.4818,
+      "step": 10862
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1580124829706e-05,
+      "loss": 0.463,
+      "step": 10863
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1578752799780114e-05,
+      "loss": 0.4967,
+      "step": 10864
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1577380739374376e-05,
+      "loss": 0.4763,
+      "step": 10865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576008648515286e-05,
+      "loss": 0.4594,
+      "step": 10866
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157463652722933e-05,
+      "loss": 0.4765,
+      "step": 10867
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573264375543002e-05,
+      "loss": 0.4805,
+      "step": 10868
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571892193482786e-05,
+      "loss": 0.4811,
+      "step": 10869
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1570519981075178e-05,
+      "loss": 0.4758,
+      "step": 10870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1569147738346673e-05,
+      "loss": 0.487,
+      "step": 10871
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567775465323758e-05,
+      "loss": 0.4873,
+      "step": 10872
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156640316203293e-05,
+      "loss": 0.4819,
+      "step": 10873
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1565030828500685e-05,
+      "loss": 0.4798,
+      "step": 10874
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563658464753516e-05,
+      "loss": 0.4619,
+      "step": 10875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1562286070817914e-05,
+      "loss": 0.4716,
+      "step": 10876
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1560913646720379e-05,
+      "loss": 0.4827,
+      "step": 10877
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155954119248741e-05,
+      "loss": 0.4491,
+      "step": 10878
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1558168708145497e-05,
+      "loss": 0.4977,
+      "step": 10879
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1556796193721141e-05,
+      "loss": 0.4854,
+      "step": 10880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1555423649240845e-05,
+      "loss": 0.4815,
+      "step": 10881
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1554051074731099e-05,
+      "loss": 0.4822,
+      "step": 10882
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1552678470218406e-05,
+      "loss": 0.5117,
+      "step": 10883
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1551305835729267e-05,
+      "loss": 0.4407,
+      "step": 10884
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549933171290184e-05,
+      "loss": 0.4713,
+      "step": 10885
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1548560476927654e-05,
+      "loss": 0.4502,
+      "step": 10886
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1547187752668181e-05,
+      "loss": 0.5096,
+      "step": 10887
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1545814998538267e-05,
+      "loss": 0.475,
+      "step": 10888
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1544442214564415e-05,
+      "loss": 0.4926,
+      "step": 10889
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543069400773126e-05,
+      "loss": 0.488,
+      "step": 10890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1541696557190906e-05,
+      "loss": 0.5222,
+      "step": 10891
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154032368384426e-05,
+      "loss": 0.5185,
+      "step": 10892
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1538950780759694e-05,
+      "loss": 0.5037,
+      "step": 10893
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1537577847963711e-05,
+      "loss": 0.4906,
+      "step": 10894
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536204885482816e-05,
+      "loss": 0.473,
+      "step": 10895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1534831893343518e-05,
+      "loss": 0.4881,
+      "step": 10896
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.153345887157233e-05,
+      "loss": 0.4635,
+      "step": 10897
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532085820195748e-05,
+      "loss": 0.4778,
+      "step": 10898
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1530712739240289e-05,
+      "loss": 0.499,
+      "step": 10899
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1529339628732462e-05,
+      "loss": 0.4906,
+      "step": 10900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1527966488698772e-05,
+      "loss": 0.509,
+      "step": 10901
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1526593319165731e-05,
+      "loss": 0.4637,
+      "step": 10902
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1525220120159852e-05,
+      "loss": 0.4949,
+      "step": 10903
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523846891707646e-05,
+      "loss": 0.4782,
+      "step": 10904
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1522473633835623e-05,
+      "loss": 0.4512,
+      "step": 10905
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1521100346570295e-05,
+      "loss": 0.4931,
+      "step": 10906
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151972702993818e-05,
+      "loss": 0.4698,
+      "step": 10907
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1518353683965786e-05,
+      "loss": 0.4808,
+      "step": 10908
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1516980308679628e-05,
+      "loss": 0.4816,
+      "step": 10909
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1515606904106222e-05,
+      "loss": 0.4885,
+      "step": 10910
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1514233470272085e-05,
+      "loss": 0.4885,
+      "step": 10911
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1512860007203734e-05,
+      "loss": 0.4657,
+      "step": 10912
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1511486514927676e-05,
+      "loss": 0.4679,
+      "step": 10913
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1510112993470442e-05,
+      "loss": 0.4932,
+      "step": 10914
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.150873944285854e-05,
+      "loss": 0.4631,
+      "step": 10915
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.150736586311849e-05,
+      "loss": 0.4753,
+      "step": 10916
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1505992254276808e-05,
+      "loss": 0.4686,
+      "step": 10917
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150461861636002e-05,
+      "loss": 0.4971,
+      "step": 10918
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150324494939464e-05,
+      "loss": 0.4903,
+      "step": 10919
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1501871253407196e-05,
+      "loss": 0.4881,
+      "step": 10920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1500497528424199e-05,
+      "loss": 0.4688,
+      "step": 10921
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1499123774472176e-05,
+      "loss": 0.5317,
+      "step": 10922
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1497749991577652e-05,
+      "loss": 0.4736,
+      "step": 10923
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496376179767139e-05,
+      "loss": 0.4636,
+      "step": 10924
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1495002339067173e-05,
+      "loss": 0.4704,
+      "step": 10925
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149362846950427e-05,
+      "loss": 0.4943,
+      "step": 10926
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1492254571104959e-05,
+      "loss": 0.4809,
+      "step": 10927
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490880643895757e-05,
+      "loss": 0.4784,
+      "step": 10928
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1489506687903197e-05,
+      "loss": 0.4789,
+      "step": 10929
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1488132703153805e-05,
+      "loss": 0.4672,
+      "step": 10930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1486758689674102e-05,
+      "loss": 0.4672,
+      "step": 10931
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1485384647490618e-05,
+      "loss": 0.4751,
+      "step": 10932
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1484010576629882e-05,
+      "loss": 0.4846,
+      "step": 10933
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148263647711842e-05,
+      "loss": 0.5037,
+      "step": 10934
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1481262348982762e-05,
+      "loss": 0.4762,
+      "step": 10935
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1479888192249437e-05,
+      "loss": 0.4683,
+      "step": 10936
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147851400694498e-05,
+      "loss": 0.4947,
+      "step": 10937
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1477139793095909e-05,
+      "loss": 0.485,
+      "step": 10938
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1475765550728764e-05,
+      "loss": 0.4929,
+      "step": 10939
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1474391279870076e-05,
+      "loss": 0.4672,
+      "step": 10940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1473016980546377e-05,
+      "loss": 0.4847,
+      "step": 10941
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471642652784198e-05,
+      "loss": 0.4536,
+      "step": 10942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1470268296610069e-05,
+      "loss": 0.4611,
+      "step": 10943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1468893912050531e-05,
+      "loss": 0.4716,
+      "step": 10944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467519499132113e-05,
+      "loss": 0.4879,
+      "step": 10945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1466145057881353e-05,
+      "loss": 0.4848,
+      "step": 10946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464770588324782e-05,
+      "loss": 0.4703,
+      "step": 10947
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1463396090488943e-05,
+      "loss": 0.4585,
+      "step": 10948
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1462021564400366e-05,
+      "loss": 0.4761,
+      "step": 10949
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1460647010085588e-05,
+      "loss": 0.499,
+      "step": 10950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459272427571148e-05,
+      "loss": 0.5067,
+      "step": 10951
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145789781688359e-05,
+      "loss": 0.5046,
+      "step": 10952
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1456523178049443e-05,
+      "loss": 0.4998,
+      "step": 10953
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455148511095252e-05,
+      "loss": 0.486,
+      "step": 10954
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1453773816047554e-05,
+      "loss": 0.4806,
+      "step": 10955
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145239909293289e-05,
+      "loss": 0.4761,
+      "step": 10956
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1451024341777801e-05,
+      "loss": 0.4587,
+      "step": 10957
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1449649562608828e-05,
+      "loss": 0.4705,
+      "step": 10958
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1448274755452515e-05,
+      "loss": 0.4568,
+      "step": 10959
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1446899920335407e-05,
+      "loss": 0.4808,
+      "step": 10960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1445525057284038e-05,
+      "loss": 0.4919,
+      "step": 10961
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1444150166324954e-05,
+      "loss": 0.481,
+      "step": 10962
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1442775247484707e-05,
+      "loss": 0.4797,
+      "step": 10963
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1441400300789831e-05,
+      "loss": 0.4995,
+      "step": 10964
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1440025326266877e-05,
+      "loss": 0.4957,
+      "step": 10965
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143865032394239e-05,
+      "loss": 0.4709,
+      "step": 10966
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1437275293842917e-05,
+      "loss": 0.509,
+      "step": 10967
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435900235995004e-05,
+      "loss": 0.4727,
+      "step": 10968
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434525150425195e-05,
+      "loss": 0.4735,
+      "step": 10969
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1433150037160043e-05,
+      "loss": 0.5315,
+      "step": 10970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1431774896226093e-05,
+      "loss": 0.5087,
+      "step": 10971
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1430399727649895e-05,
+      "loss": 0.495,
+      "step": 10972
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1429024531457998e-05,
+      "loss": 0.4667,
+      "step": 10973
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1427649307676949e-05,
+      "loss": 0.4749,
+      "step": 10974
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1426274056333307e-05,
+      "loss": 0.487,
+      "step": 10975
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1424898777453615e-05,
+      "loss": 0.4928,
+      "step": 10976
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1423523471064425e-05,
+      "loss": 0.4967,
+      "step": 10977
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1422148137192293e-05,
+      "loss": 0.5004,
+      "step": 10978
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1420772775863771e-05,
+      "loss": 0.4632,
+      "step": 10979
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419397387105405e-05,
+      "loss": 0.4944,
+      "step": 10980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141802197094376e-05,
+      "loss": 0.4612,
+      "step": 10981
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1416646527405382e-05,
+      "loss": 0.4794,
+      "step": 10982
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1415271056516833e-05,
+      "loss": 0.4751,
+      "step": 10983
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413895558304658e-05,
+      "loss": 0.4849,
+      "step": 10984
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141252003279542e-05,
+      "loss": 0.4989,
+      "step": 10985
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1411144480015674e-05,
+      "loss": 0.5179,
+      "step": 10986
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1409768899991973e-05,
+      "loss": 0.4655,
+      "step": 10987
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1408393292750883e-05,
+      "loss": 0.4802,
+      "step": 10988
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1407017658318952e-05,
+      "loss": 0.4736,
+      "step": 10989
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1405641996722748e-05,
+      "loss": 0.5183,
+      "step": 10990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404266307988819e-05,
+      "loss": 0.4649,
+      "step": 10991
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1402890592143733e-05,
+      "loss": 0.4687,
+      "step": 10992
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1401514849214048e-05,
+      "loss": 0.4712,
+      "step": 10993
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1400139079226325e-05,
+      "loss": 0.459,
+      "step": 10994
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1398763282207118e-05,
+      "loss": 0.5176,
+      "step": 10995
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1397387458182998e-05,
+      "loss": 0.4915,
+      "step": 10996
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396011607180524e-05,
+      "loss": 0.4587,
+      "step": 10997
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1394635729226255e-05,
+      "loss": 0.4816,
+      "step": 10998
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1393259824346758e-05,
+      "loss": 0.4588,
+      "step": 10999
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1391883892568599e-05,
+      "loss": 0.4791,
+      "step": 11000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390507933918336e-05,
+      "loss": 0.5136,
+      "step": 11001
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389131948422534e-05,
+      "loss": 0.4933,
+      "step": 11002
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1387755936107762e-05,
+      "loss": 0.4866,
+      "step": 11003
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386379897000585e-05,
+      "loss": 0.4964,
+      "step": 11004
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1385003831127573e-05,
+      "loss": 0.4531,
+      "step": 11005
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138362773851528e-05,
+      "loss": 0.4851,
+      "step": 11006
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1382251619190284e-05,
+      "loss": 0.4763,
+      "step": 11007
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138087547317915e-05,
+      "loss": 0.18,
+      "step": 11008
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1379499300508446e-05,
+      "loss": 0.506,
+      "step": 11009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378123101204742e-05,
+      "loss": 0.4767,
+      "step": 11010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1376746875294606e-05,
+      "loss": 0.4838,
+      "step": 11011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137537062280461e-05,
+      "loss": 0.5099,
+      "step": 11012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1373994343761318e-05,
+      "loss": 0.4859,
+      "step": 11013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1372618038191307e-05,
+      "loss": 0.4963,
+      "step": 11014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137124170612115e-05,
+      "loss": 0.4667,
+      "step": 11015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136986534757741e-05,
+      "loss": 0.4954,
+      "step": 11016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136848896258667e-05,
+      "loss": 0.4908,
+      "step": 11017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1367112551175494e-05,
+      "loss": 0.4963,
+      "step": 11018
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1365736113370463e-05,
+      "loss": 0.5196,
+      "step": 11019
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1364359649198145e-05,
+      "loss": 0.5055,
+      "step": 11020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1362983158685115e-05,
+      "loss": 0.4645,
+      "step": 11021
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136160664185795e-05,
+      "loss": 0.4851,
+      "step": 11022
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1360230098743225e-05,
+      "loss": 0.4791,
+      "step": 11023
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1358853529367515e-05,
+      "loss": 0.4986,
+      "step": 11024
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1357476933757398e-05,
+      "loss": 0.4736,
+      "step": 11025
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1356100311939452e-05,
+      "loss": 0.4661,
+      "step": 11026
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135472366394025e-05,
+      "loss": 0.4878,
+      "step": 11027
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135334698978637e-05,
+      "loss": 0.4939,
+      "step": 11028
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351970289504396e-05,
+      "loss": 0.4693,
+      "step": 11029
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1350593563120907e-05,
+      "loss": 0.4686,
+      "step": 11030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349216810662476e-05,
+      "loss": 0.5081,
+      "step": 11031
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1347840032155686e-05,
+      "loss": 0.4756,
+      "step": 11032
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1346463227627115e-05,
+      "loss": 0.4673,
+      "step": 11033
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1345086397103353e-05,
+      "loss": 0.4779,
+      "step": 11034
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1343709540610971e-05,
+      "loss": 0.4707,
+      "step": 11035
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342332658176556e-05,
+      "loss": 0.4859,
+      "step": 11036
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134095574982669e-05,
+      "loss": 0.4736,
+      "step": 11037
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1339578815587957e-05,
+      "loss": 0.4735,
+      "step": 11038
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1338201855486936e-05,
+      "loss": 0.4754,
+      "step": 11039
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336824869550216e-05,
+      "loss": 0.5024,
+      "step": 11040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1335447857804377e-05,
+      "loss": 0.4856,
+      "step": 11041
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1334070820276013e-05,
+      "loss": 0.494,
+      "step": 11042
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332693756991697e-05,
+      "loss": 0.4986,
+      "step": 11043
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1331316667978026e-05,
+      "loss": 0.1755,
+      "step": 11044
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1329939553261575e-05,
+      "loss": 0.4829,
+      "step": 11045
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328562412868944e-05,
+      "loss": 0.5063,
+      "step": 11046
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1327185246826712e-05,
+      "loss": 0.474,
+      "step": 11047
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325808055161465e-05,
+      "loss": 0.4725,
+      "step": 11048
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1324430837899801e-05,
+      "loss": 0.514,
+      "step": 11049
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1323053595068296e-05,
+      "loss": 0.4599,
+      "step": 11050
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1321676326693553e-05,
+      "loss": 0.4604,
+      "step": 11051
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1320299032802153e-05,
+      "loss": 0.4524,
+      "step": 11052
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1318921713420691e-05,
+      "loss": 0.4935,
+      "step": 11053
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1317544368575752e-05,
+      "loss": 0.4756,
+      "step": 11054
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1316166998293937e-05,
+      "loss": 0.4939,
+      "step": 11055
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314789602601828e-05,
+      "loss": 0.4695,
+      "step": 11056
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1313412181526025e-05,
+      "loss": 0.4988,
+      "step": 11057
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1312034735093114e-05,
+      "loss": 0.4808,
+      "step": 11058
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1310657263329695e-05,
+      "loss": 0.4764,
+      "step": 11059
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1309279766262355e-05,
+      "loss": 0.4975,
+      "step": 11060
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307902243917694e-05,
+      "loss": 0.4767,
+      "step": 11061
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1306524696322304e-05,
+      "loss": 0.4843,
+      "step": 11062
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1305147123502784e-05,
+      "loss": 0.4603,
+      "step": 11063
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1303769525485726e-05,
+      "loss": 0.4698,
+      "step": 11064
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1302391902297726e-05,
+      "loss": 0.502,
+      "step": 11065
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301014253965383e-05,
+      "loss": 0.4973,
+      "step": 11066
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1299636580515294e-05,
+      "loss": 0.4995,
+      "step": 11067
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1298258881974056e-05,
+      "loss": 0.5073,
+      "step": 11068
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1296881158368268e-05,
+      "loss": 0.4744,
+      "step": 11069
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1295503409724526e-05,
+      "loss": 0.4833,
+      "step": 11070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294125636069433e-05,
+      "loss": 0.5035,
+      "step": 11071
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1292747837429586e-05,
+      "loss": 0.4657,
+      "step": 11072
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1291370013831587e-05,
+      "loss": 0.4323,
+      "step": 11073
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1289992165302036e-05,
+      "loss": 0.5038,
+      "step": 11074
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1288614291867532e-05,
+      "loss": 0.4623,
+      "step": 11075
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128723639355468e-05,
+      "loss": 0.4708,
+      "step": 11076
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1285858470390079e-05,
+      "loss": 0.4847,
+      "step": 11077
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1284480522400337e-05,
+      "loss": 0.4847,
+      "step": 11078
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128310254961205e-05,
+      "loss": 0.4597,
+      "step": 11079
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1281724552051824e-05,
+      "loss": 0.4706,
+      "step": 11080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280346529746264e-05,
+      "loss": 0.4777,
+      "step": 11081
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1278968482721978e-05,
+      "loss": 0.4975,
+      "step": 11082
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277590411005566e-05,
+      "loss": 0.504,
+      "step": 11083
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1276212314623634e-05,
+      "loss": 0.5042,
+      "step": 11084
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274834193602786e-05,
+      "loss": 0.4664,
+      "step": 11085
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1273456047969635e-05,
+      "loss": 0.4776,
+      "step": 11086
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1272077877750782e-05,
+      "loss": 0.4423,
+      "step": 11087
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1270699682972836e-05,
+      "loss": 0.4758,
+      "step": 11088
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269321463662406e-05,
+      "loss": 0.4866,
+      "step": 11089
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12679432198461e-05,
+      "loss": 0.4594,
+      "step": 11090
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1266564951550523e-05,
+      "loss": 0.4687,
+      "step": 11091
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265186658802289e-05,
+      "loss": 0.4692,
+      "step": 11092
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1263808341628009e-05,
+      "loss": 0.4696,
+      "step": 11093
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1262430000054286e-05,
+      "loss": 0.5325,
+      "step": 11094
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261051634107735e-05,
+      "loss": 0.4801,
+      "step": 11095
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259673243814967e-05,
+      "loss": 0.4775,
+      "step": 11096
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1258294829202598e-05,
+      "loss": 0.456,
+      "step": 11097
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1256916390297232e-05,
+      "loss": 0.5204,
+      "step": 11098
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255537927125485e-05,
+      "loss": 0.4855,
+      "step": 11099
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125415943971397e-05,
+      "loss": 0.4774,
+      "step": 11100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1252780928089301e-05,
+      "loss": 0.494,
+      "step": 11101
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1251402392278093e-05,
+      "loss": 0.5115,
+      "step": 11102
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250023832306956e-05,
+      "loss": 0.4656,
+      "step": 11103
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124864524820251e-05,
+      "loss": 0.5159,
+      "step": 11104
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124726663999137e-05,
+      "loss": 0.4905,
+      "step": 11105
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245888007700143e-05,
+      "loss": 0.4517,
+      "step": 11106
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244509351355458e-05,
+      "loss": 0.4957,
+      "step": 11107
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1243130670983927e-05,
+      "loss": 0.4771,
+      "step": 11108
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1241751966612165e-05,
+      "loss": 0.4979,
+      "step": 11109
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1240373238266786e-05,
+      "loss": 0.4914,
+      "step": 11110
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238994485974417e-05,
+      "loss": 0.4933,
+      "step": 11111
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1237615709761673e-05,
+      "loss": 0.4748,
+      "step": 11112
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1236236909655172e-05,
+      "loss": 0.4771,
+      "step": 11113
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234858085681533e-05,
+      "loss": 0.4488,
+      "step": 11114
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1233479237867378e-05,
+      "loss": 0.4713,
+      "step": 11115
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1232100366239327e-05,
+      "loss": 0.4528,
+      "step": 11116
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230721470823995e-05,
+      "loss": 0.5338,
+      "step": 11117
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1229342551648014e-05,
+      "loss": 0.475,
+      "step": 11118
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1227963608738e-05,
+      "loss": 0.4749,
+      "step": 11119
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226584642120574e-05,
+      "loss": 0.4661,
+      "step": 11120
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1225205651822359e-05,
+      "loss": 0.1756,
+      "step": 11121
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223826637869984e-05,
+      "loss": 0.4892,
+      "step": 11122
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1222447600290066e-05,
+      "loss": 0.4774,
+      "step": 11123
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122106853910923e-05,
+      "loss": 0.4781,
+      "step": 11124
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1219689454354104e-05,
+      "loss": 0.4578,
+      "step": 11125
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121831034605131e-05,
+      "loss": 0.4622,
+      "step": 11126
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1216931214227474e-05,
+      "loss": 0.5126,
+      "step": 11127
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1215552058909222e-05,
+      "loss": 0.4722,
+      "step": 11128
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1214172880123179e-05,
+      "loss": 0.4652,
+      "step": 11129
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212793677895976e-05,
+      "loss": 0.4596,
+      "step": 11130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1211414452254239e-05,
+      "loss": 0.4947,
+      "step": 11131
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121003520322459e-05,
+      "loss": 0.4659,
+      "step": 11132
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208655930833666e-05,
+      "loss": 0.4668,
+      "step": 11133
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207276635108088e-05,
+      "loss": 0.4801,
+      "step": 11134
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1205897316074487e-05,
+      "loss": 0.4462,
+      "step": 11135
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204517973759497e-05,
+      "loss": 0.5169,
+      "step": 11136
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1203138608189745e-05,
+      "loss": 0.4553,
+      "step": 11137
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1201759219391858e-05,
+      "loss": 0.4774,
+      "step": 11138
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1200379807392472e-05,
+      "loss": 0.4965,
+      "step": 11139
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1199000372218214e-05,
+      "loss": 0.4797,
+      "step": 11140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1197620913895721e-05,
+      "loss": 0.4778,
+      "step": 11141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1196241432451623e-05,
+      "loss": 0.4916,
+      "step": 11142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1194861927912547e-05,
+      "loss": 0.4633,
+      "step": 11143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193482400305136e-05,
+      "loss": 0.5006,
+      "step": 11144
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1192102849656016e-05,
+      "loss": 0.475,
+      "step": 11145
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1190723275991824e-05,
+      "loss": 0.5061,
+      "step": 11146
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1189343679339195e-05,
+      "loss": 0.4765,
+      "step": 11147
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.118796405972476e-05,
+      "loss": 0.502,
+      "step": 11148
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186584417175162e-05,
+      "loss": 0.4757,
+      "step": 11149
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118520475171703e-05,
+      "loss": 0.4742,
+      "step": 11150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1183825063376999e-05,
+      "loss": 0.4957,
+      "step": 11151
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118244535218171e-05,
+      "loss": 0.4715,
+      "step": 11152
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1181065618157802e-05,
+      "loss": 0.472,
+      "step": 11153
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1179685861331908e-05,
+      "loss": 0.4991,
+      "step": 11154
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1178306081730666e-05,
+      "loss": 0.4845,
+      "step": 11155
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1176926279380718e-05,
+      "loss": 0.4925,
+      "step": 11156
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1175546454308699e-05,
+      "loss": 0.506,
+      "step": 11157
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117416660654125e-05,
+      "loss": 0.5002,
+      "step": 11158
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117278673610501e-05,
+      "loss": 0.4879,
+      "step": 11159
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1171406843026625e-05,
+      "loss": 0.4996,
+      "step": 11160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1170026927332724e-05,
+      "loss": 0.4887,
+      "step": 11161
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1168646989049958e-05,
+      "loss": 0.4891,
+      "step": 11162
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167267028204963e-05,
+      "loss": 0.4638,
+      "step": 11163
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165887044824386e-05,
+      "loss": 0.482,
+      "step": 11164
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1164507038934864e-05,
+      "loss": 0.5093,
+      "step": 11165
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1163127010563042e-05,
+      "loss": 0.4849,
+      "step": 11166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1161746959735563e-05,
+      "loss": 0.4757,
+      "step": 11167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1160366886479073e-05,
+      "loss": 0.4639,
+      "step": 11168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115898679082021e-05,
+      "loss": 0.4913,
+      "step": 11169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157606672785627e-05,
+      "loss": 0.4843,
+      "step": 11170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1156226532401962e-05,
+      "loss": 0.5146,
+      "step": 11171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154846369695864e-05,
+      "loss": 0.4933,
+      "step": 11172
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1153466184693973e-05,
+      "loss": 0.4427,
+      "step": 11173
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1152085977422946e-05,
+      "loss": 0.4841,
+      "step": 11174
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1150705747909423e-05,
+      "loss": 0.487,
+      "step": 11175
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1149325496180047e-05,
+      "loss": 0.4825,
+      "step": 11176
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1147945222261474e-05,
+      "loss": 0.4709,
+      "step": 11177
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1146564926180347e-05,
+      "loss": 0.4895,
+      "step": 11178
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1145184607963318e-05,
+      "loss": 0.4523,
+      "step": 11179
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1143804267637028e-05,
+      "loss": 0.5056,
+      "step": 11180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142423905228134e-05,
+      "loss": 0.4895,
+      "step": 11181
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1141043520763284e-05,
+      "loss": 0.4786,
+      "step": 11182
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113966311426913e-05,
+      "loss": 0.4666,
+      "step": 11183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138282685772315e-05,
+      "loss": 0.48,
+      "step": 11184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136902235299495e-05,
+      "loss": 0.4416,
+      "step": 11185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1135521762877323e-05,
+      "loss": 0.4735,
+      "step": 11186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134141268532451e-05,
+      "loss": 0.4627,
+      "step": 11187
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1132760752291524e-05,
+      "loss": 0.4714,
+      "step": 11188
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131380214181205e-05,
+      "loss": 0.4934,
+      "step": 11189
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129999654228139e-05,
+      "loss": 0.4721,
+      "step": 11190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128619072458982e-05,
+      "loss": 0.4989,
+      "step": 11191
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1127238468900387e-05,
+      "loss": 0.4823,
+      "step": 11192
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125857843579013e-05,
+      "loss": 0.4613,
+      "step": 11193
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.112447719652151e-05,
+      "loss": 0.4697,
+      "step": 11194
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123096527754533e-05,
+      "loss": 0.4545,
+      "step": 11195
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1121715837304743e-05,
+      "loss": 0.4851,
+      "step": 11196
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1120335125198786e-05,
+      "loss": 0.5012,
+      "step": 11197
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1118954391463328e-05,
+      "loss": 0.4705,
+      "step": 11198
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117573636125021e-05,
+      "loss": 0.5111,
+      "step": 11199
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1116192859210527e-05,
+      "loss": 0.4777,
+      "step": 11200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1114812060746496e-05,
+      "loss": 0.4665,
+      "step": 11201
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113431240759592e-05,
+      "loss": 0.4864,
+      "step": 11202
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111205039927647e-05,
+      "loss": 0.4841,
+      "step": 11203
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1110669536323793e-05,
+      "loss": 0.4887,
+      "step": 11204
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1109288651928216e-05,
+      "loss": 0.4901,
+      "step": 11205
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1107907746116402e-05,
+      "loss": 0.4778,
+      "step": 11206
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1106526818915008e-05,
+      "loss": 0.5038,
+      "step": 11207
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11051458703507e-05,
+      "loss": 0.5072,
+      "step": 11208
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1103764900450133e-05,
+      "loss": 0.4913,
+      "step": 11209
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102383909239971e-05,
+      "loss": 0.4776,
+      "step": 11210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1101002896746876e-05,
+      "loss": 0.4911,
+      "step": 11211
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1099621862997511e-05,
+      "loss": 0.5115,
+      "step": 11212
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098240808018534e-05,
+      "loss": 0.467,
+      "step": 11213
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1096859731836615e-05,
+      "loss": 0.4983,
+      "step": 11214
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1095478634478412e-05,
+      "loss": 0.5129,
+      "step": 11215
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094097515970592e-05,
+      "loss": 0.4854,
+      "step": 11216
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1092716376339818e-05,
+      "loss": 0.5039,
+      "step": 11217
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1091335215612752e-05,
+      "loss": 0.4666,
+      "step": 11218
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089954033816066e-05,
+      "loss": 0.4513,
+      "step": 11219
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108857283097642e-05,
+      "loss": 0.4602,
+      "step": 11220
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108719160712048e-05,
+      "loss": 0.4839,
+      "step": 11221
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1085810362274913e-05,
+      "loss": 0.4755,
+      "step": 11222
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108442909646639e-05,
+      "loss": 0.4649,
+      "step": 11223
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108304780972157e-05,
+      "loss": 0.4718,
+      "step": 11224
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1081666502067126e-05,
+      "loss": 0.5047,
+      "step": 11225
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1080285173529724e-05,
+      "loss": 0.5131,
+      "step": 11226
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078903824136035e-05,
+      "loss": 0.4707,
+      "step": 11227
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1077522453912724e-05,
+      "loss": 0.4808,
+      "step": 11228
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1076141062886463e-05,
+      "loss": 0.4769,
+      "step": 11229
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.107475965108392e-05,
+      "loss": 0.5101,
+      "step": 11230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1073378218531766e-05,
+      "loss": 0.4712,
+      "step": 11231
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1071996765256667e-05,
+      "loss": 0.464,
+      "step": 11232
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1070615291285303e-05,
+      "loss": 0.4951,
+      "step": 11233
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1069233796644334e-05,
+      "loss": 0.4821,
+      "step": 11234
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1067852281360443e-05,
+      "loss": 0.4816,
+      "step": 11235
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106647074546029e-05,
+      "loss": 0.4996,
+      "step": 11236
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065089188970557e-05,
+      "loss": 0.4918,
+      "step": 11237
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106370761191791e-05,
+      "loss": 0.4862,
+      "step": 11238
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106232601432903e-05,
+      "loss": 0.5168,
+      "step": 11239
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1060944396230583e-05,
+      "loss": 0.5276,
+      "step": 11240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1059562757649243e-05,
+      "loss": 0.4965,
+      "step": 11241
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058181098611691e-05,
+      "loss": 0.4692,
+      "step": 11242
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1056799419144592e-05,
+      "loss": 0.4981,
+      "step": 11243
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055417719274633e-05,
+      "loss": 0.4909,
+      "step": 11244
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1054035999028478e-05,
+      "loss": 0.4945,
+      "step": 11245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1052654258432813e-05,
+      "loss": 0.4719,
+      "step": 11246
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1051272497514302e-05,
+      "loss": 0.4904,
+      "step": 11247
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1049890716299636e-05,
+      "loss": 0.4752,
+      "step": 11248
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048508914815481e-05,
+      "loss": 0.4838,
+      "step": 11249
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1047127093088521e-05,
+      "loss": 0.4752,
+      "step": 11250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045745251145428e-05,
+      "loss": 0.4907,
+      "step": 11251
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1044363389012887e-05,
+      "loss": 0.4802,
+      "step": 11252
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104298150671757e-05,
+      "loss": 0.4955,
+      "step": 11253
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104159960428616e-05,
+      "loss": 0.486,
+      "step": 11254
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1040217681745335e-05,
+      "loss": 0.4638,
+      "step": 11255
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1038835739121779e-05,
+      "loss": 0.4931,
+      "step": 11256
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1037453776442164e-05,
+      "loss": 0.4842,
+      "step": 11257
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1036071793733177e-05,
+      "loss": 0.4682,
+      "step": 11258
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034689791021495e-05,
+      "loss": 0.4603,
+      "step": 11259
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033307768333806e-05,
+      "loss": 0.5243,
+      "step": 11260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1031925725696782e-05,
+      "loss": 0.4717,
+      "step": 11261
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1030543663137112e-05,
+      "loss": 0.4601,
+      "step": 11262
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029161580681478e-05,
+      "loss": 0.5305,
+      "step": 11263
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102777947835656e-05,
+      "loss": 0.4866,
+      "step": 11264
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102639735618904e-05,
+      "loss": 0.4826,
+      "step": 11265
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1025015214205608e-05,
+      "loss": 0.4847,
+      "step": 11266
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023633052432945e-05,
+      "loss": 0.4395,
+      "step": 11267
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022250870897732e-05,
+      "loss": 0.4594,
+      "step": 11268
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1020868669626659e-05,
+      "loss": 0.4807,
+      "step": 11269
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1019486448646407e-05,
+      "loss": 0.4664,
+      "step": 11270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1018104207983665e-05,
+      "loss": 0.4899,
+      "step": 11271
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016721947665115e-05,
+      "loss": 0.4678,
+      "step": 11272
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015339667717446e-05,
+      "loss": 0.4772,
+      "step": 11273
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1013957368167343e-05,
+      "loss": 0.4756,
+      "step": 11274
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1012575049041499e-05,
+      "loss": 0.5098,
+      "step": 11275
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1011192710366592e-05,
+      "loss": 0.4974,
+      "step": 11276
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009810352169316e-05,
+      "loss": 0.4944,
+      "step": 11277
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1008427974476357e-05,
+      "loss": 0.485,
+      "step": 11278
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1007045577314409e-05,
+      "loss": 0.499,
+      "step": 11279
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1005663160710152e-05,
+      "loss": 0.4765,
+      "step": 11280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1004280724690277e-05,
+      "loss": 0.4785,
+      "step": 11281
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002898269281481e-05,
+      "loss": 0.4874,
+      "step": 11282
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1001515794510449e-05,
+      "loss": 0.4776,
+      "step": 11283
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1000133300403869e-05,
+      "loss": 0.4862,
+      "step": 11284
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0998750786988435e-05,
+      "loss": 0.477,
+      "step": 11285
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099736825429084e-05,
+      "loss": 0.4911,
+      "step": 11286
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995985702337773e-05,
+      "loss": 0.4994,
+      "step": 11287
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0994603131155925e-05,
+      "loss": 0.5014,
+      "step": 11288
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099322054077199e-05,
+      "loss": 0.4755,
+      "step": 11289
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099183793121266e-05,
+      "loss": 0.4803,
+      "step": 11290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099045530250463e-05,
+      "loss": 0.4982,
+      "step": 11291
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0989072654674589e-05,
+      "loss": 0.4465,
+      "step": 11292
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0987689987749235e-05,
+      "loss": 0.5052,
+      "step": 11293
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986307301755261e-05,
+      "loss": 0.4917,
+      "step": 11294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098492459671936e-05,
+      "loss": 0.4283,
+      "step": 11295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0983541872668228e-05,
+      "loss": 0.4743,
+      "step": 11296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0982159129628563e-05,
+      "loss": 0.4718,
+      "step": 11297
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0980776367627057e-05,
+      "loss": 0.5129,
+      "step": 11298
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0979393586690403e-05,
+      "loss": 0.4735,
+      "step": 11299
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0978010786845306e-05,
+      "loss": 0.4904,
+      "step": 11300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0976627968118457e-05,
+      "loss": 0.4954,
+      "step": 11301
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0975245130536555e-05,
+      "loss": 0.4816,
+      "step": 11302
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0973862274126295e-05,
+      "loss": 0.5164,
+      "step": 11303
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0972479398914378e-05,
+      "loss": 0.5332,
+      "step": 11304
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971096504927498e-05,
+      "loss": 0.5077,
+      "step": 11305
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0969713592192356e-05,
+      "loss": 0.4762,
+      "step": 11306
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0968330660735655e-05,
+      "loss": 0.4902,
+      "step": 11307
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0966947710584086e-05,
+      "loss": 0.4773,
+      "step": 11308
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0965564741764357e-05,
+      "loss": 0.4916,
+      "step": 11309
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964181754303159e-05,
+      "loss": 0.4654,
+      "step": 11310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0962798748227202e-05,
+      "loss": 0.4774,
+      "step": 11311
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0961415723563178e-05,
+      "loss": 0.4945,
+      "step": 11312
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960032680337796e-05,
+      "loss": 0.5044,
+      "step": 11313
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0958649618577747e-05,
+      "loss": 0.5159,
+      "step": 11314
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0957266538309748e-05,
+      "loss": 0.4808,
+      "step": 11315
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955883439560487e-05,
+      "loss": 0.4859,
+      "step": 11316
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0954500322356672e-05,
+      "loss": 0.4686,
+      "step": 11317
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0953117186725007e-05,
+      "loss": 0.5104,
+      "step": 11318
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0951734032692194e-05,
+      "loss": 0.4853,
+      "step": 11319
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0950350860284935e-05,
+      "loss": 0.488,
+      "step": 11320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948967669529935e-05,
+      "loss": 0.4959,
+      "step": 11321
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.09475844604539e-05,
+      "loss": 0.5079,
+      "step": 11322
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0946201233083533e-05,
+      "loss": 0.4686,
+      "step": 11323
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944817987445542e-05,
+      "loss": 0.5139,
+      "step": 11324
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943434723566624e-05,
+      "loss": 0.509,
+      "step": 11325
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942051441473493e-05,
+      "loss": 0.4745,
+      "step": 11326
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0940668141192855e-05,
+      "loss": 0.467,
+      "step": 11327
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0939284822751408e-05,
+      "loss": 0.4946,
+      "step": 11328
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937901486175868e-05,
+      "loss": 0.4939,
+      "step": 11329
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093651813149294e-05,
+      "loss": 0.4726,
+      "step": 11330
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0935134758729327e-05,
+      "loss": 0.4767,
+      "step": 11331
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093375136791174e-05,
+      "loss": 0.5283,
+      "step": 11332
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0932367959066885e-05,
+      "loss": 0.4742,
+      "step": 11333
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930984532221477e-05,
+      "loss": 0.4998,
+      "step": 11334
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929601087402217e-05,
+      "loss": 0.5005,
+      "step": 11335
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0928217624635817e-05,
+      "loss": 0.4858,
+      "step": 11336
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0926834143948989e-05,
+      "loss": 0.4571,
+      "step": 11337
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092545064536844e-05,
+      "loss": 0.4991,
+      "step": 11338
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924067128920882e-05,
+      "loss": 0.4864,
+      "step": 11339
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 0.4477,
+      "step": 11340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0921300042531572e-05,
+      "loss": 0.4929,
+      "step": 11341
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091991647264325e-05,
+      "loss": 0.4727,
+      "step": 11342
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091853288499476e-05,
+      "loss": 0.4784,
+      "step": 11343
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917149279612813e-05,
+      "loss": 0.4862,
+      "step": 11344
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0915765656524127e-05,
+      "loss": 0.4835,
+      "step": 11345
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0914382015755412e-05,
+      "loss": 0.4683,
+      "step": 11346
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091299835733338e-05,
+      "loss": 0.4899,
+      "step": 11347
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0911614681284746e-05,
+      "loss": 0.5061,
+      "step": 11348
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0910230987636225e-05,
+      "loss": 0.484,
+      "step": 11349
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908847276414527e-05,
+      "loss": 0.4712,
+      "step": 11350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0907463547646366e-05,
+      "loss": 0.4662,
+      "step": 11351
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090607980135846e-05,
+      "loss": 0.4899,
+      "step": 11352
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0904696037577527e-05,
+      "loss": 0.4654,
+      "step": 11353
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903312256330273e-05,
+      "loss": 0.4895,
+      "step": 11354
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090192845764342e-05,
+      "loss": 0.4766,
+      "step": 11355
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0900544641543685e-05,
+      "loss": 0.5083,
+      "step": 11356
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899160808057783e-05,
+      "loss": 0.4745,
+      "step": 11357
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0897776957212427e-05,
+      "loss": 0.4899,
+      "step": 11358
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0896393089034336e-05,
+      "loss": 0.47,
+      "step": 11359
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0895009203550231e-05,
+      "loss": 0.4664,
+      "step": 11360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0893625300786828e-05,
+      "loss": 0.4536,
+      "step": 11361
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0892241380770838e-05,
+      "loss": 0.4832,
+      "step": 11362
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089085744352899e-05,
+      "loss": 0.46,
+      "step": 11363
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0889473489087999e-05,
+      "loss": 0.4466,
+      "step": 11364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088808951747458e-05,
+      "loss": 0.4958,
+      "step": 11365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0886705528715456e-05,
+      "loss": 0.4879,
+      "step": 11366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0885321522837346e-05,
+      "loss": 0.463,
+      "step": 11367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0883937499866972e-05,
+      "loss": 0.4892,
+      "step": 11368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0882553459831049e-05,
+      "loss": 0.4744,
+      "step": 11369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0881169402756305e-05,
+      "loss": 0.478,
+      "step": 11370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0879785328669454e-05,
+      "loss": 0.4939,
+      "step": 11371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0878401237597223e-05,
+      "loss": 0.5,
+      "step": 11372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0877017129566328e-05,
+      "loss": 0.4634,
+      "step": 11373
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0875633004603497e-05,
+      "loss": 0.4607,
+      "step": 11374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0874248862735445e-05,
+      "loss": 0.4678,
+      "step": 11375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0872864703988903e-05,
+      "loss": 0.4931,
+      "step": 11376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0871480528390585e-05,
+      "loss": 0.4626,
+      "step": 11377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0870096335967226e-05,
+      "loss": 0.4843,
+      "step": 11378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868712126745537e-05,
+      "loss": 0.4485,
+      "step": 11379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0867327900752252e-05,
+      "loss": 0.1797,
+      "step": 11380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0865943658014087e-05,
+      "loss": 0.4946,
+      "step": 11381
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0864559398557774e-05,
+      "loss": 0.4769,
+      "step": 11382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0863175122410033e-05,
+      "loss": 0.4754,
+      "step": 11383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.086179082959759e-05,
+      "loss": 0.4819,
+      "step": 11384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0860406520147171e-05,
+      "loss": 0.4404,
+      "step": 11385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0859022194085505e-05,
+      "loss": 0.4765,
+      "step": 11386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857637851439311e-05,
+      "loss": 0.4706,
+      "step": 11387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0856253492235323e-05,
+      "loss": 0.4709,
+      "step": 11388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854869116500265e-05,
+      "loss": 0.4876,
+      "step": 11389
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0853484724260861e-05,
+      "loss": 0.4954,
+      "step": 11390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0852100315543841e-05,
+      "loss": 0.4917,
+      "step": 11391
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0850715890375934e-05,
+      "loss": 0.5024,
+      "step": 11392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849331448783869e-05,
+      "loss": 0.4752,
+      "step": 11393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.084794699079437e-05,
+      "loss": 0.4802,
+      "step": 11394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846562516434167e-05,
+      "loss": 0.4401,
+      "step": 11395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845178025729992e-05,
+      "loss": 0.492,
+      "step": 11396
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0843793518708576e-05,
+      "loss": 0.4658,
+      "step": 11397
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0842408995396641e-05,
+      "loss": 0.4661,
+      "step": 11398
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841024455820924e-05,
+      "loss": 0.4656,
+      "step": 11399
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.083963990000815e-05,
+      "loss": 0.488,
+      "step": 11400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0838255327985055e-05,
+      "loss": 0.4803,
+      "step": 11401
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0836870739778365e-05,
+      "loss": 0.4526,
+      "step": 11402
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0835486135414812e-05,
+      "loss": 0.4718,
+      "step": 11403
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834101514921132e-05,
+      "loss": 0.4619,
+      "step": 11404
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0832716878324055e-05,
+      "loss": 0.5118,
+      "step": 11405
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831332225650309e-05,
+      "loss": 0.4934,
+      "step": 11406
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082994755692663e-05,
+      "loss": 0.4774,
+      "step": 11407
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082856287217975e-05,
+      "loss": 0.4601,
+      "step": 11408
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0827178171436406e-05,
+      "loss": 0.4823,
+      "step": 11409
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.4953,
+      "step": 11410
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824408722067244e-05,
+      "loss": 0.4833,
+      "step": 11411
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0823023973494898e-05,
+      "loss": 0.5007,
+      "step": 11412
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0821639209033017e-05,
+      "loss": 0.4738,
+      "step": 11413
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0820254428708341e-05,
+      "loss": 0.4818,
+      "step": 11414
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0818869632547602e-05,
+      "loss": 0.467,
+      "step": 11415
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0817484820577537e-05,
+      "loss": 0.4806,
+      "step": 11416
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816099992824878e-05,
+      "loss": 0.497,
+      "step": 11417
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0814715149316363e-05,
+      "loss": 0.4802,
+      "step": 11418
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0813330290078732e-05,
+      "loss": 0.4849,
+      "step": 11419
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0811945415138717e-05,
+      "loss": 0.5071,
+      "step": 11420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0810560524523055e-05,
+      "loss": 0.4836,
+      "step": 11421
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809175618258482e-05,
+      "loss": 0.488,
+      "step": 11422
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807790696371738e-05,
+      "loss": 0.4765,
+      "step": 11423
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0806405758889565e-05,
+      "loss": 0.4826,
+      "step": 11424
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0805020805838688e-05,
+      "loss": 0.4498,
+      "step": 11425
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803635837245858e-05,
+      "loss": 0.4883,
+      "step": 11426
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0802250853137808e-05,
+      "loss": 0.4657,
+      "step": 11427
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.080086585354128e-05,
+      "loss": 0.4969,
+      "step": 11428
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799480838483006e-05,
+      "loss": 0.5282,
+      "step": 11429
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0798095807989735e-05,
+      "loss": 0.4708,
+      "step": 11430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0796710762088201e-05,
+      "loss": 0.4667,
+      "step": 11431
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0795325700805142e-05,
+      "loss": 0.477,
+      "step": 11432
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0793940624167304e-05,
+      "loss": 0.4781,
+      "step": 11433
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0792555532201425e-05,
+      "loss": 0.4825,
+      "step": 11434
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0791170424934248e-05,
+      "loss": 0.4728,
+      "step": 11435
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0789785302392507e-05,
+      "loss": 0.4796,
+      "step": 11436
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0788400164602957e-05,
+      "loss": 0.5047,
+      "step": 11437
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0787015011592328e-05,
+      "loss": 0.4787,
+      "step": 11438
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0785629843387366e-05,
+      "loss": 0.4728,
+      "step": 11439
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784244660014813e-05,
+      "loss": 0.4897,
+      "step": 11440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782859461501413e-05,
+      "loss": 0.4893,
+      "step": 11441
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0781474247873908e-05,
+      "loss": 0.5143,
+      "step": 11442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780089019159044e-05,
+      "loss": 0.4674,
+      "step": 11443
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0778703775383559e-05,
+      "loss": 0.4917,
+      "step": 11444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0777318516574205e-05,
+      "loss": 0.4601,
+      "step": 11445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0775933242757716e-05,
+      "loss": 0.4727,
+      "step": 11446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0774547953960843e-05,
+      "loss": 0.4777,
+      "step": 11447
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0773162650210332e-05,
+      "loss": 0.4848,
+      "step": 11448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771777331532927e-05,
+      "loss": 0.4862,
+      "step": 11449
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0770391997955367e-05,
+      "loss": 0.4764,
+      "step": 11450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0769006649504406e-05,
+      "loss": 0.4672,
+      "step": 11451
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0767621286206786e-05,
+      "loss": 0.4705,
+      "step": 11452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0766235908089253e-05,
+      "loss": 0.5241,
+      "step": 11453
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0764850515178555e-05,
+      "loss": 0.4739,
+      "step": 11454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0763465107501438e-05,
+      "loss": 0.5147,
+      "step": 11455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0762079685084651e-05,
+      "loss": 0.4799,
+      "step": 11456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0760694247954937e-05,
+      "loss": 0.4961,
+      "step": 11457
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0759308796139048e-05,
+      "loss": 0.4859,
+      "step": 11458
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.075792332966373e-05,
+      "loss": 0.4602,
+      "step": 11459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.075653784855573e-05,
+      "loss": 0.4633,
+      "step": 11460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0755152352841798e-05,
+      "loss": 0.5046,
+      "step": 11461
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0753766842548683e-05,
+      "loss": 0.4886,
+      "step": 11462
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0752381317703133e-05,
+      "loss": 0.4685,
+      "step": 11463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0750995778331901e-05,
+      "loss": 0.4772,
+      "step": 11464
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074961022446173e-05,
+      "loss": 0.4652,
+      "step": 11465
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0748224656119372e-05,
+      "loss": 0.4798,
+      "step": 11466
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746839073331583e-05,
+      "loss": 0.4623,
+      "step": 11467
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074545347612511e-05,
+      "loss": 0.4478,
+      "step": 11468
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744067864526697e-05,
+      "loss": 0.4791,
+      "step": 11469
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0742682238563105e-05,
+      "loss": 0.4749,
+      "step": 11470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0741296598261079e-05,
+      "loss": 0.461,
+      "step": 11471
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0739910943647376e-05,
+      "loss": 0.5066,
+      "step": 11472
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073852527474874e-05,
+      "loss": 0.4876,
+      "step": 11473
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073713959159193e-05,
+      "loss": 0.4595,
+      "step": 11474
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735753894203698e-05,
+      "loss": 0.4537,
+      "step": 11475
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734368182610792e-05,
+      "loss": 0.4893,
+      "step": 11476
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0732982456839969e-05,
+      "loss": 0.4455,
+      "step": 11477
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0731596716917978e-05,
+      "loss": 0.4725,
+      "step": 11478
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730210962871579e-05,
+      "loss": 0.4872,
+      "step": 11479
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072882519472752e-05,
+      "loss": 0.4935,
+      "step": 11480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0727439412512557e-05,
+      "loss": 0.483,
+      "step": 11481
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0726053616253444e-05,
+      "loss": 0.4504,
+      "step": 11482
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0724667805976939e-05,
+      "loss": 0.4875,
+      "step": 11483
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723281981709789e-05,
+      "loss": 0.5026,
+      "step": 11484
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0721896143478759e-05,
+      "loss": 0.4762,
+      "step": 11485
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0720510291310595e-05,
+      "loss": 0.494,
+      "step": 11486
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0719124425232059e-05,
+      "loss": 0.4774,
+      "step": 11487
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717738545269902e-05,
+      "loss": 0.4744,
+      "step": 11488
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0716352651450888e-05,
+      "loss": 0.4864,
+      "step": 11489
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0714966743801768e-05,
+      "loss": 0.4716,
+      "step": 11490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713580822349296e-05,
+      "loss": 0.4527,
+      "step": 11491
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0712194887120232e-05,
+      "loss": 0.4872,
+      "step": 11492
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0710808938141337e-05,
+      "loss": 0.4782,
+      "step": 11493
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0709422975439362e-05,
+      "loss": 0.4675,
+      "step": 11494
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708036999041072e-05,
+      "loss": 0.489,
+      "step": 11495
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0706651008973216e-05,
+      "loss": 0.4522,
+      "step": 11496
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070526500526256e-05,
+      "loss": 0.4736,
+      "step": 11497
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0703878987935859e-05,
+      "loss": 0.4769,
+      "step": 11498
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070249295701987e-05,
+      "loss": 0.4523,
+      "step": 11499
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0701106912541357e-05,
+      "loss": 0.4656,
+      "step": 11500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0699720854527078e-05,
+      "loss": 0.4824,
+      "step": 11501
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0698334783003792e-05,
+      "loss": 0.4584,
+      "step": 11502
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0696948697998253e-05,
+      "loss": 0.4842,
+      "step": 11503
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0695562599537232e-05,
+      "loss": 0.483,
+      "step": 11504
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0694176487647482e-05,
+      "loss": 0.4859,
+      "step": 11505
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692790362355766e-05,
+      "loss": 0.4466,
+      "step": 11506
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0691404223688847e-05,
+      "loss": 0.4886,
+      "step": 11507
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690018071673482e-05,
+      "loss": 0.4732,
+      "step": 11508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0688631906336436e-05,
+      "loss": 0.48,
+      "step": 11509
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687245727704465e-05,
+      "loss": 0.4793,
+      "step": 11510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685859535804336e-05,
+      "loss": 0.4595,
+      "step": 11511
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0684473330662815e-05,
+      "loss": 0.4668,
+      "step": 11512
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683087112306656e-05,
+      "loss": 0.507,
+      "step": 11513
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681700880762624e-05,
+      "loss": 0.5304,
+      "step": 11514
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0680314636057486e-05,
+      "loss": 0.4621,
+      "step": 11515
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0678928378218e-05,
+      "loss": 0.4772,
+      "step": 11516
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677542107270935e-05,
+      "loss": 0.4969,
+      "step": 11517
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067615582324305e-05,
+      "loss": 0.4673,
+      "step": 11518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674769526161112e-05,
+      "loss": 0.4561,
+      "step": 11519
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673383216051883e-05,
+      "loss": 0.4615,
+      "step": 11520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671996892942127e-05,
+      "loss": 0.4909,
+      "step": 11521
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.067061055685861e-05,
+      "loss": 0.4858,
+      "step": 11522
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0669224207828102e-05,
+      "loss": 0.4801,
+      "step": 11523
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066783784587736e-05,
+      "loss": 0.4858,
+      "step": 11524
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0666451471033153e-05,
+      "loss": 0.4707,
+      "step": 11525
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0665065083322248e-05,
+      "loss": 0.4904,
+      "step": 11526
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066367868277141e-05,
+      "loss": 0.4943,
+      "step": 11527
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0662292269407404e-05,
+      "loss": 0.4419,
+      "step": 11528
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0660905843256995e-05,
+      "loss": 0.4683,
+      "step": 11529
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0659519404346955e-05,
+      "loss": 0.1789,
+      "step": 11530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0658132952704048e-05,
+      "loss": 0.4485,
+      "step": 11531
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0656746488355042e-05,
+      "loss": 0.4457,
+      "step": 11532
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0655360011326701e-05,
+      "loss": 0.453,
+      "step": 11533
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0653973521645797e-05,
+      "loss": 0.469,
+      "step": 11534
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0652587019339099e-05,
+      "loss": 0.4904,
+      "step": 11535
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.065120050443337e-05,
+      "loss": 0.466,
+      "step": 11536
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.064981397695538e-05,
+      "loss": 0.4669,
+      "step": 11537
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0648427436931904e-05,
+      "loss": 0.4805,
+      "step": 11538
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0647040884389702e-05,
+      "loss": 0.4731,
+      "step": 11539
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0645654319355546e-05,
+      "loss": 0.4978,
+      "step": 11540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0644267741856208e-05,
+      "loss": 0.4754,
+      "step": 11541
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0642881151918457e-05,
+      "loss": 0.1778,
+      "step": 11542
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064149454956906e-05,
+      "loss": 0.4768,
+      "step": 11543
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064010793483479e-05,
+      "loss": 0.1948,
+      "step": 11544
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638721307742414e-05,
+      "loss": 0.4817,
+      "step": 11545
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0637334668318708e-05,
+      "loss": 0.4782,
+      "step": 11546
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0635948016590439e-05,
+      "loss": 0.4785,
+      "step": 11547
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0634561352584378e-05,
+      "loss": 0.4864,
+      "step": 11548
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633174676327298e-05,
+      "loss": 0.4714,
+      "step": 11549
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0631787987845971e-05,
+      "loss": 0.4917,
+      "step": 11550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0630401287167165e-05,
+      "loss": 0.4782,
+      "step": 11551
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0629014574317657e-05,
+      "loss": 0.4796,
+      "step": 11552
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627627849324217e-05,
+      "loss": 0.4623,
+      "step": 11553
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0626241112213618e-05,
+      "loss": 0.4627,
+      "step": 11554
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624854363012629e-05,
+      "loss": 0.4771,
+      "step": 11555
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062346760174803e-05,
+      "loss": 0.4514,
+      "step": 11556
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0622080828446588e-05,
+      "loss": 0.4589,
+      "step": 11557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062069404313508e-05,
+      "loss": 0.4929,
+      "step": 11558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619307245840279e-05,
+      "loss": 0.4664,
+      "step": 11559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0617920436588955e-05,
+      "loss": 0.4815,
+      "step": 11560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061653361540789e-05,
+      "loss": 0.4706,
+      "step": 11561
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0615146782323849e-05,
+      "loss": 0.4904,
+      "step": 11562
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0613759937363617e-05,
+      "loss": 0.4818,
+      "step": 11563
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061237308055396e-05,
+      "loss": 0.4925,
+      "step": 11564
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0610986211921657e-05,
+      "loss": 0.4932,
+      "step": 11565
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060959933149348e-05,
+      "loss": 0.1837,
+      "step": 11566
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060821243929621e-05,
+      "loss": 0.1891,
+      "step": 11567
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060682553535662e-05,
+      "loss": 0.4748,
+      "step": 11568
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0605438619701487e-05,
+      "loss": 0.4546,
+      "step": 11569
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060405169235758e-05,
+      "loss": 0.4757,
+      "step": 11570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0602664753351689e-05,
+      "loss": 0.473,
+      "step": 11571
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0601277802710576e-05,
+      "loss": 0.5043,
+      "step": 11572
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059989084046103e-05,
+      "loss": 0.4737,
+      "step": 11573
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598503866629819e-05,
+      "loss": 0.478,
+      "step": 11574
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0597116881243728e-05,
+      "loss": 0.4634,
+      "step": 11575
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059572988432953e-05,
+      "loss": 0.4787,
+      "step": 11576
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0594342875914e-05,
+      "loss": 0.5182,
+      "step": 11577
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592955856023923e-05,
+      "loss": 0.4868,
+      "step": 11578
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0591568824686073e-05,
+      "loss": 0.4864,
+      "step": 11579
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0590181781927229e-05,
+      "loss": 0.4822,
+      "step": 11580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058879472777417e-05,
+      "loss": 0.4711,
+      "step": 11581
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0587407662253674e-05,
+      "loss": 0.4536,
+      "step": 11582
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0586020585392523e-05,
+      "loss": 0.5081,
+      "step": 11583
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0584633497217492e-05,
+      "loss": 0.5025,
+      "step": 11584
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058324639775536e-05,
+      "loss": 0.4892,
+      "step": 11585
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581859287032914e-05,
+      "loss": 0.473,
+      "step": 11586
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0580472165076928e-05,
+      "loss": 0.4788,
+      "step": 11587
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057908503191418e-05,
+      "loss": 0.4833,
+      "step": 11588
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0577697887571455e-05,
+      "loss": 0.4862,
+      "step": 11589
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0576310732075534e-05,
+      "loss": 0.4845,
+      "step": 11590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574923565453197e-05,
+      "loss": 0.475,
+      "step": 11591
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057353638773122e-05,
+      "loss": 0.5078,
+      "step": 11592
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0572149198936391e-05,
+      "loss": 0.4901,
+      "step": 11593
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057076199909549e-05,
+      "loss": 0.4886,
+      "step": 11594
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569374788235298e-05,
+      "loss": 0.4844,
+      "step": 11595
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0567987566382594e-05,
+      "loss": 0.4921,
+      "step": 11596
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0566600333564163e-05,
+      "loss": 0.4687,
+      "step": 11597
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0565213089806787e-05,
+      "loss": 0.4429,
+      "step": 11598
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056382583513725e-05,
+      "loss": 0.4702,
+      "step": 11599
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056243856958233e-05,
+      "loss": 0.4688,
+      "step": 11600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0561051293168815e-05,
+      "loss": 0.4658,
+      "step": 11601
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0559664005923487e-05,
+      "loss": 0.4857,
+      "step": 11602
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0558276707873128e-05,
+      "loss": 0.468,
+      "step": 11603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0556889399044518e-05,
+      "loss": 0.4605,
+      "step": 11604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055550207946445e-05,
+      "loss": 0.4459,
+      "step": 11605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05541147491597e-05,
+      "loss": 0.4756,
+      "step": 11606
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0552727408157058e-05,
+      "loss": 0.5009,
+      "step": 11607
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05513400564833e-05,
+      "loss": 0.4658,
+      "step": 11608
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0549952694165224e-05,
+      "loss": 0.48,
+      "step": 11609
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05485653212296e-05,
+      "loss": 0.4635,
+      "step": 11610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0547177937703223e-05,
+      "loss": 0.4608,
+      "step": 11611
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545790543612873e-05,
+      "loss": 0.4816,
+      "step": 11612
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.054440313898534e-05,
+      "loss": 0.4802,
+      "step": 11613
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0543015723847402e-05,
+      "loss": 0.4681,
+      "step": 11614
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541628298225856e-05,
+      "loss": 0.48,
+      "step": 11615
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.054024086214748e-05,
+      "loss": 0.4961,
+      "step": 11616
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0538853415639064e-05,
+      "loss": 0.4521,
+      "step": 11617
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0537465958727387e-05,
+      "loss": 0.4832,
+      "step": 11618
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0536078491439245e-05,
+      "loss": 0.4982,
+      "step": 11619
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053469101380142e-05,
+      "loss": 0.4902,
+      "step": 11620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0533303525840703e-05,
+      "loss": 0.4754,
+      "step": 11621
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0531916027583877e-05,
+      "loss": 0.4521,
+      "step": 11622
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053052851905773e-05,
+      "loss": 0.4657,
+      "step": 11623
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529141000289051e-05,
+      "loss": 0.4815,
+      "step": 11624
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0527753471304625e-05,
+      "loss": 0.4927,
+      "step": 11625
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0526365932131245e-05,
+      "loss": 0.5084,
+      "step": 11626
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524978382795695e-05,
+      "loss": 0.4967,
+      "step": 11627
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0523590823324766e-05,
+      "loss": 0.46,
+      "step": 11628
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0522203253745242e-05,
+      "loss": 0.4724,
+      "step": 11629
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520815674083919e-05,
+      "loss": 0.5195,
+      "step": 11630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0519428084367583e-05,
+      "loss": 0.4821,
+      "step": 11631
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518040484623022e-05,
+      "loss": 0.4646,
+      "step": 11632
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051665287487702e-05,
+      "loss": 0.487,
+      "step": 11633
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051526525515638e-05,
+      "loss": 0.4609,
+      "step": 11634
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051387762548788e-05,
+      "loss": 0.4989,
+      "step": 11635
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0512489985898315e-05,
+      "loss": 0.4587,
+      "step": 11636
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511102336414475e-05,
+      "loss": 0.4998,
+      "step": 11637
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0509714677063152e-05,
+      "loss": 0.4643,
+      "step": 11638
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0508327007871133e-05,
+      "loss": 0.4776,
+      "step": 11639
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506939328865207e-05,
+      "loss": 0.4666,
+      "step": 11640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0505551640072169e-05,
+      "loss": 0.4792,
+      "step": 11641
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0504163941518812e-05,
+      "loss": 0.4578,
+      "step": 11642
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0502776233231925e-05,
+      "loss": 0.4505,
+      "step": 11643
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0501388515238293e-05,
+      "loss": 0.455,
+      "step": 11644
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0500000787564718e-05,
+      "loss": 0.4878,
+      "step": 11645
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498613050237987e-05,
+      "loss": 0.438,
+      "step": 11646
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049722530328489e-05,
+      "loss": 0.4777,
+      "step": 11647
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0495837546732224e-05,
+      "loss": 0.4599,
+      "step": 11648
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0494449780606778e-05,
+      "loss": 0.4663,
+      "step": 11649
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0493062004935346e-05,
+      "loss": 0.4611,
+      "step": 11650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491674219744721e-05,
+      "loss": 0.4849,
+      "step": 11651
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0490286425061692e-05,
+      "loss": 0.4738,
+      "step": 11652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0488898620913062e-05,
+      "loss": 0.4494,
+      "step": 11653
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0487510807325615e-05,
+      "loss": 0.4988,
+      "step": 11654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0486122984326146e-05,
+      "loss": 0.4704,
+      "step": 11655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.048473515194145e-05,
+      "loss": 0.4912,
+      "step": 11656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0483347310198324e-05,
+      "loss": 0.4775,
+      "step": 11657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0481959459123557e-05,
+      "loss": 0.509,
+      "step": 11658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0480571598743943e-05,
+      "loss": 0.4659,
+      "step": 11659
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479183729086279e-05,
+      "loss": 0.4697,
+      "step": 11660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0477795850177363e-05,
+      "loss": 0.5,
+      "step": 11661
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476407962043986e-05,
+      "loss": 0.478,
+      "step": 11662
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047502006471294e-05,
+      "loss": 0.4899,
+      "step": 11663
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0473632158211025e-05,
+      "loss": 0.4847,
+      "step": 11664
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472244242565035e-05,
+      "loss": 0.4863,
+      "step": 11665
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0470856317801766e-05,
+      "loss": 0.4563,
+      "step": 11666
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046946838394801e-05,
+      "loss": 0.5014,
+      "step": 11667
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0468080441030568e-05,
+      "loss": 0.5049,
+      "step": 11668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0466692489076232e-05,
+      "loss": 0.4997,
+      "step": 11669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465304528111797e-05,
+      "loss": 0.5099,
+      "step": 11670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0463916558164065e-05,
+      "loss": 0.5015,
+      "step": 11671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0462528579259832e-05,
+      "loss": 0.4781,
+      "step": 11672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461140591425888e-05,
+      "loss": 0.5107,
+      "step": 11673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0459752594689037e-05,
+      "loss": 0.4821,
+      "step": 11674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0458364589076071e-05,
+      "loss": 0.4653,
+      "step": 11675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0456976574613793e-05,
+      "loss": 0.4928,
+      "step": 11676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045558855132899e-05,
+      "loss": 0.4702,
+      "step": 11677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454200519248473e-05,
+      "loss": 0.5035,
+      "step": 11678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452812478399032e-05,
+      "loss": 0.4819,
+      "step": 11679
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0451424428807467e-05,
+      "loss": 0.4633,
+      "step": 11680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0450036370500571e-05,
+      "loss": 0.4797,
+      "step": 11681
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044864830350515e-05,
+      "loss": 0.453,
+      "step": 11682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0447260227847997e-05,
+      "loss": 0.5043,
+      "step": 11683
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0445872143555916e-05,
+      "loss": 0.4724,
+      "step": 11684
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444484050655697e-05,
+      "loss": 0.4567,
+      "step": 11685
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044309594917415e-05,
+      "loss": 0.5021,
+      "step": 11686
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0441707839138064e-05,
+      "loss": 0.4494,
+      "step": 11687
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0440319720574241e-05,
+      "loss": 0.4891,
+      "step": 11688
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0438931593509485e-05,
+      "loss": 0.4745,
+      "step": 11689
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0437543457970591e-05,
+      "loss": 0.4923,
+      "step": 11690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436155313984362e-05,
+      "loss": 0.4649,
+      "step": 11691
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434767161577593e-05,
+      "loss": 0.4816,
+      "step": 11692
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043337900077709e-05,
+      "loss": 0.4745,
+      "step": 11693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0431990831609648e-05,
+      "loss": 0.4644,
+      "step": 11694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0430602654102071e-05,
+      "loss": 0.4611,
+      "step": 11695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0429214468281157e-05,
+      "loss": 0.4853,
+      "step": 11696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042782627417371e-05,
+      "loss": 0.4511,
+      "step": 11697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0426438071806527e-05,
+      "loss": 0.46,
+      "step": 11698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042504986120641e-05,
+      "loss": 0.4373,
+      "step": 11699
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423661642400163e-05,
+      "loss": 0.4686,
+      "step": 11700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0422273415414586e-05,
+      "loss": 0.492,
+      "step": 11701
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0420885180276476e-05,
+      "loss": 0.4713,
+      "step": 11702
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041949693701264e-05,
+      "loss": 0.5009,
+      "step": 11703
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0418108685649879e-05,
+      "loss": 0.4841,
+      "step": 11704
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0416720426214996e-05,
+      "loss": 0.4777,
+      "step": 11705
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041533215873479e-05,
+      "loss": 0.4758,
+      "step": 11706
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413943883236063e-05,
+      "loss": 0.4966,
+      "step": 11707
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412555599745623e-05,
+      "loss": 0.4771,
+      "step": 11708
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0411167308290266e-05,
+      "loss": 0.4938,
+      "step": 11709
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.04097790088968e-05,
+      "loss": 0.5043,
+      "step": 11710
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408390701592021e-05,
+      "loss": 0.4733,
+      "step": 11711
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040700238640274e-05,
+      "loss": 0.4362,
+      "step": 11712
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0405614063355756e-05,
+      "loss": 0.4412,
+      "step": 11713
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0404225732477871e-05,
+      "loss": 0.4429,
+      "step": 11714
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402837393795891e-05,
+      "loss": 0.4644,
+      "step": 11715
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0401449047336622e-05,
+      "loss": 0.4759,
+      "step": 11716
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040006069312686e-05,
+      "loss": 0.4634,
+      "step": 11717
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398672331193417e-05,
+      "loss": 0.4799,
+      "step": 11718
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0397283961563094e-05,
+      "loss": 0.4754,
+      "step": 11719
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0395895584262696e-05,
+      "loss": 0.4681,
+      "step": 11720
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0394507199319025e-05,
+      "loss": 0.5202,
+      "step": 11721
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0393118806758888e-05,
+      "loss": 0.4494,
+      "step": 11722
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0391730406609089e-05,
+      "loss": 0.478,
+      "step": 11723
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0390341998896434e-05,
+      "loss": 0.4763,
+      "step": 11724
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388953583647725e-05,
+      "loss": 0.4687,
+      "step": 11725
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0387565160889768e-05,
+      "loss": 0.4985,
+      "step": 11726
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0386176730649368e-05,
+      "loss": 0.4458,
+      "step": 11727
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0384788292953336e-05,
+      "loss": 0.4819,
+      "step": 11728
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038339984782847e-05,
+      "loss": 0.4477,
+      "step": 11729
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038201139530158e-05,
+      "loss": 0.4444,
+      "step": 11730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380622935399474e-05,
+      "loss": 0.4734,
+      "step": 11731
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0379234468148949e-05,
+      "loss": 0.4903,
+      "step": 11732
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0377845993576819e-05,
+      "loss": 0.4915,
+      "step": 11733
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0376457511709888e-05,
+      "loss": 0.4793,
+      "step": 11734
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375069022574965e-05,
+      "loss": 0.4711,
+      "step": 11735
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037368052619885e-05,
+      "loss": 0.5189,
+      "step": 11736
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0372292022608355e-05,
+      "loss": 0.4666,
+      "step": 11737
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0370903511830286e-05,
+      "loss": 0.4618,
+      "step": 11738
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0369514993891451e-05,
+      "loss": 0.4735,
+      "step": 11739
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0368126468818656e-05,
+      "loss": 0.4837,
+      "step": 11740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366737936638704e-05,
+      "loss": 0.4534,
+      "step": 11741
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036534939737841e-05,
+      "loss": 0.4753,
+      "step": 11742
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363960851064578e-05,
+      "loss": 0.4894,
+      "step": 11743
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362572297724012e-05,
+      "loss": 0.4606,
+      "step": 11744
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0361183737383526e-05,
+      "loss": 0.4681,
+      "step": 11745
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0359795170069926e-05,
+      "loss": 0.4791,
+      "step": 11746
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035840659581002e-05,
+      "loss": 0.4803,
+      "step": 11747
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035701801463061e-05,
+      "loss": 0.4767,
+      "step": 11748
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0355629426558519e-05,
+      "loss": 0.4741,
+      "step": 11749
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354240831620542e-05,
+      "loss": 0.456,
+      "step": 11750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0352852229843489e-05,
+      "loss": 0.4718,
+      "step": 11751
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0351463621254177e-05,
+      "loss": 0.4831,
+      "step": 11752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0350075005879406e-05,
+      "loss": 0.4882,
+      "step": 11753
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034868638374599e-05,
+      "loss": 0.4745,
+      "step": 11754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347297754880735e-05,
+      "loss": 0.4748,
+      "step": 11755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345909119310454e-05,
+      "loss": 0.479,
+      "step": 11756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0344520477061953e-05,
+      "loss": 0.485,
+      "step": 11757
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0343131828162045e-05,
+      "loss": 0.4714,
+      "step": 11758
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0341743172637533e-05,
+      "loss": 0.5031,
+      "step": 11759
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0340354510515235e-05,
+      "loss": 0.4421,
+      "step": 11760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338965841821958e-05,
+      "loss": 0.4962,
+      "step": 11761
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0337577166584509e-05,
+      "loss": 0.4494,
+      "step": 11762
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.03361884848297e-05,
+      "loss": 0.4687,
+      "step": 11763
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0334799796584345e-05,
+      "loss": 0.4914,
+      "step": 11764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033341110187525e-05,
+      "loss": 0.468,
+      "step": 11765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332022400729225e-05,
+      "loss": 0.494,
+      "step": 11766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0330633693173083e-05,
+      "loss": 0.4517,
+      "step": 11767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329244979233634e-05,
+      "loss": 0.4924,
+      "step": 11768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.032785625893769e-05,
+      "loss": 0.4812,
+      "step": 11769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0326467532312058e-05,
+      "loss": 0.4541,
+      "step": 11770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0325078799383556e-05,
+      "loss": 0.4791,
+      "step": 11771
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.032369006017899e-05,
+      "loss": 0.4916,
+      "step": 11772
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0322301314725172e-05,
+      "loss": 0.4864,
+      "step": 11773
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0320912563048915e-05,
+      "loss": 0.4739,
+      "step": 11774
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0319523805177029e-05,
+      "loss": 0.4918,
+      "step": 11775
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0318135041136327e-05,
+      "loss": 0.4827,
+      "step": 11776
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0316746270953622e-05,
+      "loss": 0.5036,
+      "step": 11777
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0315357494655722e-05,
+      "loss": 0.4835,
+      "step": 11778
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0313968712269444e-05,
+      "loss": 0.5295,
+      "step": 11779
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0312579923821598e-05,
+      "loss": 0.4768,
+      "step": 11780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0311191129338993e-05,
+      "loss": 0.5309,
+      "step": 11781
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0309802328848446e-05,
+      "loss": 0.4891,
+      "step": 11782
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030841352237677e-05,
+      "loss": 0.4722,
+      "step": 11783
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0307024709950775e-05,
+      "loss": 0.4642,
+      "step": 11784
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305635891597274e-05,
+      "loss": 0.4841,
+      "step": 11785
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030424706734308e-05,
+      "loss": 0.489,
+      "step": 11786
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030285823721501e-05,
+      "loss": 0.471,
+      "step": 11787
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0301469401239873e-05,
+      "loss": 0.4713,
+      "step": 11788
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0300080559444479e-05,
+      "loss": 0.4727,
+      "step": 11789
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029869171185565e-05,
+      "loss": 0.4774,
+      "step": 11790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0297302858500194e-05,
+      "loss": 0.484,
+      "step": 11791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295913999404923e-05,
+      "loss": 0.4563,
+      "step": 11792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0294525134596654e-05,
+      "loss": 0.4664,
+      "step": 11793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293136264102201e-05,
+      "loss": 0.4673,
+      "step": 11794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0291747387948376e-05,
+      "loss": 0.4685,
+      "step": 11795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0290358506161994e-05,
+      "loss": 0.4761,
+      "step": 11796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0288969618769871e-05,
+      "loss": 0.472,
+      "step": 11797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.028758072579882e-05,
+      "loss": 0.4304,
+      "step": 11798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0286191827275652e-05,
+      "loss": 0.502,
+      "step": 11799
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284802923227183e-05,
+      "loss": 0.4973,
+      "step": 11800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0283414013680233e-05,
+      "loss": 0.4434,
+      "step": 11801
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282025098661613e-05,
+      "loss": 0.4947,
+      "step": 11802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0280636178198134e-05,
+      "loss": 0.4839,
+      "step": 11803
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279247252316614e-05,
+      "loss": 0.4668,
+      "step": 11804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0277858321043868e-05,
+      "loss": 0.4668,
+      "step": 11805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276469384406714e-05,
+      "loss": 0.4798,
+      "step": 11806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0275080442431959e-05,
+      "loss": 0.4616,
+      "step": 11807
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0273691495146428e-05,
+      "loss": 0.493,
+      "step": 11808
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272302542576931e-05,
+      "loss": 0.4668,
+      "step": 11809
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270913584750287e-05,
+      "loss": 0.4755,
+      "step": 11810
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0269524621693302e-05,
+      "loss": 0.487,
+      "step": 11811
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0268135653432806e-05,
+      "loss": 0.4887,
+      "step": 11812
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0266746679995608e-05,
+      "loss": 0.5015,
+      "step": 11813
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0265357701408521e-05,
+      "loss": 0.4873,
+      "step": 11814
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0263968717698365e-05,
+      "loss": 0.4799,
+      "step": 11815
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262579728891954e-05,
+      "loss": 0.5006,
+      "step": 11816
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0261190735016106e-05,
+      "loss": 0.4647,
+      "step": 11817
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0259801736097634e-05,
+      "loss": 0.4633,
+      "step": 11818
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0258412732163358e-05,
+      "loss": 0.4802,
+      "step": 11819
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257023723240093e-05,
+      "loss": 0.4835,
+      "step": 11820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255634709354659e-05,
+      "loss": 0.4675,
+      "step": 11821
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254245690533863e-05,
+      "loss": 0.4852,
+      "step": 11822
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0252856666804534e-05,
+      "loss": 0.4907,
+      "step": 11823
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0251467638193479e-05,
+      "loss": 0.4676,
+      "step": 11824
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0250078604727523e-05,
+      "loss": 0.4888,
+      "step": 11825
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0248689566433477e-05,
+      "loss": 0.4901,
+      "step": 11826
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0247300523338162e-05,
+      "loss": 0.4794,
+      "step": 11827
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024591147546839e-05,
+      "loss": 0.4952,
+      "step": 11828
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244522422850987e-05,
+      "loss": 0.5016,
+      "step": 11829
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024313336551276e-05,
+      "loss": 0.4708,
+      "step": 11830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0241744303480539e-05,
+      "loss": 0.5001,
+      "step": 11831
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0240355236781128e-05,
+      "loss": 0.4858,
+      "step": 11832
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0238966165441354e-05,
+      "loss": 0.4927,
+      "step": 11833
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0237577089488032e-05,
+      "loss": 0.4802,
+      "step": 11834
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023618800894798e-05,
+      "loss": 0.4764,
+      "step": 11835
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234798923848016e-05,
+      "loss": 0.5034,
+      "step": 11836
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0233409834214956e-05,
+      "loss": 0.4549,
+      "step": 11837
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0232020740075625e-05,
+      "loss": 0.4648,
+      "step": 11838
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0230631641456833e-05,
+      "loss": 0.4766,
+      "step": 11839
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.02292425383854e-05,
+      "loss": 0.4926,
+      "step": 11840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227853430888148e-05,
+      "loss": 0.5128,
+      "step": 11841
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226464318991896e-05,
+      "loss": 0.4727,
+      "step": 11842
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0225075202723458e-05,
+      "loss": 0.4543,
+      "step": 11843
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0223686082109655e-05,
+      "loss": 0.4711,
+      "step": 11844
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0222296957177305e-05,
+      "loss": 0.4835,
+      "step": 11845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022090782795323e-05,
+      "loss": 0.4646,
+      "step": 11846
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0219518694464243e-05,
+      "loss": 0.4694,
+      "step": 11847
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0218129556737169e-05,
+      "loss": 0.4636,
+      "step": 11848
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0216740414798822e-05,
+      "loss": 0.492,
+      "step": 11849
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0215351268676028e-05,
+      "loss": 0.4421,
+      "step": 11850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0213962118395599e-05,
+      "loss": 0.4839,
+      "step": 11851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0212572963984358e-05,
+      "loss": 0.454,
+      "step": 11852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0211183805469122e-05,
+      "loss": 0.4731,
+      "step": 11853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0209794642876713e-05,
+      "loss": 0.4813,
+      "step": 11854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0208405476233952e-05,
+      "loss": 0.4905,
+      "step": 11855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0207016305567652e-05,
+      "loss": 0.4668,
+      "step": 11856
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0205627130904642e-05,
+      "loss": 0.4766,
+      "step": 11857
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0204237952271733e-05,
+      "loss": 0.4674,
+      "step": 11858
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020284876969575e-05,
+      "loss": 0.4998,
+      "step": 11859
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0201459583203512e-05,
+      "loss": 0.484,
+      "step": 11860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020007039282184e-05,
+      "loss": 0.4727,
+      "step": 11861
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019868119857755e-05,
+      "loss": 0.4778,
+      "step": 11862
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197292000497465e-05,
+      "loss": 0.4835,
+      "step": 11863
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0195902798608406e-05,
+      "loss": 0.4781,
+      "step": 11864
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0194513592937196e-05,
+      "loss": 0.4843,
+      "step": 11865
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0193124383510648e-05,
+      "loss": 0.4643,
+      "step": 11866
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191735170355585e-05,
+      "loss": 0.451,
+      "step": 11867
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0190345953498832e-05,
+      "loss": 0.4551,
+      "step": 11868
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188956732967208e-05,
+      "loss": 0.479,
+      "step": 11869
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187567508787526e-05,
+      "loss": 0.475,
+      "step": 11870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018617828098662e-05,
+      "loss": 0.5071,
+      "step": 11871
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01847890495913e-05,
+      "loss": 0.4797,
+      "step": 11872
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0183399814628394e-05,
+      "loss": 0.4995,
+      "step": 11873
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182010576124715e-05,
+      "loss": 0.4783,
+      "step": 11874
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018062133410709e-05,
+      "loss": 0.4517,
+      "step": 11875
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017923208860234e-05,
+      "loss": 0.4957,
+      "step": 11876
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0177842839637287e-05,
+      "loss": 0.4633,
+      "step": 11877
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176453587238748e-05,
+      "loss": 0.478,
+      "step": 11878
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0175064331433547e-05,
+      "loss": 0.4642,
+      "step": 11879
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173675072248504e-05,
+      "loss": 0.4864,
+      "step": 11880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0172285809710438e-05,
+      "loss": 0.4774,
+      "step": 11881
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0170896543846181e-05,
+      "loss": 0.4848,
+      "step": 11882
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169507274682541e-05,
+      "loss": 0.4781,
+      "step": 11883
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.016811800224635e-05,
+      "loss": 0.4745,
+      "step": 11884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166728726564424e-05,
+      "loss": 0.4439,
+      "step": 11885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0165339447663586e-05,
+      "loss": 0.4585,
+      "step": 11886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.016395016557066e-05,
+      "loss": 0.4598,
+      "step": 11887
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162560880312462e-05,
+      "loss": 0.4897,
+      "step": 11888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161171591915822e-05,
+      "loss": 0.4639,
+      "step": 11889
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0159782300407556e-05,
+      "loss": 0.4842,
+      "step": 11890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0158393005814492e-05,
+      "loss": 0.4689,
+      "step": 11891
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0157003708163442e-05,
+      "loss": 0.4401,
+      "step": 11892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155614407481238e-05,
+      "loss": 0.4457,
+      "step": 11893
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01542251037947e-05,
+      "loss": 0.485,
+      "step": 11894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152835797130644e-05,
+      "loss": 0.184,
+      "step": 11895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01514464875159e-05,
+      "loss": 0.4999,
+      "step": 11896
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150057174977287e-05,
+      "loss": 0.4548,
+      "step": 11897
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014866785954163e-05,
+      "loss": 0.4926,
+      "step": 11898
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0147278541235747e-05,
+      "loss": 0.4822,
+      "step": 11899
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0145889220086465e-05,
+      "loss": 0.4799,
+      "step": 11900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144499896120604e-05,
+      "loss": 0.4685,
+      "step": 11901
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143110569364987e-05,
+      "loss": 0.4496,
+      "step": 11902
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0141721239846436e-05,
+      "loss": 0.484,
+      "step": 11903
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0140331907591776e-05,
+      "loss": 0.5147,
+      "step": 11904
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013894257262783e-05,
+      "loss": 0.4646,
+      "step": 11905
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013755323498142e-05,
+      "loss": 0.4848,
+      "step": 11906
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0136163894679365e-05,
+      "loss": 0.459,
+      "step": 11907
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0134774551748494e-05,
+      "loss": 0.4784,
+      "step": 11908
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133385206215628e-05,
+      "loss": 0.4703,
+      "step": 11909
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013199585810759e-05,
+      "loss": 0.4701,
+      "step": 11910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0130606507451202e-05,
+      "loss": 0.4722,
+      "step": 11911
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129217154273288e-05,
+      "loss": 0.4581,
+      "step": 11912
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127827798600673e-05,
+      "loss": 0.5056,
+      "step": 11913
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0126438440460178e-05,
+      "loss": 0.4776,
+      "step": 11914
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0125049079878623e-05,
+      "loss": 0.4575,
+      "step": 11915
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0123659716882839e-05,
+      "loss": 0.4335,
+      "step": 11916
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0122270351499648e-05,
+      "loss": 0.4753,
+      "step": 11917
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0120880983755868e-05,
+      "loss": 0.4538,
+      "step": 11918
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119491613678325e-05,
+      "loss": 0.4685,
+      "step": 11919
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118102241293848e-05,
+      "loss": 0.5045,
+      "step": 11920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011671286662925e-05,
+      "loss": 0.479,
+      "step": 11921
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0115323489711364e-05,
+      "loss": 0.5026,
+      "step": 11922
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011393411056701e-05,
+      "loss": 0.4772,
+      "step": 11923
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112544729223013e-05,
+      "loss": 0.4831,
+      "step": 11924
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0111155345706193e-05,
+      "loss": 0.1727,
+      "step": 11925
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109765960043379e-05,
+      "loss": 0.4894,
+      "step": 11926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.010837657226139e-05,
+      "loss": 0.466,
+      "step": 11927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0106987182387055e-05,
+      "loss": 0.4975,
+      "step": 11928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0105597790447195e-05,
+      "loss": 0.4609,
+      "step": 11929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0104208396468634e-05,
+      "loss": 0.4914,
+      "step": 11930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0102819000478194e-05,
+      "loss": 0.4964,
+      "step": 11931
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101429602502708e-05,
+      "loss": 0.473,
+      "step": 11932
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0100040202568986e-05,
+      "loss": 0.4752,
+      "step": 11933
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098650800703864e-05,
+      "loss": 0.4729,
+      "step": 11934
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097261396934161e-05,
+      "loss": 0.4927,
+      "step": 11935
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0095871991286705e-05,
+      "loss": 0.446,
+      "step": 11936
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0094482583788311e-05,
+      "loss": 0.4596,
+      "step": 11937
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093093174465816e-05,
+      "loss": 0.4614,
+      "step": 11938
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0091703763346035e-05,
+      "loss": 0.499,
+      "step": 11939
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0090314350455796e-05,
+      "loss": 0.4829,
+      "step": 11940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0088924935821922e-05,
+      "loss": 0.471,
+      "step": 11941
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0087535519471241e-05,
+      "loss": 0.4961,
+      "step": 11942
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0086146101430573e-05,
+      "loss": 0.4685,
+      "step": 11943
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0084756681726741e-05,
+      "loss": 0.463,
+      "step": 11944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083367260386578e-05,
+      "loss": 0.4828,
+      "step": 11945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00819778374369e-05,
+      "loss": 0.4763,
+      "step": 11946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080588412904537e-05,
+      "loss": 0.4422,
+      "step": 11947
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079198986816307e-05,
+      "loss": 0.4621,
+      "step": 11948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0077809559199043e-05,
+      "loss": 0.4481,
+      "step": 11949
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076420130079565e-05,
+      "loss": 0.4754,
+      "step": 11950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00750306994847e-05,
+      "loss": 0.5066,
+      "step": 11951
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0073641267441268e-05,
+      "loss": 0.4822,
+      "step": 11952
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0072251833976098e-05,
+      "loss": 0.4785,
+      "step": 11953
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070862399116016e-05,
+      "loss": 0.4776,
+      "step": 11954
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069472962887843e-05,
+      "loss": 0.4582,
+      "step": 11955
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0068083525318404e-05,
+      "loss": 0.4697,
+      "step": 11956
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006669408643453e-05,
+      "loss": 0.4769,
+      "step": 11957
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0065304646263036e-05,
+      "loss": 0.4721,
+      "step": 11958
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063915204830754e-05,
+      "loss": 0.4615,
+      "step": 11959
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0062525762164507e-05,
+      "loss": 0.4766,
+      "step": 11960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0061136318291122e-05,
+      "loss": 0.4559,
+      "step": 11961
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005974687323742e-05,
+      "loss": 0.489,
+      "step": 11962
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0058357427030228e-05,
+      "loss": 0.4676,
+      "step": 11963
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056967979696372e-05,
+      "loss": 0.4618,
+      "step": 11964
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0055578531262677e-05,
+      "loss": 0.4616,
+      "step": 11965
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054189081755965e-05,
+      "loss": 0.4716,
+      "step": 11966
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0052799631203066e-05,
+      "loss": 0.4904,
+      "step": 11967
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0051410179630801e-05,
+      "loss": 0.4557,
+      "step": 11968
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050020727065997e-05,
+      "loss": 0.4675,
+      "step": 11969
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0048631273535476e-05,
+      "loss": 0.5179,
+      "step": 11970
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0047241819066069e-05,
+      "loss": 0.4631,
+      "step": 11971
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00458523636846e-05,
+      "loss": 0.4673,
+      "step": 11972
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044462907417888e-05,
+      "loss": 0.4655,
+      "step": 11973
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043073450292766e-05,
+      "loss": 0.4727,
+      "step": 11974
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041683992336053e-05,
+      "loss": 0.4712,
+      "step": 11975
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004029453357458e-05,
+      "loss": 0.4594,
+      "step": 11976
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038905074035167e-05,
+      "loss": 0.4609,
+      "step": 11977
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0037515613744643e-05,
+      "loss": 0.4856,
+      "step": 11978
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036126152729832e-05,
+      "loss": 0.4938,
+      "step": 11979
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003473669101756e-05,
+      "loss": 0.4819,
+      "step": 11980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0033347228634651e-05,
+      "loss": 0.4685,
+      "step": 11981
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003195776560793e-05,
+      "loss": 0.4747,
+      "step": 11982
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0030568301964223e-05,
+      "loss": 0.4591,
+      "step": 11983
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002917883773036e-05,
+      "loss": 0.5026,
+      "step": 11984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002778937293316e-05,
+      "loss": 0.4711,
+      "step": 11985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026399907599447e-05,
+      "loss": 0.4608,
+      "step": 11986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0025010441756056e-05,
+      "loss": 0.4707,
+      "step": 11987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0023620975429803e-05,
+      "loss": 0.4847,
+      "step": 11988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0022231508647515e-05,
+      "loss": 0.5141,
+      "step": 11989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002084204143602e-05,
+      "loss": 0.4515,
+      "step": 11990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019452573822146e-05,
+      "loss": 0.4487,
+      "step": 11991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0018063105832711e-05,
+      "loss": 0.4661,
+      "step": 11992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0016673637494548e-05,
+      "loss": 0.4479,
+      "step": 11993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015284168834476e-05,
+      "loss": 0.4722,
+      "step": 11994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013894699879327e-05,
+      "loss": 0.4975,
+      "step": 11995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001250523065592e-05,
+      "loss": 0.489,
+      "step": 11996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0011115761191085e-05,
+      "loss": 0.4562,
+      "step": 11997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009726291511645e-05,
+      "loss": 0.4814,
+      "step": 11998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000833682164443e-05,
+      "loss": 0.4916,
+      "step": 11999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0006947351616256e-05,
+      "loss": 0.4534,
+      "step": 12000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000555788145396e-05,
+      "loss": 0.4594,
+      "step": 12001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0004168411184359e-05,
+      "loss": 0.4728,
+      "step": 12002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0002778940834284e-05,
+      "loss": 0.4977,
+      "step": 12003
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0001389470430554e-05,
+      "loss": 0.4758,
+      "step": 12004
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.4868,
+      "step": 12005
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.998610529569447e-06,
+      "loss": 0.4838,
+      "step": 12006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.997221059165721e-06,
+      "loss": 0.4916,
+      "step": 12007
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.995831588815644e-06,
+      "loss": 0.4793,
+      "step": 12008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994442118546042e-06,
+      "loss": 0.4954,
+      "step": 12009
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993052648383744e-06,
+      "loss": 0.4389,
+      "step": 12010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.991663178355576e-06,
+      "loss": 0.4766,
+      "step": 12011
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990273708488357e-06,
+      "loss": 0.4787,
+      "step": 12012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.988884238808916e-06,
+      "loss": 0.4749,
+      "step": 12013
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987494769344082e-06,
+      "loss": 0.471,
+      "step": 12014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986105300120678e-06,
+      "loss": 0.4666,
+      "step": 12015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984715831165525e-06,
+      "loss": 0.4749,
+      "step": 12016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.983326362505455e-06,
+      "loss": 0.4736,
+      "step": 12017
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98193689416729e-06,
+      "loss": 0.4811,
+      "step": 12018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980547426177859e-06,
+      "loss": 0.524,
+      "step": 12019
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.979157958563983e-06,
+      "loss": 0.4788,
+      "step": 12020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977768491352488e-06,
+      "loss": 0.4885,
+      "step": 12021
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.976379024570202e-06,
+      "loss": 0.4802,
+      "step": 12022
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97498955824395e-06,
+      "loss": 0.4656,
+      "step": 12023
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973600092400556e-06,
+      "loss": 0.4591,
+      "step": 12024
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972210627066846e-06,
+      "loss": 0.4958,
+      "step": 12025
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970821162269644e-06,
+      "loss": 0.4831,
+      "step": 12026
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969431698035779e-06,
+      "loss": 0.4886,
+      "step": 12027
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968042234392072e-06,
+      "loss": 0.4547,
+      "step": 12028
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.966652771365352e-06,
+      "loss": 0.4905,
+      "step": 12029
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965263308982445e-06,
+      "loss": 0.472,
+      "step": 12030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.963873847270171e-06,
+      "loss": 0.5112,
+      "step": 12031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96248438625536e-06,
+      "loss": 0.4342,
+      "step": 12032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961094925964834e-06,
+      "loss": 0.4782,
+      "step": 12033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.959705466425424e-06,
+      "loss": 0.495,
+      "step": 12034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95831600766395e-06,
+      "loss": 0.4723,
+      "step": 12035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956926549707238e-06,
+      "loss": 0.4826,
+      "step": 12036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955537092582113e-06,
+      "loss": 0.4927,
+      "step": 12037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954147636315406e-06,
+      "loss": 0.473,
+      "step": 12038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.952758180933933e-06,
+      "loss": 0.485,
+      "step": 12039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951368726464526e-06,
+      "loss": 0.4678,
+      "step": 12040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949979272934004e-06,
+      "loss": 0.4687,
+      "step": 12041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.948589820369204e-06,
+      "loss": 0.4958,
+      "step": 12042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94720036879694e-06,
+      "loss": 0.4908,
+      "step": 12043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945810918244038e-06,
+      "loss": 0.4589,
+      "step": 12044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.944421468737325e-06,
+      "loss": 0.477,
+      "step": 12045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943032020303631e-06,
+      "loss": 0.4661,
+      "step": 12046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941642572969775e-06,
+      "loss": 0.4753,
+      "step": 12047
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940253126762582e-06,
+      "loss": 0.4823,
+      "step": 12048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.938863681708883e-06,
+      "loss": 0.4638,
+      "step": 12049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937474237835496e-06,
+      "loss": 0.5121,
+      "step": 12050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.936084795169248e-06,
+      "loss": 0.4926,
+      "step": 12051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934695353736964e-06,
+      "loss": 0.494,
+      "step": 12052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933305913565477e-06,
+      "loss": 0.4853,
+      "step": 12053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.931916474681598e-06,
+      "loss": 0.4668,
+      "step": 12054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93052703711216e-06,
+      "loss": 0.4734,
+      "step": 12055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.929137600883986e-06,
+      "loss": 0.4707,
+      "step": 12056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.927748166023905e-06,
+      "loss": 0.4782,
+      "step": 12057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.926358732558735e-06,
+      "loss": 0.4454,
+      "step": 12058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.924969300515302e-06,
+      "loss": 0.4826,
+      "step": 12059
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.923579869920435e-06,
+      "loss": 0.4659,
+      "step": 12060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.922190440800963e-06,
+      "loss": 0.4783,
+      "step": 12061
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920801013183696e-06,
+      "loss": 0.4797,
+      "step": 12062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919411587095467e-06,
+      "loss": 0.4559,
+      "step": 12063
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918022162563101e-06,
+      "loss": 0.4933,
+      "step": 12064
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916632739613428e-06,
+      "loss": 0.4634,
+      "step": 12065
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.915243318273262e-06,
+      "loss": 0.4806,
+      "step": 12066
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91385389856943e-06,
+      "loss": 0.4593,
+      "step": 12067
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91246448052876e-06,
+      "loss": 0.4555,
+      "step": 12068
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.911075064178081e-06,
+      "loss": 0.4906,
+      "step": 12069
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909685649544205e-06,
+      "loss": 0.4989,
+      "step": 12070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.908296236653965e-06,
+      "loss": 0.4529,
+      "step": 12071
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906906825534189e-06,
+      "loss": 0.4748,
+      "step": 12072
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.90551741621169e-06,
+      "loss": 0.4622,
+      "step": 12073
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904128008713298e-06,
+      "loss": 0.5087,
+      "step": 12074
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.902738603065839e-06,
+      "loss": 0.4445,
+      "step": 12075
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.90134919929614e-06,
+      "loss": 0.4671,
+      "step": 12076
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899959797431018e-06,
+      "loss": 0.4795,
+      "step": 12077
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.898570397497295e-06,
+      "loss": 0.4395,
+      "step": 12078
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.897180999521805e-06,
+      "loss": 0.489,
+      "step": 12079
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.895791603531371e-06,
+      "loss": 0.4739,
+      "step": 12080
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.894402209552809e-06,
+      "loss": 0.476,
+      "step": 12081
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893012817612946e-06,
+      "loss": 0.4876,
+      "step": 12082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891623427738611e-06,
+      "loss": 0.4691,
+      "step": 12083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.890234039956626e-06,
+      "loss": 0.4756,
+      "step": 12084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88884465429381e-06,
+      "loss": 0.4779,
+      "step": 12085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88745527077699e-06,
+      "loss": 0.4797,
+      "step": 12086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886065889432992e-06,
+      "loss": 0.4699,
+      "step": 12087
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884676510288641e-06,
+      "loss": 0.4558,
+      "step": 12088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.883287133370752e-06,
+      "loss": 0.4857,
+      "step": 12089
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.881897758706155e-06,
+      "loss": 0.4583,
+      "step": 12090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88050838632168e-06,
+      "loss": 0.452,
+      "step": 12091
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.879119016244136e-06,
+      "loss": 0.4797,
+      "step": 12092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877729648500355e-06,
+      "loss": 0.4566,
+      "step": 12093
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876340283117161e-06,
+      "loss": 0.4702,
+      "step": 12094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87495092012138e-06,
+      "loss": 0.4514,
+      "step": 12095
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873561559539826e-06,
+      "loss": 0.4594,
+      "step": 12096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872172201399329e-06,
+      "loss": 0.5,
+      "step": 12097
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870782845726714e-06,
+      "loss": 0.4968,
+      "step": 12098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.869393492548802e-06,
+      "loss": 0.4789,
+      "step": 12099
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.868004141892412e-06,
+      "loss": 0.4635,
+      "step": 12100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866614793784373e-06,
+      "loss": 0.5008,
+      "step": 12101
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865225448251506e-06,
+      "loss": 0.4612,
+      "step": 12102
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863836105320636e-06,
+      "loss": 0.5147,
+      "step": 12103
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862446765018584e-06,
+      "loss": 0.464,
+      "step": 12104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.861057427372171e-06,
+      "loss": 0.4901,
+      "step": 12105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.859668092408224e-06,
+      "loss": 0.4691,
+      "step": 12106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.858278760153567e-06,
+      "loss": 0.5031,
+      "step": 12107
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856889430635017e-06,
+      "loss": 0.4762,
+      "step": 12108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855500103879398e-06,
+      "loss": 0.4614,
+      "step": 12109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.854110779913537e-06,
+      "loss": 0.4407,
+      "step": 12110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.852721458764256e-06,
+      "loss": 0.4653,
+      "step": 12111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.851332140458372e-06,
+      "loss": 0.5123,
+      "step": 12112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.849942825022715e-06,
+      "loss": 0.4734,
+      "step": 12113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848553512484104e-06,
+      "loss": 0.5103,
+      "step": 12114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847164202869359e-06,
+      "loss": 0.4419,
+      "step": 12115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.845774896205304e-06,
+      "loss": 0.4823,
+      "step": 12116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.844385592518763e-06,
+      "loss": 0.4742,
+      "step": 12117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84299629183656e-06,
+      "loss": 0.4922,
+      "step": 12118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841606994185513e-06,
+      "loss": 0.4441,
+      "step": 12119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.840217699592446e-06,
+      "loss": 0.4521,
+      "step": 12120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83882840808418e-06,
+      "loss": 0.4699,
+      "step": 12121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83743911968754e-06,
+      "loss": 0.4656,
+      "step": 12122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.836049834429344e-06,
+      "loss": 0.4883,
+      "step": 12123
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.834660552336415e-06,
+      "loss": 0.4982,
+      "step": 12124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833271273435578e-06,
+      "loss": 0.4816,
+      "step": 12125
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.831881997753654e-06,
+      "loss": 0.4694,
+      "step": 12126
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83049272531746e-06,
+      "loss": 0.4338,
+      "step": 12127
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.829103456153824e-06,
+      "loss": 0.4776,
+      "step": 12128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827714190289562e-06,
+      "loss": 0.4861,
+      "step": 12129
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826324927751501e-06,
+      "loss": 0.48,
+      "step": 12130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.824935668566458e-06,
+      "loss": 0.4614,
+      "step": 12131
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823546412761255e-06,
+      "loss": 0.478,
+      "step": 12132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.822157160362718e-06,
+      "loss": 0.4941,
+      "step": 12133
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.820767911397664e-06,
+      "loss": 0.4685,
+      "step": 12134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819378665892911e-06,
+      "loss": 0.4667,
+      "step": 12135
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.817989423875287e-06,
+      "loss": 0.4501,
+      "step": 12136
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.816600185371612e-06,
+      "loss": 0.4583,
+      "step": 12137
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.815210950408703e-06,
+      "loss": 0.4716,
+      "step": 12138
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813821719013384e-06,
+      "loss": 0.4776,
+      "step": 12139
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812432491212474e-06,
+      "loss": 0.4755,
+      "step": 12140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811043267032797e-06,
+      "loss": 0.4976,
+      "step": 12141
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80965404650117e-06,
+      "loss": 0.4429,
+      "step": 12142
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.808264829644417e-06,
+      "loss": 0.4677,
+      "step": 12143
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.806875616489355e-06,
+      "loss": 0.4649,
+      "step": 12144
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.805486407062809e-06,
+      "loss": 0.4691,
+      "step": 12145
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.804097201391595e-06,
+      "loss": 0.458,
+      "step": 12146
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802707999502538e-06,
+      "loss": 0.4469,
+      "step": 12147
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.801318801422452e-06,
+      "loss": 0.4405,
+      "step": 12148
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.799929607178165e-06,
+      "loss": 0.5136,
+      "step": 12149
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798540416796491e-06,
+      "loss": 0.4481,
+      "step": 12150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797151230304253e-06,
+      "loss": 0.4598,
+      "step": 12151
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.795762047728268e-06,
+      "loss": 0.4709,
+      "step": 12152
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.794372869095363e-06,
+      "loss": 0.4998,
+      "step": 12153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79298369443235e-06,
+      "loss": 0.4658,
+      "step": 12154
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.791594523766052e-06,
+      "loss": 0.4693,
+      "step": 12155
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79020535712329e-06,
+      "loss": 0.4591,
+      "step": 12156
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788816194530882e-06,
+      "loss": 0.4681,
+      "step": 12157
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.787427036015647e-06,
+      "loss": 0.4832,
+      "step": 12158
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786037881604404e-06,
+      "loss": 0.4879,
+      "step": 12159
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.784648731323977e-06,
+      "loss": 0.4896,
+      "step": 12160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.78325958520118e-06,
+      "loss": 0.457,
+      "step": 12161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.781870443262834e-06,
+      "loss": 0.4846,
+      "step": 12162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.780481305535759e-06,
+      "loss": 0.4666,
+      "step": 12163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.779092172046775e-06,
+      "loss": 0.4518,
+      "step": 12164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.777703042822699e-06,
+      "loss": 0.4545,
+      "step": 12165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.776313917890347e-06,
+      "loss": 0.477,
+      "step": 12166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.774924797276542e-06,
+      "loss": 0.4845,
+      "step": 12167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773535681008108e-06,
+      "loss": 0.1936,
+      "step": 12168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.772146569111854e-06,
+      "loss": 0.4786,
+      "step": 12169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770757461614601e-06,
+      "loss": 0.4983,
+      "step": 12170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769368358543169e-06,
+      "loss": 0.4491,
+      "step": 12171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76797925992438e-06,
+      "loss": 0.4676,
+      "step": 12172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.766590165785046e-06,
+      "loss": 0.4784,
+      "step": 12173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.765201076151988e-06,
+      "loss": 0.4713,
+      "step": 12174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763811991052021e-06,
+      "loss": 0.4489,
+      "step": 12175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762422910511971e-06,
+      "loss": 0.4848,
+      "step": 12176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76103383455865e-06,
+      "loss": 0.4677,
+      "step": 12177
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.759644763218872e-06,
+      "loss": 0.4518,
+      "step": 12178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758255696519468e-06,
+      "loss": 0.4659,
+      "step": 12179
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.756866634487243e-06,
+      "loss": 0.4714,
+      "step": 12180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755477577149018e-06,
+      "loss": 0.4862,
+      "step": 12181
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.754088524531608e-06,
+      "loss": 0.4609,
+      "step": 12182
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.752699476661843e-06,
+      "loss": 0.4901,
+      "step": 12183
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.751310433566526e-06,
+      "loss": 0.4518,
+      "step": 12184
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74992139527248e-06,
+      "loss": 0.1834,
+      "step": 12185
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74853236180652e-06,
+      "loss": 0.4745,
+      "step": 12186
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.747143333195471e-06,
+      "loss": 0.1624,
+      "step": 12187
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.745754309466138e-06,
+      "loss": 0.5072,
+      "step": 12188
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744365290645345e-06,
+      "loss": 0.4887,
+      "step": 12189
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.742976276759907e-06,
+      "loss": 0.4696,
+      "step": 12190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.741587267836645e-06,
+      "loss": 0.462,
+      "step": 12191
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74019826390237e-06,
+      "loss": 0.4797,
+      "step": 12192
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.738809264983897e-06,
+      "loss": 0.475,
+      "step": 12193
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737420271108047e-06,
+      "loss": 0.4671,
+      "step": 12194
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73603128230164e-06,
+      "loss": 0.4626,
+      "step": 12195
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734642298591482e-06,
+      "loss": 0.4735,
+      "step": 12196
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733253320004394e-06,
+      "loss": 0.4864,
+      "step": 12197
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.731864346567197e-06,
+      "loss": 0.4819,
+      "step": 12198
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7304753783067e-06,
+      "loss": 0.5043,
+      "step": 12199
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729086415249716e-06,
+      "loss": 0.4575,
+      "step": 12200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72769745742307e-06,
+      "loss": 0.4754,
+      "step": 12201
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.726308504853577e-06,
+      "loss": 0.4948,
+      "step": 12202
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.724919557568044e-06,
+      "loss": 0.4584,
+      "step": 12203
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.723530615593291e-06,
+      "loss": 0.469,
+      "step": 12204
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.722141678956134e-06,
+      "loss": 0.4721,
+      "step": 12205
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720752747683393e-06,
+      "loss": 0.4797,
+      "step": 12206
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.719363821801871e-06,
+      "loss": 0.5011,
+      "step": 12207
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717974901338392e-06,
+      "loss": 0.4682,
+      "step": 12208
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.716585986319769e-06,
+      "loss": 0.4842,
+      "step": 12209
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71519707677282e-06,
+      "loss": 0.4567,
+      "step": 12210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71380817272435e-06,
+      "loss": 0.4485,
+      "step": 12211
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.712419274201183e-06,
+      "loss": 0.4788,
+      "step": 12212
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71103038123013e-06,
+      "loss": 0.4648,
+      "step": 12213
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70964149383801e-06,
+      "loss": 0.4964,
+      "step": 12214
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.708252612051625e-06,
+      "loss": 0.5029,
+      "step": 12215
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7068637358978e-06,
+      "loss": 0.4789,
+      "step": 12216
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.705474865403346e-06,
+      "loss": 0.4861,
+      "step": 12217
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70408600059508e-06,
+      "loss": 0.4678,
+      "step": 12218
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.702697141499811e-06,
+      "loss": 0.4925,
+      "step": 12219
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.701308288144352e-06,
+      "loss": 0.4759,
+      "step": 12220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.699919440555526e-06,
+      "loss": 0.4899,
+      "step": 12221
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.698530598760132e-06,
+      "loss": 0.4956,
+      "step": 12222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697141762784993e-06,
+      "loss": 0.5046,
+      "step": 12223
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69575293265692e-06,
+      "loss": 0.179,
+      "step": 12224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69436410840273e-06,
+      "loss": 0.4558,
+      "step": 12225
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.692975290049228e-06,
+      "loss": 0.48,
+      "step": 12226
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691586477623233e-06,
+      "loss": 0.4655,
+      "step": 12227
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.690197671151554e-06,
+      "loss": 0.488,
+      "step": 12228
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.688808870661009e-06,
+      "loss": 0.4434,
+      "step": 12229
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.687420076178407e-06,
+      "loss": 0.5055,
+      "step": 12230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686031287730558e-06,
+      "loss": 0.4745,
+      "step": 12231
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68464250534428e-06,
+      "loss": 0.4996,
+      "step": 12232
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.683253729046383e-06,
+      "loss": 0.463,
+      "step": 12233
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.681864958863674e-06,
+      "loss": 0.4866,
+      "step": 12234
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680476194822973e-06,
+      "loss": 0.4702,
+      "step": 12235
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.679087436951087e-06,
+      "loss": 0.1684,
+      "step": 12236
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.677698685274832e-06,
+      "loss": 0.4727,
+      "step": 12237
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.676309939821013e-06,
+      "loss": 0.488,
+      "step": 12238
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.674921200616445e-06,
+      "loss": 0.4588,
+      "step": 12239
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.673532467687943e-06,
+      "loss": 0.4651,
+      "step": 12240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.672143741062313e-06,
+      "loss": 0.477,
+      "step": 12241
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670755020766369e-06,
+      "loss": 0.4742,
+      "step": 12242
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.669366306826919e-06,
+      "loss": 0.461,
+      "step": 12243
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667977599270779e-06,
+      "loss": 0.4924,
+      "step": 12244
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666588898124754e-06,
+      "loss": 0.4987,
+      "step": 12245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.665200203415657e-06,
+      "loss": 0.4705,
+      "step": 12246
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.6638115151703e-06,
+      "loss": 0.4905,
+      "step": 12247
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.662422833415493e-06,
+      "loss": 0.4745,
+      "step": 12248
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.661034158178045e-06,
+      "loss": 0.4787,
+      "step": 12249
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.659645489484766e-06,
+      "loss": 0.4472,
+      "step": 12250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.658256827362467e-06,
+      "loss": 0.4787,
+      "step": 12251
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65686817183796e-06,
+      "loss": 0.4764,
+      "step": 12252
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.655479522938049e-06,
+      "loss": 0.4758,
+      "step": 12253
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654090880689548e-06,
+      "loss": 0.4514,
+      "step": 12254
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652702245119267e-06,
+      "loss": 0.4831,
+      "step": 12255
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651313616254014e-06,
+      "loss": 0.493,
+      "step": 12256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.649924994120598e-06,
+      "loss": 0.4883,
+      "step": 12257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.648536378745828e-06,
+      "loss": 0.499,
+      "step": 12258
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647147770156511e-06,
+      "loss": 0.4682,
+      "step": 12259
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.645759168379463e-06,
+      "loss": 0.5094,
+      "step": 12260
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644370573441485e-06,
+      "loss": 0.4678,
+      "step": 12261
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.642981985369388e-06,
+      "loss": 0.4669,
+      "step": 12262
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.641593404189986e-06,
+      "loss": 0.474,
+      "step": 12263
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640204829930077e-06,
+      "loss": 0.4808,
+      "step": 12264
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.638816262616476e-06,
+      "loss": 0.4827,
+      "step": 12265
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.637427702275988e-06,
+      "loss": 0.486,
+      "step": 12266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636039148935428e-06,
+      "loss": 0.4951,
+      "step": 12267
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.634650602621593e-06,
+      "loss": 0.5162,
+      "step": 12268
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.633262063361298e-06,
+      "loss": 0.414,
+      "step": 12269
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631873531181349e-06,
+      "loss": 0.4726,
+      "step": 12270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.630485006108554e-06,
+      "loss": 0.4752,
+      "step": 12271
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.629096488169717e-06,
+      "loss": 0.4777,
+      "step": 12272
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.627707977391647e-06,
+      "loss": 0.4896,
+      "step": 12273
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.626319473801152e-06,
+      "loss": 0.4918,
+      "step": 12274
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62493097742504e-06,
+      "loss": 0.4664,
+      "step": 12275
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.623542488290115e-06,
+      "loss": 0.4534,
+      "step": 12276
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.622154006423185e-06,
+      "loss": 0.4743,
+      "step": 12277
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.620765531851055e-06,
+      "loss": 0.4736,
+      "step": 12278
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619377064600533e-06,
+      "loss": 0.4835,
+      "step": 12279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.617988604698423e-06,
+      "loss": 0.17,
+      "step": 12280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.616600152171531e-06,
+      "loss": 0.4732,
+      "step": 12281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.615211707046667e-06,
+      "loss": 0.4619,
+      "step": 12282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.613823269350633e-06,
+      "loss": 0.4545,
+      "step": 12283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.612434839110236e-06,
+      "loss": 0.4807,
+      "step": 12284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611046416352278e-06,
+      "loss": 0.4722,
+      "step": 12285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.609658001103571e-06,
+      "loss": 0.4676,
+      "step": 12286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.608269593390914e-06,
+      "loss": 0.5021,
+      "step": 12287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.606881193241115e-06,
+      "loss": 0.4737,
+      "step": 12288
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605492800680976e-06,
+      "loss": 0.4821,
+      "step": 12289
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604104415737309e-06,
+      "loss": 0.4649,
+      "step": 12290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60271603843691e-06,
+      "loss": 0.4652,
+      "step": 12291
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.601327668806584e-06,
+      "loss": 0.4988,
+      "step": 12292
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.599939306873141e-06,
+      "loss": 0.4642,
+      "step": 12293
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.598550952663383e-06,
+      "loss": 0.4624,
+      "step": 12294
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597162606204112e-06,
+      "loss": 0.4727,
+      "step": 12295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.595774267522132e-06,
+      "loss": 0.489,
+      "step": 12296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.594385936644245e-06,
+      "loss": 0.4464,
+      "step": 12297
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.592997613597265e-06,
+      "loss": 0.4771,
+      "step": 12298
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591609298407982e-06,
+      "loss": 0.1674,
+      "step": 12299
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.590220991103206e-06,
+      "loss": 0.4816,
+      "step": 12300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.588832691709734e-06,
+      "loss": 0.4686,
+      "step": 12301
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58744440025438e-06,
+      "loss": 0.4615,
+      "step": 12302
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.586056116763938e-06,
+      "loss": 0.4518,
+      "step": 12303
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58466784126521e-06,
+      "loss": 0.4852,
+      "step": 12304
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.583279573785009e-06,
+      "loss": 0.4805,
+      "step": 12305
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.581891314350123e-06,
+      "loss": 0.4466,
+      "step": 12306
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.580503062987361e-06,
+      "loss": 0.481,
+      "step": 12307
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.579114819723524e-06,
+      "loss": 0.4332,
+      "step": 12308
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57772658458542e-06,
+      "loss": 0.4749,
+      "step": 12309
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.576338357599842e-06,
+      "loss": 0.478,
+      "step": 12310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574950138793593e-06,
+      "loss": 0.5007,
+      "step": 12311
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.573561928193475e-06,
+      "loss": 0.4896,
+      "step": 12312
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572173725826296e-06,
+      "loss": 0.4735,
+      "step": 12313
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.570785531718846e-06,
+      "loss": 0.4762,
+      "step": 12314
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.56939734589793e-06,
+      "loss": 0.4722,
+      "step": 12315
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.568009168390353e-06,
+      "loss": 0.4769,
+      "step": 12316
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.566620999222915e-06,
+      "loss": 0.4918,
+      "step": 12317
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56523283842241e-06,
+      "loss": 0.4487,
+      "step": 12318
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56384468601564e-06,
+      "loss": 0.5031,
+      "step": 12319
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.562456542029409e-06,
+      "loss": 0.4679,
+      "step": 12320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.561068406490518e-06,
+      "loss": 0.5101,
+      "step": 12321
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.559680279425762e-06,
+      "loss": 0.4751,
+      "step": 12322
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.558292160861939e-06,
+      "loss": 0.4706,
+      "step": 12323
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.556904050825852e-06,
+      "loss": 0.4791,
+      "step": 12324
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555515949344307e-06,
+      "loss": 0.4613,
+      "step": 12325
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554127856444086e-06,
+      "loss": 0.4667,
+      "step": 12326
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.552739772152002e-06,
+      "loss": 0.484,
+      "step": 12327
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.551351696494854e-06,
+      "loss": 0.4883,
+      "step": 12328
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.549963629499432e-06,
+      "loss": 0.4689,
+      "step": 12329
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.548575571192535e-06,
+      "loss": 0.5051,
+      "step": 12330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54718752160097e-06,
+      "loss": 0.4915,
+      "step": 12331
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545799480751532e-06,
+      "loss": 0.4753,
+      "step": 12332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.544411448671012e-06,
+      "loss": 0.4648,
+      "step": 12333
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54302342538621e-06,
+      "loss": 0.4523,
+      "step": 12334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.541635410923929e-06,
+      "loss": 0.482,
+      "step": 12335
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54024740531097e-06,
+      "loss": 0.4636,
+      "step": 12336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538859408574115e-06,
+      "loss": 0.479,
+      "step": 12337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.537471420740173e-06,
+      "loss": 0.4898,
+      "step": 12338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536083441835936e-06,
+      "loss": 0.4734,
+      "step": 12339
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534695471888208e-06,
+      "loss": 0.4849,
+      "step": 12340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.533307510923773e-06,
+      "loss": 0.4798,
+      "step": 12341
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.531919558969435e-06,
+      "loss": 0.4282,
+      "step": 12342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530531616051992e-06,
+      "loss": 0.4917,
+      "step": 12343
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52914368219824e-06,
+      "loss": 0.4742,
+      "step": 12344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.527755757434968e-06,
+      "loss": 0.457,
+      "step": 12345
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.526367841788975e-06,
+      "loss": 0.471,
+      "step": 12346
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524979935287064e-06,
+      "loss": 0.4661,
+      "step": 12347
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523592037956017e-06,
+      "loss": 0.4547,
+      "step": 12348
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.522204149822638e-06,
+      "loss": 0.4744,
+      "step": 12349
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52081627091372e-06,
+      "loss": 0.465,
+      "step": 12350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.519428401256062e-06,
+      "loss": 0.4779,
+      "step": 12351
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.518040540876448e-06,
+      "loss": 0.4845,
+      "step": 12352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.516652689801679e-06,
+      "loss": 0.4839,
+      "step": 12353
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51526484805855e-06,
+      "loss": 0.4793,
+      "step": 12354
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513877015673858e-06,
+      "loss": 0.4766,
+      "step": 12355
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.512489192674389e-06,
+      "loss": 0.4577,
+      "step": 12356
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511101379086941e-06,
+      "loss": 0.4862,
+      "step": 12357
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.509713574938306e-06,
+      "loss": 0.4672,
+      "step": 12358
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.508325780255282e-06,
+      "loss": 0.4676,
+      "step": 12359
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506937995064655e-06,
+      "loss": 0.496,
+      "step": 12360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.505550219393223e-06,
+      "loss": 0.4931,
+      "step": 12361
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.504162453267776e-06,
+      "loss": 0.4473,
+      "step": 12362
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.502774696715111e-06,
+      "loss": 0.4996,
+      "step": 12363
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.501386949762016e-06,
+      "loss": 0.4483,
+      "step": 12364
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.499999212435283e-06,
+      "loss": 0.462,
+      "step": 12365
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.498611484761707e-06,
+      "loss": 0.4591,
+      "step": 12366
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49722376676808e-06,
+      "loss": 0.5128,
+      "step": 12367
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49583605848119e-06,
+      "loss": 0.4715,
+      "step": 12368
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49444835992783e-06,
+      "loss": 0.4842,
+      "step": 12369
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.493060671134795e-06,
+      "loss": 0.4587,
+      "step": 12370
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49167299212887e-06,
+      "loss": 0.4519,
+      "step": 12371
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.490285322936851e-06,
+      "loss": 0.4561,
+      "step": 12372
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488897663585525e-06,
+      "loss": 0.5009,
+      "step": 12373
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.487510014101687e-06,
+      "loss": 0.4796,
+      "step": 12374
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.486122374512122e-06,
+      "loss": 0.4809,
+      "step": 12375
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.484734744843622e-06,
+      "loss": 0.4574,
+      "step": 12376
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48334712512298e-06,
+      "loss": 0.1776,
+      "step": 12377
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481959515376985e-06,
+      "loss": 0.4534,
+      "step": 12378
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.480571915632422e-06,
+      "loss": 0.492,
+      "step": 12379
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479184325916083e-06,
+      "loss": 0.4692,
+      "step": 12380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.477796746254758e-06,
+      "loss": 0.4678,
+      "step": 12381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47640917667524e-06,
+      "loss": 0.4846,
+      "step": 12382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475021617204308e-06,
+      "loss": 0.4809,
+      "step": 12383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.473634067868759e-06,
+      "loss": 0.4927,
+      "step": 12384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.472246528695377e-06,
+      "loss": 0.4469,
+      "step": 12385
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.470858999710954e-06,
+      "loss": 0.4798,
+      "step": 12386
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.469471480942274e-06,
+      "loss": 0.1889,
+      "step": 12387
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.468083972416126e-06,
+      "loss": 0.4453,
+      "step": 12388
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.4666964741593e-06,
+      "loss": 0.4894,
+      "step": 12389
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465308986198581e-06,
+      "loss": 0.5008,
+      "step": 12390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.463921508560757e-06,
+      "loss": 0.4967,
+      "step": 12391
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.462534041272615e-06,
+      "loss": 0.4604,
+      "step": 12392
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.461146584360941e-06,
+      "loss": 0.4801,
+      "step": 12393
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.459759137852523e-06,
+      "loss": 0.4848,
+      "step": 12394
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.458371701774145e-06,
+      "loss": 0.4415,
+      "step": 12395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.456984276152598e-06,
+      "loss": 0.461,
+      "step": 12396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.455596861014665e-06,
+      "loss": 0.4679,
+      "step": 12397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45420945638713e-06,
+      "loss": 0.4699,
+      "step": 12398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45282206229678e-06,
+      "loss": 0.4839,
+      "step": 12399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.451434678770402e-06,
+      "loss": 0.4556,
+      "step": 12400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.450047305834782e-06,
+      "loss": 0.4505,
+      "step": 12401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.448659943516701e-06,
+      "loss": 0.5247,
+      "step": 12402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.447272591842946e-06,
+      "loss": 0.4786,
+      "step": 12403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.445885250840301e-06,
+      "loss": 0.4835,
+      "step": 12404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444497920535554e-06,
+      "loss": 0.4682,
+      "step": 12405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.443110600955483e-06,
+      "loss": 0.4807,
+      "step": 12406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.441723292126877e-06,
+      "loss": 0.4829,
+      "step": 12407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.440335994076516e-06,
+      "loss": 0.4851,
+      "step": 12408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.438948706831188e-06,
+      "loss": 0.4806,
+      "step": 12409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437561430417672e-06,
+      "loss": 0.4965,
+      "step": 12410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436174164862754e-06,
+      "loss": 0.4725,
+      "step": 12411
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.434786910193216e-06,
+      "loss": 0.4768,
+      "step": 12412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43339966643584e-06,
+      "loss": 0.4571,
+      "step": 12413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43201243361741e-06,
+      "loss": 0.4873,
+      "step": 12414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.430625211764706e-06,
+      "loss": 0.4408,
+      "step": 12415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.429238000904514e-06,
+      "loss": 0.4746,
+      "step": 12416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427850801063612e-06,
+      "loss": 0.4767,
+      "step": 12417
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.426463612268781e-06,
+      "loss": 0.4711,
+      "step": 12418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425076434546808e-06,
+      "loss": 0.4738,
+      "step": 12419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423689267924469e-06,
+      "loss": 0.4613,
+      "step": 12420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.422302112428547e-06,
+      "loss": 0.4767,
+      "step": 12421
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.420914968085822e-06,
+      "loss": 0.4815,
+      "step": 12422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.419527834923074e-06,
+      "loss": 0.4523,
+      "step": 12423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418140712967091e-06,
+      "loss": 0.4727,
+      "step": 12424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.416753602244643e-06,
+      "loss": 0.4916,
+      "step": 12425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415366502782513e-06,
+      "loss": 0.5139,
+      "step": 12426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.413979414607479e-06,
+      "loss": 0.491,
+      "step": 12427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412592337746329e-06,
+      "loss": 0.4607,
+      "step": 12428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.411205272225833e-06,
+      "loss": 0.4582,
+      "step": 12429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.409818218072774e-06,
+      "loss": 0.4781,
+      "step": 12430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40843117531393e-06,
+      "loss": 0.4579,
+      "step": 12431
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40704414397608e-06,
+      "loss": 0.4412,
+      "step": 12432
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.405657124086001e-06,
+      "loss": 0.4891,
+      "step": 12433
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404270115670472e-06,
+      "loss": 0.465,
+      "step": 12434
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.402883118756277e-06,
+      "loss": 0.4481,
+      "step": 12435
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401496133370183e-06,
+      "loss": 0.4888,
+      "step": 12436
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.400109159538975e-06,
+      "loss": 0.4777,
+      "step": 12437
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.398722197289424e-06,
+      "loss": 0.4674,
+      "step": 12438
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.397335246648318e-06,
+      "loss": 0.4801,
+      "step": 12439
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395948307642423e-06,
+      "loss": 0.5133,
+      "step": 12440
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.394561380298516e-06,
+      "loss": 0.4753,
+      "step": 12441
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.393174464643381e-06,
+      "loss": 0.4896,
+      "step": 12442
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.391787560703795e-06,
+      "loss": 0.494,
+      "step": 12443
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.390400668506523e-06,
+      "loss": 0.4896,
+      "step": 12444
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389013788078344e-06,
+      "loss": 0.4765,
+      "step": 12445
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38762691944604e-06,
+      "loss": 0.4388,
+      "step": 12446
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.386240062636388e-06,
+      "loss": 0.4541,
+      "step": 12447
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.384853217676153e-06,
+      "loss": 0.4973,
+      "step": 12448
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.383466384592113e-06,
+      "loss": 0.5145,
+      "step": 12449
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.382079563411045e-06,
+      "loss": 0.5051,
+      "step": 12450
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.380692754159726e-06,
+      "loss": 0.4723,
+      "step": 12451
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.379305956864922e-06,
+      "loss": 0.4734,
+      "step": 12452
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.377919171553412e-06,
+      "loss": 0.4819,
+      "step": 12453
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.376532398251974e-06,
+      "loss": 0.4815,
+      "step": 12454
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375145636987375e-06,
+      "loss": 0.4325,
+      "step": 12455
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.373758887786384e-06,
+      "loss": 0.4999,
+      "step": 12456
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372372150675784e-06,
+      "loss": 0.1812,
+      "step": 12457
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370985425682347e-06,
+      "loss": 0.4817,
+      "step": 12458
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.369598712832839e-06,
+      "loss": 0.4654,
+      "step": 12459
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36821201215403e-06,
+      "loss": 0.4554,
+      "step": 12460
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366825323672702e-06,
+      "loss": 0.4384,
+      "step": 12461
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.365438647415627e-06,
+      "loss": 0.4698,
+      "step": 12462
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364051983409563e-06,
+      "loss": 0.4929,
+      "step": 12463
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362665331681294e-06,
+      "loss": 0.4612,
+      "step": 12464
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361278692257588e-06,
+      "loss": 0.4912,
+      "step": 12465
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359892065165217e-06,
+      "loss": 0.4392,
+      "step": 12466
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.358505450430944e-06,
+      "loss": 0.4832,
+      "step": 12467
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357118848081547e-06,
+      "loss": 0.4822,
+      "step": 12468
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.355732258143794e-06,
+      "loss": 0.4727,
+      "step": 12469
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.354345680644459e-06,
+      "loss": 0.499,
+      "step": 12470
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.352959115610302e-06,
+      "loss": 0.4831,
+      "step": 12471
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3515725630681e-06,
+      "loss": 0.4644,
+      "step": 12472
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.35018602304462e-06,
+      "loss": 0.4454,
+      "step": 12473
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.348799495566633e-06,
+      "loss": 0.497,
+      "step": 12474
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347412980660903e-06,
+      "loss": 0.4779,
+      "step": 12475
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.346026478354203e-06,
+      "loss": 0.4685,
+      "step": 12476
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.344639988673304e-06,
+      "loss": 0.4873,
+      "step": 12477
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.343253511644962e-06,
+      "loss": 0.4865,
+      "step": 12478
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.341867047295954e-06,
+      "loss": 0.5062,
+      "step": 12479
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340480595653047e-06,
+      "loss": 0.4756,
+      "step": 12480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.339094156743007e-06,
+      "loss": 0.1858,
+      "step": 12481
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3377077305926e-06,
+      "loss": 0.4702,
+      "step": 12482
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.336321317228594e-06,
+      "loss": 0.4889,
+      "step": 12483
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.334934916677754e-06,
+      "loss": 0.4711,
+      "step": 12484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.333548528966849e-06,
+      "loss": 0.5096,
+      "step": 12485
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.332162154122642e-06,
+      "loss": 0.4637,
+      "step": 12486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.3307757921719e-06,
+      "loss": 0.4596,
+      "step": 12487
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32938944314139e-06,
+      "loss": 0.4678,
+      "step": 12488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.328003107057874e-06,
+      "loss": 0.4843,
+      "step": 12489
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32661678394812e-06,
+      "loss": 0.4662,
+      "step": 12490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325230473838891e-06,
+      "loss": 0.4451,
+      "step": 12491
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32384417675695e-06,
+      "loss": 0.4773,
+      "step": 12492
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322457892729069e-06,
+      "loss": 0.4824,
+      "step": 12493
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321071621782001e-06,
+      "loss": 0.4927,
+      "step": 12494
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.319685363942516e-06,
+      "loss": 0.4537,
+      "step": 12495
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318299119237378e-06,
+      "loss": 0.4577,
+      "step": 12496
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316912887693348e-06,
+      "loss": 0.4746,
+      "step": 12497
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.315526669337189e-06,
+      "loss": 0.4696,
+      "step": 12498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314140464195663e-06,
+      "loss": 0.4582,
+      "step": 12499
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312754272295538e-06,
+      "loss": 0.4659,
+      "step": 12500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.311368093663568e-06,
+      "loss": 0.5015,
+      "step": 12501
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30998192832652e-06,
+      "loss": 0.487,
+      "step": 12502
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.308595776311155e-06,
+      "loss": 0.4964,
+      "step": 12503
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307209637644235e-06,
+      "loss": 0.4405,
+      "step": 12504
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30582351235252e-06,
+      "loss": 0.4774,
+      "step": 12505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.304437400462771e-06,
+      "loss": 0.4526,
+      "step": 12506
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.303051302001747e-06,
+      "loss": 0.4842,
+      "step": 12507
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301665216996214e-06,
+      "loss": 0.4839,
+      "step": 12508
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300279145472925e-06,
+      "loss": 0.4811,
+      "step": 12509
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.298893087458645e-06,
+      "loss": 0.4744,
+      "step": 12510
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297507042980132e-06,
+      "loss": 0.4765,
+      "step": 12511
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.296121012064146e-06,
+      "loss": 0.4598,
+      "step": 12512
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.294734994737444e-06,
+      "loss": 0.4499,
+      "step": 12513
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293348991026787e-06,
+      "loss": 0.4653,
+      "step": 12514
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.291963000958932e-06,
+      "loss": 0.1823,
+      "step": 12515
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.290577024560641e-06,
+      "loss": 0.4686,
+      "step": 12516
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.289191061858665e-06,
+      "loss": 0.4456,
+      "step": 12517
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28780511287977e-06,
+      "loss": 0.4699,
+      "step": 12518
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286419177650707e-06,
+      "loss": 0.4582,
+      "step": 12519
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285033256198237e-06,
+      "loss": 0.4661,
+      "step": 12520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.283647348549113e-06,
+      "loss": 0.4288,
+      "step": 12521
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282261454730098e-06,
+      "loss": 0.4945,
+      "step": 12522
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280875574767945e-06,
+      "loss": 0.4619,
+      "step": 12523
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279489708689407e-06,
+      "loss": 0.1674,
+      "step": 12524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.278103856521246e-06,
+      "loss": 0.4314,
+      "step": 12525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276718018290213e-06,
+      "loss": 0.4804,
+      "step": 12526
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.275332194023066e-06,
+      "loss": 0.4957,
+      "step": 12527
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27394638374656e-06,
+      "loss": 0.4624,
+      "step": 12528
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.272560587487447e-06,
+      "loss": 0.4919,
+      "step": 12529
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271174805272482e-06,
+      "loss": 0.4868,
+      "step": 12530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269789037128426e-06,
+      "loss": 0.4638,
+      "step": 12531
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268403283082025e-06,
+      "loss": 0.4739,
+      "step": 12532
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.267017543160034e-06,
+      "loss": 0.4814,
+      "step": 12533
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.265631817389211e-06,
+      "loss": 0.4676,
+      "step": 12534
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264246105796307e-06,
+      "loss": 0.484,
+      "step": 12535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262860408408072e-06,
+      "loss": 0.489,
+      "step": 12536
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.261474725251261e-06,
+      "loss": 0.4593,
+      "step": 12537
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26008905635263e-06,
+      "loss": 0.4645,
+      "step": 12538
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258703401738924e-06,
+      "loss": 0.4851,
+      "step": 12539
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257317761436899e-06,
+      "loss": 0.4634,
+      "step": 12540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255932135473304e-06,
+      "loss": 0.4382,
+      "step": 12541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.254546523874897e-06,
+      "loss": 0.4695,
+      "step": 12542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25316092666842e-06,
+      "loss": 0.4746,
+      "step": 12543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25177534388063e-06,
+      "loss": 0.4623,
+      "step": 12544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.250389775538273e-06,
+      "loss": 0.4797,
+      "step": 12545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.249004221668106e-06,
+      "loss": 0.4788,
+      "step": 12546
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24761868229687e-06,
+      "loss": 0.1727,
+      "step": 12547
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.246233157451319e-06,
+      "loss": 0.4729,
+      "step": 12548
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.244847647158203e-06,
+      "loss": 0.4852,
+      "step": 12549
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.243462151444275e-06,
+      "loss": 0.4737,
+      "step": 12550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.242076670336276e-06,
+      "loss": 0.4546,
+      "step": 12551
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.240691203860955e-06,
+      "loss": 0.4578,
+      "step": 12552
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.239305752045062e-06,
+      "loss": 0.4773,
+      "step": 12553
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.237920314915354e-06,
+      "loss": 0.4827,
+      "step": 12554
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.236534892498564e-06,
+      "loss": 0.4752,
+      "step": 12555
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235149484821447e-06,
+      "loss": 0.4586,
+      "step": 12556
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233764091910747e-06,
+      "loss": 0.4621,
+      "step": 12557
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23237871379322e-06,
+      "loss": 0.4981,
+      "step": 12558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.230993350495597e-06,
+      "loss": 0.4466,
+      "step": 12559
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.229608002044633e-06,
+      "loss": 0.4545,
+      "step": 12560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22822266846708e-06,
+      "loss": 0.484,
+      "step": 12561
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226837349789671e-06,
+      "loss": 0.4725,
+      "step": 12562
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225452046039158e-06,
+      "loss": 0.4716,
+      "step": 12563
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.224066757242284e-06,
+      "loss": 0.4737,
+      "step": 12564
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.222681483425801e-06,
+      "loss": 0.4761,
+      "step": 12565
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221296224616443e-06,
+      "loss": 0.4594,
+      "step": 12566
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21991098084096e-06,
+      "loss": 0.4777,
+      "step": 12567
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21852575212609e-06,
+      "loss": 0.4688,
+      "step": 12568
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21714053849859e-06,
+      "loss": 0.4655,
+      "step": 12569
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21575533998519e-06,
+      "loss": 0.4968,
+      "step": 12570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214370156612635e-06,
+      "loss": 0.4721,
+      "step": 12571
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.212984988407673e-06,
+      "loss": 0.4493,
+      "step": 12572
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.211599835397048e-06,
+      "loss": 0.4734,
+      "step": 12573
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210214697607494e-06,
+      "loss": 0.5004,
+      "step": 12574
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.208829575065754e-06,
+      "loss": 0.4899,
+      "step": 12575
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207444467798575e-06,
+      "loss": 0.4788,
+      "step": 12576
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.2060593758327e-06,
+      "loss": 0.4835,
+      "step": 12577
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204674299194862e-06,
+      "loss": 0.4793,
+      "step": 12578
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203289237911802e-06,
+      "loss": 0.4828,
+      "step": 12579
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.201904192010272e-06,
+      "loss": 0.4719,
+      "step": 12580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200519161516997e-06,
+      "loss": 0.4874,
+      "step": 12581
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.199134146458723e-06,
+      "loss": 0.4554,
+      "step": 12582
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.197749146862193e-06,
+      "loss": 0.4736,
+      "step": 12583
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196364162754147e-06,
+      "loss": 0.4877,
+      "step": 12584
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194979194161315e-06,
+      "loss": 0.4543,
+      "step": 12585
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.193594241110439e-06,
+      "loss": 0.4624,
+      "step": 12586
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192209303628262e-06,
+      "loss": 0.4613,
+      "step": 12587
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190824381741523e-06,
+      "loss": 0.4695,
+      "step": 12588
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18943947547695e-06,
+      "loss": 0.4899,
+      "step": 12589
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188054584861285e-06,
+      "loss": 0.4866,
+      "step": 12590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18666970992127e-06,
+      "loss": 0.5015,
+      "step": 12591
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18528485068364e-06,
+      "loss": 0.4579,
+      "step": 12592
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183900007175123e-06,
+      "loss": 0.4409,
+      "step": 12593
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182515179422466e-06,
+      "loss": 0.4613,
+      "step": 12594
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.1811303674524e-06,
+      "loss": 0.4777,
+      "step": 12595
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.179745571291664e-06,
+      "loss": 0.4741,
+      "step": 12596
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178360790966985e-06,
+      "loss": 0.4675,
+      "step": 12597
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176976026505106e-06,
+      "loss": 0.45,
+      "step": 12598
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175591277932756e-06,
+      "loss": 0.4757,
+      "step": 12599
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.458,
+      "step": 12600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.172821828563598e-06,
+      "loss": 0.4649,
+      "step": 12601
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17143712782025e-06,
+      "loss": 0.4618,
+      "step": 12602
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170052443073374e-06,
+      "loss": 0.4758,
+      "step": 12603
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.168667774349693e-06,
+      "loss": 0.4507,
+      "step": 12604
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.167283121675949e-06,
+      "loss": 0.1783,
+      "step": 12605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16589848507887e-06,
+      "loss": 0.4786,
+      "step": 12606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.164513864585192e-06,
+      "loss": 0.4945,
+      "step": 12607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.163129260221638e-06,
+      "loss": 0.467,
+      "step": 12608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.161744672014949e-06,
+      "loss": 0.4641,
+      "step": 12609
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.160360099991852e-06,
+      "loss": 0.488,
+      "step": 12610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15897554417908e-06,
+      "loss": 0.445,
+      "step": 12611
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15759100460336e-06,
+      "loss": 0.4618,
+      "step": 12612
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.156206481291427e-06,
+      "loss": 0.4798,
+      "step": 12613
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154821974270006e-06,
+      "loss": 0.4553,
+      "step": 12614
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.153437483565835e-06,
+      "loss": 0.4607,
+      "step": 12615
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.152053009205633e-06,
+      "loss": 0.472,
+      "step": 12616
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150668551216134e-06,
+      "loss": 0.4738,
+      "step": 12617
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.149284109624066e-06,
+      "loss": 0.4988,
+      "step": 12618
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.147899684456162e-06,
+      "loss": 0.4586,
+      "step": 12619
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.146515275739142e-06,
+      "loss": 0.465,
+      "step": 12620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.14513088349974e-06,
+      "loss": 0.4456,
+      "step": 12621
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.143746507764679e-06,
+      "loss": 0.4618,
+      "step": 12622
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.142362148560692e-06,
+      "loss": 0.5382,
+      "step": 12623
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.1409778059145e-06,
+      "loss": 0.4493,
+      "step": 12624
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13959347985283e-06,
+      "loss": 0.4899,
+      "step": 12625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.138209170402414e-06,
+      "loss": 0.4885,
+      "step": 12626
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13682487758997e-06,
+      "loss": 0.4524,
+      "step": 12627
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135440601442229e-06,
+      "loss": 0.4457,
+      "step": 12628
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.134056341985913e-06,
+      "loss": 0.4803,
+      "step": 12629
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.132672099247753e-06,
+      "loss": 0.4433,
+      "step": 12630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.131287873254464e-06,
+      "loss": 0.4845,
+      "step": 12631
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129903664032777e-06,
+      "loss": 0.4837,
+      "step": 12632
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.128519471609413e-06,
+      "loss": 0.4704,
+      "step": 12633
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.127135296011102e-06,
+      "loss": 0.4812,
+      "step": 12634
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125751137264556e-06,
+      "loss": 0.45,
+      "step": 12635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.124366995396508e-06,
+      "loss": 0.4644,
+      "step": 12636
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122982870433674e-06,
+      "loss": 0.461,
+      "step": 12637
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12159876240278e-06,
+      "loss": 0.4937,
+      "step": 12638
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120214671330548e-06,
+      "loss": 0.4815,
+      "step": 12639
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118830597243697e-06,
+      "loss": 0.474,
+      "step": 12640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.117446540168951e-06,
+      "loss": 0.4814,
+      "step": 12641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116062500133031e-06,
+      "loss": 0.4843,
+      "step": 12642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.114678477162656e-06,
+      "loss": 0.4931,
+      "step": 12643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113294471284546e-06,
+      "loss": 0.4764,
+      "step": 12644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111910482525423e-06,
+      "loss": 0.4704,
+      "step": 12645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110526510912006e-06,
+      "loss": 0.1769,
+      "step": 12646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109142556471013e-06,
+      "loss": 0.4598,
+      "step": 12647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.107758619229163e-06,
+      "loss": 0.4676,
+      "step": 12648
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106374699213179e-06,
+      "loss": 0.5004,
+      "step": 12649
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104990796449774e-06,
+      "loss": 0.4809,
+      "step": 12650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103606910965666e-06,
+      "loss": 0.483,
+      "step": 12651
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102223042787577e-06,
+      "loss": 0.4721,
+      "step": 12652
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.100839191942224e-06,
+      "loss": 0.4801,
+      "step": 12653
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.099455358456319e-06,
+      "loss": 0.4831,
+      "step": 12654
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.098071542356581e-06,
+      "loss": 0.464,
+      "step": 12655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096687743669728e-06,
+      "loss": 0.1709,
+      "step": 12656
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.09530396242248e-06,
+      "loss": 0.4853,
+      "step": 12657
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.093920198641541e-06,
+      "loss": 0.4544,
+      "step": 12658
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092536452353636e-06,
+      "loss": 0.4898,
+      "step": 12659
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091152723585477e-06,
+      "loss": 0.4786,
+      "step": 12660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08976901236378e-06,
+      "loss": 0.4693,
+      "step": 12661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.088385318715257e-06,
+      "loss": 0.47,
+      "step": 12662
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087001642666622e-06,
+      "loss": 0.4991,
+      "step": 12663
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08561798424459e-06,
+      "loss": 0.4668,
+      "step": 12664
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.084234343475877e-06,
+      "loss": 0.4829,
+      "step": 12665
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.082850720387189e-06,
+      "loss": 0.4672,
+      "step": 12666
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.081467115005245e-06,
+      "loss": 0.4633,
+      "step": 12667
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.080083527356755e-06,
+      "loss": 0.4603,
+      "step": 12668
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07869995746843e-06,
+      "loss": 0.4729,
+      "step": 12669
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 0.5188,
+      "step": 12670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.075932871079123e-06,
+      "loss": 0.4854,
+      "step": 12671
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.074549354631565e-06,
+      "loss": 0.4575,
+      "step": 12672
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.073165856051015e-06,
+      "loss": 0.48,
+      "step": 12673
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.071782375364185e-06,
+      "loss": 0.4683,
+      "step": 12674
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.070398912597785e-06,
+      "loss": 0.5023,
+      "step": 12675
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069015467778528e-06,
+      "loss": 0.4731,
+      "step": 12676
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067632040933116e-06,
+      "loss": 0.4996,
+      "step": 12677
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.066248632088263e-06,
+      "loss": 0.4491,
+      "step": 12678
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.064865241270673e-06,
+      "loss": 0.4853,
+      "step": 12679
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.063481868507066e-06,
+      "loss": 0.4778,
+      "step": 12680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062098513824135e-06,
+      "loss": 0.4493,
+      "step": 12681
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.060715177248594e-06,
+      "loss": 0.4532,
+      "step": 12682
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.059331858807147e-06,
+      "loss": 0.4598,
+      "step": 12683
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05794855852651e-06,
+      "loss": 0.4586,
+      "step": 12684
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056565276433378e-06,
+      "loss": 0.497,
+      "step": 12685
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055182012554462e-06,
+      "loss": 0.4662,
+      "step": 12686
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05379876691647e-06,
+      "loss": 0.4756,
+      "step": 12687
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.052415539546104e-06,
+      "loss": 0.4949,
+      "step": 12688
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.051032330470068e-06,
+      "loss": 0.4659,
+      "step": 12689
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049649139715067e-06,
+      "loss": 0.4617,
+      "step": 12690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.048265967307813e-06,
+      "loss": 0.5115,
+      "step": 12691
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.046882813274997e-06,
+      "loss": 0.489,
+      "step": 12692
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.045499677643331e-06,
+      "loss": 0.4615,
+      "step": 12693
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044116560439515e-06,
+      "loss": 0.4696,
+      "step": 12694
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.042733461690259e-06,
+      "loss": 0.465,
+      "step": 12695
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041350381422254e-06,
+      "loss": 0.4647,
+      "step": 12696
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039967319662206e-06,
+      "loss": 0.5217,
+      "step": 12697
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.038584276436822e-06,
+      "loss": 0.444,
+      "step": 12698
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.037201251772803e-06,
+      "loss": 0.4584,
+      "step": 12699
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035818245696843e-06,
+      "loss": 0.4673,
+      "step": 12700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.034435258235645e-06,
+      "loss": 0.4937,
+      "step": 12701
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.033052289415914e-06,
+      "loss": 0.4928,
+      "step": 12702
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.03166933926435e-06,
+      "loss": 0.4387,
+      "step": 12703
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.030286407807647e-06,
+      "loss": 0.4974,
+      "step": 12704
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028903495072503e-06,
+      "loss": 0.473,
+      "step": 12705
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.027520601085624e-06,
+      "loss": 0.4775,
+      "step": 12706
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02613772587371e-06,
+      "loss": 0.4568,
+      "step": 12707
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.024754869463448e-06,
+      "loss": 0.4719,
+      "step": 12708
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.023372031881543e-06,
+      "loss": 0.4686,
+      "step": 12709
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.021989213154699e-06,
+      "loss": 0.4827,
+      "step": 12710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.020606413309599e-06,
+      "loss": 0.4577,
+      "step": 12711
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.019223632372946e-06,
+      "loss": 0.4624,
+      "step": 12712
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.017840870371439e-06,
+      "loss": 0.4764,
+      "step": 12713
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.016458127331775e-06,
+      "loss": 0.4528,
+      "step": 12714
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.015075403280644e-06,
+      "loss": 0.4932,
+      "step": 12715
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.013692698244742e-06,
+      "loss": 0.5072,
+      "step": 12716
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.012310012250767e-06,
+      "loss": 0.4841,
+      "step": 12717
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.010927345325416e-06,
+      "loss": 0.4896,
+      "step": 12718
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.009544697495373e-06,
+      "loss": 0.4418,
+      "step": 12719
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008162068787341e-06,
+      "loss": 0.4476,
+      "step": 12720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.006779459228012e-06,
+      "loss": 0.4571,
+      "step": 12721
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.00539686884408e-06,
+      "loss": 0.475,
+      "step": 12722
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.00401429766223e-06,
+      "loss": 0.45,
+      "step": 12723
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.00263174570916e-06,
+      "loss": 0.4687,
+      "step": 12724
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001249213011565e-06,
+      "loss": 0.4942,
+      "step": 12725
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.999866699596133e-06,
+      "loss": 0.4621,
+      "step": 12726
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.998484205489554e-06,
+      "loss": 0.4736,
+      "step": 12727
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99710173071852e-06,
+      "loss": 0.4657,
+      "step": 12728
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995719275309726e-06,
+      "loss": 0.5016,
+      "step": 12729
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994336839289853e-06,
+      "loss": 0.4659,
+      "step": 12730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.992954422685595e-06,
+      "loss": 0.4563,
+      "step": 12731
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.991572025523643e-06,
+      "loss": 0.4939,
+      "step": 12732
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990189647830689e-06,
+      "loss": 0.4618,
+      "step": 12733
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98880728963341e-06,
+      "loss": 0.4697,
+      "step": 12734
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987424950958505e-06,
+      "loss": 0.4789,
+      "step": 12735
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.986042631832656e-06,
+      "loss": 0.4985,
+      "step": 12736
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984660332282558e-06,
+      "loss": 0.487,
+      "step": 12737
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983278052334888e-06,
+      "loss": 0.167,
+      "step": 12738
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.981895792016339e-06,
+      "loss": 0.4844,
+      "step": 12739
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.980513551353595e-06,
+      "loss": 0.4758,
+      "step": 12740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.979131330373346e-06,
+      "loss": 0.4798,
+      "step": 12741
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97774912910227e-06,
+      "loss": 0.4799,
+      "step": 12742
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976366947567057e-06,
+      "loss": 0.4707,
+      "step": 12743
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974984785794392e-06,
+      "loss": 0.4694,
+      "step": 12744
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.973602643810962e-06,
+      "loss": 0.4707,
+      "step": 12745
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.972220521643445e-06,
+      "loss": 0.4513,
+      "step": 12746
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.970838419318525e-06,
+      "loss": 0.469,
+      "step": 12747
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96945633686289e-06,
+      "loss": 0.4938,
+      "step": 12748
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.968074274303221e-06,
+      "loss": 0.4856,
+      "step": 12749
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9666922316662e-06,
+      "loss": 0.4613,
+      "step": 12750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965310208978506e-06,
+      "loss": 0.4651,
+      "step": 12751
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.963928206266826e-06,
+      "loss": 0.4842,
+      "step": 12752
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962546223557838e-06,
+      "loss": 0.4512,
+      "step": 12753
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.961164260878224e-06,
+      "loss": 0.4654,
+      "step": 12754
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.959782318254665e-06,
+      "loss": 0.4755,
+      "step": 12755
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.958400395713842e-06,
+      "loss": 0.4825,
+      "step": 12756
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.957018493282432e-06,
+      "loss": 0.455,
+      "step": 12757
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.955636610987116e-06,
+      "loss": 0.4462,
+      "step": 12758
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954254748854572e-06,
+      "loss": 0.4564,
+      "step": 12759
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.952872906911484e-06,
+      "loss": 0.459,
+      "step": 12760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951491085184522e-06,
+      "loss": 0.4647,
+      "step": 12761
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.950109283700367e-06,
+      "loss": 0.426,
+      "step": 12762
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.948727502485698e-06,
+      "loss": 0.4646,
+      "step": 12763
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.947345741567192e-06,
+      "loss": 0.4738,
+      "step": 12764
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945964000971525e-06,
+      "loss": 0.4585,
+      "step": 12765
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94458228072537e-06,
+      "loss": 0.476,
+      "step": 12766
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.943200580855408e-06,
+      "loss": 0.4721,
+      "step": 12767
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941818901388314e-06,
+      "loss": 0.4771,
+      "step": 12768
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.940437242350759e-06,
+      "loss": 0.4651,
+      "step": 12769
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93905560376942e-06,
+      "loss": 0.4807,
+      "step": 12770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.937673985670972e-06,
+      "loss": 0.447,
+      "step": 12771
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.936292388082091e-06,
+      "loss": 0.4861,
+      "step": 12772
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934910811029446e-06,
+      "loss": 0.4609,
+      "step": 12773
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93352925453971e-06,
+      "loss": 0.4689,
+      "step": 12774
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.932147718639562e-06,
+      "loss": 0.4458,
+      "step": 12775
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.930766203355668e-06,
+      "loss": 0.458,
+      "step": 12776
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9293847087147e-06,
+      "loss": 0.4719,
+      "step": 12777
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.928003234743333e-06,
+      "loss": 0.4616,
+      "step": 12778
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92662178146824e-06,
+      "loss": 0.4493,
+      "step": 12779
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.925240348916085e-06,
+      "loss": 0.453,
+      "step": 12780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92385893711354e-06,
+      "loss": 0.485,
+      "step": 12781
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.922477546087278e-06,
+      "loss": 0.4561,
+      "step": 12782
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.92109617586397e-06,
+      "loss": 0.4613,
+      "step": 12783
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.91971482647028e-06,
+      "loss": 0.4946,
+      "step": 12784
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918333497932878e-06,
+      "loss": 0.4819,
+      "step": 12785
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.916952190278433e-06,
+      "loss": 0.4666,
+      "step": 12786
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.915570903533615e-06,
+      "loss": 0.4839,
+      "step": 12787
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.91418963772509e-06,
+      "loss": 0.4739,
+      "step": 12788
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912808392879523e-06,
+      "loss": 0.496,
+      "step": 12789
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.911427169023583e-06,
+      "loss": 0.479,
+      "step": 12790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910045966183938e-06,
+      "loss": 0.4589,
+      "step": 12791
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90866478438725e-06,
+      "loss": 0.4452,
+      "step": 12792
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.907283623660185e-06,
+      "loss": 0.4519,
+      "step": 12793
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905902484029411e-06,
+      "loss": 0.4763,
+      "step": 12794
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.904521365521591e-06,
+      "loss": 0.4891,
+      "step": 12795
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.903140268163388e-06,
+      "loss": 0.4722,
+      "step": 12796
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.901759191981467e-06,
+      "loss": 0.4514,
+      "step": 12797
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.900378137002494e-06,
+      "loss": 0.451,
+      "step": 12798
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898997103253127e-06,
+      "loss": 0.4433,
+      "step": 12799
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897616090760032e-06,
+      "loss": 0.4914,
+      "step": 12800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896235099549868e-06,
+      "loss": 0.4794,
+      "step": 12801
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.894854129649303e-06,
+      "loss": 0.4927,
+      "step": 12802
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.893473181084993e-06,
+      "loss": 0.4622,
+      "step": 12803
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.892092253883602e-06,
+      "loss": 0.492,
+      "step": 12804
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.890711348071784e-06,
+      "loss": 0.469,
+      "step": 12805
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.889330463676212e-06,
+      "loss": 0.4607,
+      "step": 12806
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887949600723533e-06,
+      "loss": 0.5014,
+      "step": 12807
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.886568759240411e-06,
+      "loss": 0.4803,
+      "step": 12808
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.885187939253504e-06,
+      "loss": 0.4679,
+      "step": 12809
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.883807140789478e-06,
+      "loss": 0.4652,
+      "step": 12810
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88242636387498e-06,
+      "loss": 0.4669,
+      "step": 12811
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.881045608536675e-06,
+      "loss": 0.4679,
+      "step": 12812
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.879664874801212e-06,
+      "loss": 0.437,
+      "step": 12813
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.878284162695263e-06,
+      "loss": 0.4973,
+      "step": 12814
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876903472245469e-06,
+      "loss": 0.457,
+      "step": 12815
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875522803478491e-06,
+      "loss": 0.4687,
+      "step": 12816
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87414215642099e-06,
+      "loss": 0.4803,
+      "step": 12817
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.872761531099615e-06,
+      "loss": 0.4868,
+      "step": 12818
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871380927541021e-06,
+      "loss": 0.4831,
+      "step": 12819
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.870000345771863e-06,
+      "loss": 0.4717,
+      "step": 12820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8686197858188e-06,
+      "loss": 0.4625,
+      "step": 12821
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.867239247708479e-06,
+      "loss": 0.4684,
+      "step": 12822
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865858731467554e-06,
+      "loss": 0.4693,
+      "step": 12823
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.864478237122677e-06,
+      "loss": 0.4832,
+      "step": 12824
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863097764700508e-06,
+      "loss": 0.4748,
+      "step": 12825
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861717314227689e-06,
+      "loss": 0.4899,
+      "step": 12826
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.860336885730873e-06,
+      "loss": 0.4548,
+      "step": 12827
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858956479236716e-06,
+      "loss": 0.4588,
+      "step": 12828
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85757609477187e-06,
+      "loss": 0.4634,
+      "step": 12829
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.856195732362975e-06,
+      "loss": 0.4759,
+      "step": 12830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.854815392036685e-06,
+      "loss": 0.4692,
+      "step": 12831
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.853435073819655e-06,
+      "loss": 0.444,
+      "step": 12832
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85205477773853e-06,
+      "loss": 0.4669,
+      "step": 12833
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850674503819954e-06,
+      "loss": 0.17,
+      "step": 12834
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84929425209058e-06,
+      "loss": 0.4812,
+      "step": 12835
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84791402257706e-06,
+      "loss": 0.4781,
+      "step": 12836
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84653381530603e-06,
+      "loss": 0.4809,
+      "step": 12837
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84515363030414e-06,
+      "loss": 0.4743,
+      "step": 12838
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84377346759804e-06,
+      "loss": 0.4678,
+      "step": 12839
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842393327214378e-06,
+      "loss": 0.4606,
+      "step": 12840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.841013209179793e-06,
+      "loss": 0.4698,
+      "step": 12841
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83963311352093e-06,
+      "loss": 0.4904,
+      "step": 12842
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838253040264439e-06,
+      "loss": 0.4715,
+      "step": 12843
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.836872989436965e-06,
+      "loss": 0.4911,
+      "step": 12844
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.835492961065139e-06,
+      "loss": 0.4533,
+      "step": 12845
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834112955175617e-06,
+      "loss": 0.454,
+      "step": 12846
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832732971795037e-06,
+      "loss": 0.4934,
+      "step": 12847
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.831353010950049e-06,
+      "loss": 0.4613,
+      "step": 12848
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.829973072667278e-06,
+      "loss": 0.4542,
+      "step": 12849
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.828593156973379e-06,
+      "loss": 0.4698,
+      "step": 12850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82721326389499e-06,
+      "loss": 0.4786,
+      "step": 12851
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.825833393458755e-06,
+      "loss": 0.4743,
+      "step": 12852
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.824453545691303e-06,
+      "loss": 0.4891,
+      "step": 12853
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.823073720619283e-06,
+      "loss": 0.4542,
+      "step": 12854
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.821693918269334e-06,
+      "loss": 0.5031,
+      "step": 12855
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.820314138668095e-06,
+      "loss": 0.4508,
+      "step": 12856
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8189343818422e-06,
+      "loss": 0.4627,
+      "step": 12857
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.81755464781829e-06,
+      "loss": 0.4698,
+      "step": 12858
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.816174936623005e-06,
+      "loss": 0.4752,
+      "step": 12859
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.814795248282974e-06,
+      "loss": 0.455,
+      "step": 12860
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813415582824842e-06,
+      "loss": 0.4748,
+      "step": 12861
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.812035940275239e-06,
+      "loss": 0.4779,
+      "step": 12862
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.810656320660809e-06,
+      "loss": 0.4455,
+      "step": 12863
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.809276724008178e-06,
+      "loss": 0.4836,
+      "step": 12864
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.807897150343985e-06,
+      "loss": 0.4443,
+      "step": 12865
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806517599694865e-06,
+      "loss": 0.4831,
+      "step": 12866
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.805138072087454e-06,
+      "loss": 0.4835,
+      "step": 12867
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80375856754838e-06,
+      "loss": 0.4641,
+      "step": 12868
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80237908610428e-06,
+      "loss": 0.4606,
+      "step": 12869
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.800999627781786e-06,
+      "loss": 0.485,
+      "step": 12870
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.799620192607531e-06,
+      "loss": 0.4875,
+      "step": 12871
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.798240780608143e-06,
+      "loss": 0.4769,
+      "step": 12872
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.796861391810259e-06,
+      "loss": 0.4918,
+      "step": 12873
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.795482026240505e-06,
+      "loss": 0.4983,
+      "step": 12874
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.794102683925515e-06,
+      "loss": 0.4619,
+      "step": 12875
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792723364891915e-06,
+      "loss": 0.4196,
+      "step": 12876
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.791344069166336e-06,
+      "loss": 0.473,
+      "step": 12877
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789964796775412e-06,
+      "loss": 0.4806,
+      "step": 12878
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.788585547745764e-06,
+      "loss": 0.4695,
+      "step": 12879
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787206322104025e-06,
+      "loss": 0.5167,
+      "step": 12880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.785827119876821e-06,
+      "loss": 0.4727,
+      "step": 12881
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.784447941090781e-06,
+      "loss": 0.4955,
+      "step": 12882
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78306878577253e-06,
+      "loss": 0.48,
+      "step": 12883
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.781689653948693e-06,
+      "loss": 0.4478,
+      "step": 12884
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780310545645897e-06,
+      "loss": 0.4615,
+      "step": 12885
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778931460890772e-06,
+      "loss": 0.4473,
+      "step": 12886
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.777552399709936e-06,
+      "loss": 0.4796,
+      "step": 12887
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.776173362130018e-06,
+      "loss": 0.4429,
+      "step": 12888
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.774794348177641e-06,
+      "loss": 0.4517,
+      "step": 12889
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.773415357879429e-06,
+      "loss": 0.519,
+      "step": 12890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772036391262003e-06,
+      "loss": 0.4779,
+      "step": 12891
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.770657448351987e-06,
+      "loss": 0.462,
+      "step": 12892
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769278529176005e-06,
+      "loss": 0.4704,
+      "step": 12893
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76789963376068e-06,
+      "loss": 0.465,
+      "step": 12894
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.766520762132627e-06,
+      "loss": 0.4633,
+      "step": 12895
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76514191431847e-06,
+      "loss": 0.4719,
+      "step": 12896
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.763763090344832e-06,
+      "loss": 0.4814,
+      "step": 12897
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76238429023833e-06,
+      "loss": 0.4552,
+      "step": 12898
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761005514025586e-06,
+      "loss": 0.471,
+      "step": 12899
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759626761733215e-06,
+      "loss": 0.4951,
+      "step": 12900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.758248033387841e-06,
+      "loss": 0.5046,
+      "step": 12901
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.756869329016078e-06,
+      "loss": 0.4948,
+      "step": 12902
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755490648644543e-06,
+      "loss": 0.4643,
+      "step": 12903
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754111992299858e-06,
+      "loss": 0.4707,
+      "step": 12904
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.752733360008635e-06,
+      "loss": 0.4494,
+      "step": 12905
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.751354751797492e-06,
+      "loss": 0.4613,
+      "step": 12906
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.749976167693045e-06,
+      "loss": 0.4883,
+      "step": 12907
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74859760772191e-06,
+      "loss": 0.4498,
+      "step": 12908
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.747219071910704e-06,
+      "loss": 0.4794,
+      "step": 12909
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.745840560286034e-06,
+      "loss": 0.4591,
+      "step": 12910
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744462072874519e-06,
+      "loss": 0.4892,
+      "step": 12911
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.743083609702772e-06,
+      "loss": 0.4784,
+      "step": 12912
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.741705170797407e-06,
+      "loss": 0.4643,
+      "step": 12913
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740326756185034e-06,
+      "loss": 0.5046,
+      "step": 12914
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.738948365892267e-06,
+      "loss": 0.4548,
+      "step": 12915
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.737569999945716e-06,
+      "loss": 0.4583,
+      "step": 12916
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736191658371998e-06,
+      "loss": 0.5022,
+      "step": 12917
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.734813341197713e-06,
+      "loss": 0.4821,
+      "step": 12918
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.733435048449479e-06,
+      "loss": 0.5033,
+      "step": 12919
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.732056780153904e-06,
+      "loss": 0.4779,
+      "step": 12920
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.730678536337599e-06,
+      "loss": 0.45,
+      "step": 12921
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.729300317027168e-06,
+      "loss": 0.482,
+      "step": 12922
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.727922122249221e-06,
+      "loss": 0.4788,
+      "step": 12923
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72654395203037e-06,
+      "loss": 0.4645,
+      "step": 12924
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725165806397216e-06,
+      "loss": 0.4671,
+      "step": 12925
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.723787685376371e-06,
+      "loss": 0.4831,
+      "step": 12926
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.722409588994437e-06,
+      "loss": 0.4642,
+      "step": 12927
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.721031517278027e-06,
+      "loss": 0.4938,
+      "step": 12928
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719653470253738e-06,
+      "loss": 0.4586,
+      "step": 12929
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.718275447948178e-06,
+      "loss": 0.4605,
+      "step": 12930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716897450387951e-06,
+      "loss": 0.4804,
+      "step": 12931
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.715519477599668e-06,
+      "loss": 0.4645,
+      "step": 12932
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.714141529609924e-06,
+      "loss": 0.4774,
+      "step": 12933
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.712763606445323e-06,
+      "loss": 0.4783,
+      "step": 12934
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71138570813247e-06,
+      "loss": 0.5078,
+      "step": 12935
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71000783469797e-06,
+      "loss": 0.4842,
+      "step": 12936
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.708629986168418e-06,
+      "loss": 0.4616,
+      "step": 12937
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.707252162570418e-06,
+      "loss": 0.4622,
+      "step": 12938
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.705874363930568e-06,
+      "loss": 0.4423,
+      "step": 12939
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.704496590275479e-06,
+      "loss": 0.4913,
+      "step": 12940
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.703118841631737e-06,
+      "loss": 0.4672,
+      "step": 12941
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.701741118025944e-06,
+      "loss": 0.4724,
+      "step": 12942
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.700363419484711e-06,
+      "loss": 0.4414,
+      "step": 12943
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69898574603462e-06,
+      "loss": 0.4475,
+      "step": 12944
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.697608097702277e-06,
+      "loss": 0.5002,
+      "step": 12945
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.696230474514274e-06,
+      "loss": 0.5005,
+      "step": 12946
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69485287649722e-06,
+      "loss": 0.4639,
+      "step": 12947
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.693475303677699e-06,
+      "loss": 0.485,
+      "step": 12948
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692097756082308e-06,
+      "loss": 0.4786,
+      "step": 12949
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.690720233737645e-06,
+      "loss": 0.4522,
+      "step": 12950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68934273667031e-06,
+      "loss": 0.4546,
+      "step": 12951
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687965264906891e-06,
+      "loss": 0.4539,
+      "step": 12952
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686587818473976e-06,
+      "loss": 0.444,
+      "step": 12953
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.685210397398172e-06,
+      "loss": 0.4665,
+      "step": 12954
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.683833001706068e-06,
+      "loss": 0.4807,
+      "step": 12955
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68245563142425e-06,
+      "loss": 0.4964,
+      "step": 12956
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68107828657931e-06,
+      "loss": 0.4771,
+      "step": 12957
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.679700967197846e-06,
+      "loss": 0.4779,
+      "step": 12958
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.678323673306452e-06,
+      "loss": 0.4807,
+      "step": 12959
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.676946404931706e-06,
+      "loss": 0.4788,
+      "step": 12960
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.675569162100202e-06,
+      "loss": 0.4799,
+      "step": 12961
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674191944838535e-06,
+      "loss": 0.4652,
+      "step": 12962
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672814753173295e-06,
+      "loss": 0.4481,
+      "step": 12963
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67143758713106e-06,
+      "loss": 0.4587,
+      "step": 12964
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670060446738425e-06,
+      "loss": 0.4746,
+      "step": 12965
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.668683332021981e-06,
+      "loss": 0.4822,
+      "step": 12966
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.667306243008307e-06,
+      "loss": 0.5145,
+      "step": 12967
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66592917972399e-06,
+      "loss": 0.4997,
+      "step": 12968
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66455214219562e-06,
+      "loss": 0.436,
+      "step": 12969
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66317513044979e-06,
+      "loss": 0.4838,
+      "step": 12970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661798144513065e-06,
+      "loss": 0.4921,
+      "step": 12971
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660421184412046e-06,
+      "loss": 0.4734,
+      "step": 12972
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65904425017331e-06,
+      "loss": 0.4823,
+      "step": 12973
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657667341823449e-06,
+      "loss": 0.446,
+      "step": 12974
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.656290459389032e-06,
+      "loss": 0.4668,
+      "step": 12975
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65491360289665e-06,
+      "loss": 0.4848,
+      "step": 12976
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.653536772372883e-06,
+      "loss": 0.4703,
+      "step": 12977
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65215996784432e-06,
+      "loss": 0.484,
+      "step": 12978
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.650783189337528e-06,
+      "loss": 0.4869,
+      "step": 12979
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.649406436879095e-06,
+      "loss": 0.4528,
+      "step": 12980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.648029710495604e-06,
+      "loss": 0.4627,
+      "step": 12981
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646653010213631e-06,
+      "loss": 0.4388,
+      "step": 12982
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.645276336059752e-06,
+      "loss": 0.4648,
+      "step": 12983
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.643899688060551e-06,
+      "loss": 0.5031,
+      "step": 12984
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642523066242607e-06,
+      "loss": 0.4829,
+      "step": 12985
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.641146470632487e-06,
+      "loss": 0.4596,
+      "step": 12986
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.639769901256778e-06,
+      "loss": 0.4836,
+      "step": 12987
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.638393358142052e-06,
+      "loss": 0.4752,
+      "step": 12988
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.637016841314891e-06,
+      "loss": 0.4669,
+      "step": 12989
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63564035080186e-06,
+      "loss": 0.4933,
+      "step": 12990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63426388662954e-06,
+      "loss": 0.4794,
+      "step": 12991
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.632887448824506e-06,
+      "loss": 0.4811,
+      "step": 12992
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.631511037413334e-06,
+      "loss": 0.4544,
+      "step": 12993
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.630134652422591e-06,
+      "loss": 0.4792,
+      "step": 12994
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.628758293878853e-06,
+      "loss": 0.4705,
+      "step": 12995
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.627381961808693e-06,
+      "loss": 0.4763,
+      "step": 12996
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.626005656238685e-06,
+      "loss": 0.4543,
+      "step": 12997
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624629377195394e-06,
+      "loss": 0.4784,
+      "step": 12998
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.623253124705397e-06,
+      "loss": 0.5045,
+      "step": 12999
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.62187689879526e-06,
+      "loss": 0.439,
+      "step": 13000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.620500699491556e-06,
+      "loss": 0.4849,
+      "step": 13001
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619124526820852e-06,
+      "loss": 0.478,
+      "step": 13002
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.617748380809718e-06,
+      "loss": 0.4757,
+      "step": 13003
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.616372261484722e-06,
+      "loss": 0.4946,
+      "step": 13004
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.614996168872434e-06,
+      "loss": 0.4674,
+      "step": 13005
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613620102999416e-06,
+      "loss": 0.465,
+      "step": 13006
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61224406389224e-06,
+      "loss": 0.4345,
+      "step": 13007
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61086805157747e-06,
+      "loss": 0.4684,
+      "step": 13008
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.609492066081668e-06,
+      "loss": 0.4734,
+      "step": 13009
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.608116107431405e-06,
+      "loss": 0.4798,
+      "step": 13010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.606740175653242e-06,
+      "loss": 0.468,
+      "step": 13011
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.605364270773747e-06,
+      "loss": 0.4785,
+      "step": 13012
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60398839281948e-06,
+      "loss": 0.1819,
+      "step": 13013
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.602612541817004e-06,
+      "loss": 0.4524,
+      "step": 13014
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.601236717792884e-06,
+      "loss": 0.4777,
+      "step": 13015
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.599860920773682e-06,
+      "loss": 0.4711,
+      "step": 13016
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.598485150785956e-06,
+      "loss": 0.4807,
+      "step": 13017
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.597109407856268e-06,
+      "loss": 0.4349,
+      "step": 13018
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.595733692011183e-06,
+      "loss": 0.5081,
+      "step": 13019
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.594358003277257e-06,
+      "loss": 0.4836,
+      "step": 13020
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.592982341681049e-06,
+      "loss": 0.4834,
+      "step": 13021
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59160670724912e-06,
+      "loss": 0.484,
+      "step": 13022
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.590231100008027e-06,
+      "loss": 0.4699,
+      "step": 13023
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.588855519984331e-06,
+      "loss": 0.4591,
+      "step": 13024
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.587479967204584e-06,
+      "loss": 0.4758,
+      "step": 13025
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586104441695346e-06,
+      "loss": 0.4792,
+      "step": 13026
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584728943483174e-06,
+      "loss": 0.4776,
+      "step": 13027
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58335347259462e-06,
+      "loss": 0.4715,
+      "step": 13028
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.581978029056242e-06,
+      "loss": 0.4518,
+      "step": 13029
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580602612894595e-06,
+      "loss": 0.4517,
+      "step": 13030
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.579227224136236e-06,
+      "loss": 0.4694,
+      "step": 13031
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577851862807712e-06,
+      "loss": 0.4861,
+      "step": 13032
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.576476528935578e-06,
+      "loss": 0.4787,
+      "step": 13033
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.575101222546389e-06,
+      "loss": 0.4776,
+      "step": 13034
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.573725943666698e-06,
+      "loss": 0.4695,
+      "step": 13035
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.572350692323053e-06,
+      "loss": 0.4636,
+      "step": 13036
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.570975468542007e-06,
+      "loss": 0.5067,
+      "step": 13037
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.569600272350109e-06,
+      "loss": 0.4728,
+      "step": 13038
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.568225103773912e-06,
+      "loss": 0.4606,
+      "step": 13039
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56684996283996e-06,
+      "loss": 0.4797,
+      "step": 13040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565474849574807e-06,
+      "loss": 0.4938,
+      "step": 13041
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.564099764004998e-06,
+      "loss": 0.489,
+      "step": 13042
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.562724706157086e-06,
+      "loss": 0.4524,
+      "step": 13043
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.561349676057612e-06,
+      "loss": 0.4814,
+      "step": 13044
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559974673733125e-06,
+      "loss": 0.476,
+      "step": 13045
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55859969921017e-06,
+      "loss": 0.4967,
+      "step": 13046
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.5572247525153e-06,
+      "loss": 0.5335,
+      "step": 13047
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.555849833675047e-06,
+      "loss": 0.4746,
+      "step": 13048
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.554474942715966e-06,
+      "loss": 0.4704,
+      "step": 13049
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.553100079664598e-06,
+      "loss": 0.4542,
+      "step": 13050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.551725244547487e-06,
+      "loss": 0.486,
+      "step": 13051
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550350437391173e-06,
+      "loss": 0.5068,
+      "step": 13052
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.5489756582222e-06,
+      "loss": 0.4584,
+      "step": 13053
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547600907067115e-06,
+      "loss": 0.4639,
+      "step": 13054
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.546226183952452e-06,
+      "loss": 0.4968,
+      "step": 13055
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.544851488904753e-06,
+      "loss": 0.4892,
+      "step": 13056
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54347682195056e-06,
+      "loss": 0.4418,
+      "step": 13057
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542102183116417e-06,
+      "loss": 0.4399,
+      "step": 13058
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.540727572428854e-06,
+      "loss": 0.4786,
+      "step": 13059
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.539352989914416e-06,
+      "loss": 0.4644,
+      "step": 13060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.537978435599636e-06,
+      "loss": 0.4844,
+      "step": 13061
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.536603909511062e-06,
+      "loss": 0.4562,
+      "step": 13062
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53522941167522e-06,
+      "loss": 0.4679,
+      "step": 13063
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53385494211865e-06,
+      "loss": 0.4634,
+      "step": 13064
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532480500867887e-06,
+      "loss": 0.4563,
+      "step": 13065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.531106087949472e-06,
+      "loss": 0.4567,
+      "step": 13066
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.529731703389933e-06,
+      "loss": 0.465,
+      "step": 13067
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528357347215804e-06,
+      "loss": 0.4782,
+      "step": 13068
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.526983019453624e-06,
+      "loss": 0.4925,
+      "step": 13069
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.525608720129928e-06,
+      "loss": 0.4709,
+      "step": 13070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.524234449271237e-06,
+      "loss": 0.487,
+      "step": 13071
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52286020690409e-06,
+      "loss": 0.4471,
+      "step": 13072
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.521485993055027e-06,
+      "loss": 0.4338,
+      "step": 13073
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.520111807750564e-06,
+      "loss": 0.4763,
+      "step": 13074
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.518737651017241e-06,
+      "loss": 0.4716,
+      "step": 13075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51736352288158e-06,
+      "loss": 0.4623,
+      "step": 13076
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515989423370123e-06,
+      "loss": 0.4387,
+      "step": 13077
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.514615352509386e-06,
+      "loss": 0.4836,
+      "step": 13078
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.5132413103259e-06,
+      "loss": 0.415,
+      "step": 13079
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511867296846197e-06,
+      "loss": 0.4702,
+      "step": 13080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.510493312096806e-06,
+      "loss": 0.4755,
+      "step": 13081
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.509119356104247e-06,
+      "loss": 0.4682,
+      "step": 13082
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.507745428895044e-06,
+      "loss": 0.4934,
+      "step": 13083
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50637153049573e-06,
+      "loss": 0.4873,
+      "step": 13084
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504997660932832e-06,
+      "loss": 0.4865,
+      "step": 13085
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.503623820232863e-06,
+      "loss": 0.4988,
+      "step": 13086
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.502250008422352e-06,
+      "loss": 0.4641,
+      "step": 13087
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.500876225527822e-06,
+      "loss": 0.4627,
+      "step": 13088
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.499502471575804e-06,
+      "loss": 0.5031,
+      "step": 13089
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.498128746592806e-06,
+      "loss": 0.4831,
+      "step": 13090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.496755050605358e-06,
+      "loss": 0.1892,
+      "step": 13091
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.495381383639985e-06,
+      "loss": 0.4652,
+      "step": 13092
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.494007745723197e-06,
+      "loss": 0.4674,
+      "step": 13093
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.492634136881514e-06,
+      "loss": 0.4814,
+      "step": 13094
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.491260557141462e-06,
+      "loss": 0.475,
+      "step": 13095
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.489887006529564e-06,
+      "loss": 0.4915,
+      "step": 13096
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.488513485072325e-06,
+      "loss": 0.469,
+      "step": 13097
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48713999279627e-06,
+      "loss": 0.4479,
+      "step": 13098
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.485766529727915e-06,
+      "loss": 0.48,
+      "step": 13099
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.484393095893781e-06,
+      "loss": 0.4916,
+      "step": 13100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.483019691320374e-06,
+      "loss": 0.4679,
+      "step": 13101
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.481646316034218e-06,
+      "loss": 0.4642,
+      "step": 13102
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.480272970061822e-06,
+      "loss": 0.1792,
+      "step": 13103
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.478899653429708e-06,
+      "loss": 0.5123,
+      "step": 13104
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.477526366164378e-06,
+      "loss": 0.4663,
+      "step": 13105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476153108292356e-06,
+      "loss": 0.4993,
+      "step": 13106
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47477987984015e-06,
+      "loss": 0.4722,
+      "step": 13107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47340668083427e-06,
+      "loss": 0.4315,
+      "step": 13108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472033511301232e-06,
+      "loss": 0.4855,
+      "step": 13109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47066037126754e-06,
+      "loss": 0.4372,
+      "step": 13110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.469287260759713e-06,
+      "loss": 0.4506,
+      "step": 13111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.467914179804255e-06,
+      "loss": 0.4413,
+      "step": 13112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466541128427674e-06,
+      "loss": 0.4872,
+      "step": 13113
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.465168106656482e-06,
+      "loss": 0.4476,
+      "step": 13114
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.463795114517189e-06,
+      "loss": 0.4739,
+      "step": 13115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.462422152036292e-06,
+      "loss": 0.4514,
+      "step": 13116
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461049219240309e-06,
+      "loss": 0.4549,
+      "step": 13117
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.459676316155741e-06,
+      "loss": 0.4612,
+      "step": 13118
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.458303442809096e-06,
+      "loss": 0.4888,
+      "step": 13119
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456930599226876e-06,
+      "loss": 0.4426,
+      "step": 13120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455557785435588e-06,
+      "loss": 0.481,
+      "step": 13121
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.454185001461735e-06,
+      "loss": 0.4729,
+      "step": 13122
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452812247331822e-06,
+      "loss": 0.4635,
+      "step": 13123
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.451439523072347e-06,
+      "loss": 0.4584,
+      "step": 13124
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450066828709818e-06,
+      "loss": 0.4707,
+      "step": 13125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.448694164270733e-06,
+      "loss": 0.4611,
+      "step": 13126
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.447321529781597e-06,
+      "loss": 0.4498,
+      "step": 13127
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.445948925268906e-06,
+      "loss": 0.4419,
+      "step": 13128
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444576350759159e-06,
+      "loss": 0.4776,
+      "step": 13129
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.443203806278859e-06,
+      "loss": 0.5063,
+      "step": 13130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.441831291854507e-06,
+      "loss": 0.4615,
+      "step": 13131
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.440458807512594e-06,
+      "loss": 0.464,
+      "step": 13132
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.439086353279623e-06,
+      "loss": 0.4705,
+      "step": 13133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43771392918209e-06,
+      "loss": 0.4751,
+      "step": 13134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.436341535246489e-06,
+      "loss": 0.4778,
+      "step": 13135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.434969171499316e-06,
+      "loss": 0.4379,
+      "step": 13136
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43359683796707e-06,
+      "loss": 0.4802,
+      "step": 13137
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432224534676244e-06,
+      "loss": 0.4799,
+      "step": 13138
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43085226165333e-06,
+      "loss": 0.4506,
+      "step": 13139
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.429480018924823e-06,
+      "loss": 0.4591,
+      "step": 13140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428107806517216e-06,
+      "loss": 0.4774,
+      "step": 13141
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426735624457003e-06,
+      "loss": 0.5013,
+      "step": 13142
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.425363472770672e-06,
+      "loss": 0.4391,
+      "step": 13143
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.423991351484715e-06,
+      "loss": 0.4633,
+      "step": 13144
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.422619260625626e-06,
+      "loss": 0.4399,
+      "step": 13145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.421247200219893e-06,
+      "loss": 0.4694,
+      "step": 13146
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.419875170294003e-06,
+      "loss": 0.4627,
+      "step": 13147
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418503170874446e-06,
+      "loss": 0.4437,
+      "step": 13148
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417131201987711e-06,
+      "loss": 0.4822,
+      "step": 13149
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41575926366029e-06,
+      "loss": 0.4755,
+      "step": 13150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.414387355918662e-06,
+      "loss": 0.4852,
+      "step": 13151
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.413015478789317e-06,
+      "loss": 0.4652,
+      "step": 13152
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41164363229874e-06,
+      "loss": 0.4351,
+      "step": 13153
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41027181647342e-06,
+      "loss": 0.4827,
+      "step": 13154
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408900031339838e-06,
+      "loss": 0.4415,
+      "step": 13155
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.407528276924478e-06,
+      "loss": 0.4843,
+      "step": 13156
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406156553253826e-06,
+      "loss": 0.4344,
+      "step": 13157
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40478486035436e-06,
+      "loss": 0.4734,
+      "step": 13158
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40341319825257e-06,
+      "loss": 0.4902,
+      "step": 13159
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40204156697493e-06,
+      "loss": 0.4357,
+      "step": 13160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.400669966547925e-06,
+      "loss": 0.4582,
+      "step": 13161
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.399298396998034e-06,
+      "loss": 0.4897,
+      "step": 13162
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39792685835174e-06,
+      "loss": 0.4457,
+      "step": 13163
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396555350635516e-06,
+      "loss": 0.4768,
+      "step": 13164
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.395183873875849e-06,
+      "loss": 0.4663,
+      "step": 13165
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.393812428099212e-06,
+      "loss": 0.5004,
+      "step": 13166
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.392441013332082e-06,
+      "loss": 0.4923,
+      "step": 13167
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.391069629600937e-06,
+      "loss": 0.4701,
+      "step": 13168
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.389698276932257e-06,
+      "loss": 0.4682,
+      "step": 13169
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.38832695535251e-06,
+      "loss": 0.465,
+      "step": 13170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.386955664888178e-06,
+      "loss": 0.4811,
+      "step": 13171
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.385584405565732e-06,
+      "loss": 0.4686,
+      "step": 13172
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.384213177411648e-06,
+      "loss": 0.4907,
+      "step": 13173
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.382841980452398e-06,
+      "loss": 0.4524,
+      "step": 13174
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.381470814714455e-06,
+      "loss": 0.4663,
+      "step": 13175
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.380099680224291e-06,
+      "loss": 0.4932,
+      "step": 13176
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378728577008376e-06,
+      "loss": 0.4716,
+      "step": 13177
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.377357505093183e-06,
+      "loss": 0.489,
+      "step": 13178
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.375986464505181e-06,
+      "loss": 0.4661,
+      "step": 13179
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.374615455270843e-06,
+      "loss": 0.4922,
+      "step": 13180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.373244477416634e-06,
+      "loss": 0.4698,
+      "step": 13181
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.371873530969025e-06,
+      "loss": 0.4784,
+      "step": 13182
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370502615954481e-06,
+      "loss": 0.4579,
+      "step": 13183
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.369131732399473e-06,
+      "loss": 0.4644,
+      "step": 13184
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367760880330465e-06,
+      "loss": 0.4726,
+      "step": 13185
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366390059773924e-06,
+      "loss": 0.4539,
+      "step": 13186
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.365019270756313e-06,
+      "loss": 0.4677,
+      "step": 13187
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.363648513304106e-06,
+      "loss": 0.4688,
+      "step": 13188
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.362277787443754e-06,
+      "loss": 0.4751,
+      "step": 13189
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36090709320173e-06,
+      "loss": 0.4507,
+      "step": 13190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35953643060449e-06,
+      "loss": 0.4685,
+      "step": 13191
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.358165799678506e-06,
+      "loss": 0.4749,
+      "step": 13192
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356795200450232e-06,
+      "loss": 0.4348,
+      "step": 13193
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355424632946129e-06,
+      "loss": 0.4804,
+      "step": 13194
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35405409719266e-06,
+      "loss": 0.4662,
+      "step": 13195
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.352683593216291e-06,
+      "loss": 0.462,
+      "step": 13196
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35131312104347e-06,
+      "loss": 0.4766,
+      "step": 13197
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.34994268070066e-06,
+      "loss": 0.4606,
+      "step": 13198
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.348572272214326e-06,
+      "loss": 0.4424,
+      "step": 13199
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347201895610915e-06,
+      "loss": 0.4699,
+      "step": 13200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345831550916887e-06,
+      "loss": 0.4779,
+      "step": 13201
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3444612381587e-06,
+      "loss": 0.4747,
+      "step": 13202
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.343090957362815e-06,
+      "loss": 0.4757,
+      "step": 13203
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341720708555676e-06,
+      "loss": 0.4667,
+      "step": 13204
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.340350491763745e-06,
+      "loss": 0.4498,
+      "step": 13205
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33898030701347e-06,
+      "loss": 0.4569,
+      "step": 13206
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.337610154331312e-06,
+      "loss": 0.4522,
+      "step": 13207
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.336240033743718e-06,
+      "loss": 0.4652,
+      "step": 13208
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.334869945277138e-06,
+      "loss": 0.464,
+      "step": 13209
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33349988895803e-06,
+      "loss": 0.4908,
+      "step": 13210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.332129864812843e-06,
+      "loss": 0.4663,
+      "step": 13211
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330759872868022e-06,
+      "loss": 0.4704,
+      "step": 13212
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.329389913150018e-06,
+      "loss": 0.4723,
+      "step": 13213
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328019985685286e-06,
+      "loss": 0.4704,
+      "step": 13214
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32665009050027e-06,
+      "loss": 0.4785,
+      "step": 13215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.325280227621412e-06,
+      "loss": 0.4488,
+      "step": 13216
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.323910397075169e-06,
+      "loss": 0.4351,
+      "step": 13217
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.322540598887979e-06,
+      "loss": 0.4813,
+      "step": 13218
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.321170833086298e-06,
+      "loss": 0.437,
+      "step": 13219
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.319801099696558e-06,
+      "loss": 0.4714,
+      "step": 13220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31843139874521e-06,
+      "loss": 0.4565,
+      "step": 13221
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317061730258704e-06,
+      "loss": 0.494,
+      "step": 13222
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.315692094263471e-06,
+      "loss": 0.4901,
+      "step": 13223
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31432249078596e-06,
+      "loss": 0.4813,
+      "step": 13224
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.312952919852612e-06,
+      "loss": 0.5033,
+      "step": 13225
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311583381489874e-06,
+      "loss": 0.4868,
+      "step": 13226
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.310213875724175e-06,
+      "loss": 0.4391,
+      "step": 13227
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308844402581964e-06,
+      "loss": 0.4659,
+      "step": 13228
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307474962089676e-06,
+      "loss": 0.4762,
+      "step": 13229
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.306105554273758e-06,
+      "loss": 0.4674,
+      "step": 13230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.304736179160635e-06,
+      "loss": 0.4719,
+      "step": 13231
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.303366836776753e-06,
+      "loss": 0.4902,
+      "step": 13232
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.301997527148549e-06,
+      "loss": 0.515,
+      "step": 13233
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.300628250302462e-06,
+      "loss": 0.4592,
+      "step": 13234
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299259006264915e-06,
+      "loss": 0.452,
+      "step": 13235
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.297889795062355e-06,
+      "loss": 0.4802,
+      "step": 13236
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.296520616721212e-06,
+      "loss": 0.4413,
+      "step": 13237
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.295151471267921e-06,
+      "loss": 0.4688,
+      "step": 13238
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.293782358728915e-06,
+      "loss": 0.4581,
+      "step": 13239
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.292413279130625e-06,
+      "loss": 0.4929,
+      "step": 13240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.291044232499488e-06,
+      "loss": 0.4989,
+      "step": 13241
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.289675218861924e-06,
+      "loss": 0.4621,
+      "step": 13242
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.288306238244374e-06,
+      "loss": 0.4658,
+      "step": 13243
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.286937290673266e-06,
+      "loss": 0.4955,
+      "step": 13244
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285568376175026e-06,
+      "loss": 0.4609,
+      "step": 13245
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.284199494776083e-06,
+      "loss": 0.4615,
+      "step": 13246
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.28283064650287e-06,
+      "loss": 0.5027,
+      "step": 13247
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.281461831381808e-06,
+      "loss": 0.4679,
+      "step": 13248
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.280093049439329e-06,
+      "loss": 0.4773,
+      "step": 13249
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.278724300701855e-06,
+      "loss": 0.4899,
+      "step": 13250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.277355585195812e-06,
+      "loss": 0.4713,
+      "step": 13251
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.275986902947628e-06,
+      "loss": 0.482,
+      "step": 13252
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.274618253983727e-06,
+      "loss": 0.4915,
+      "step": 13253
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.273249638330525e-06,
+      "loss": 0.4278,
+      "step": 13254
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.271881056014454e-06,
+      "loss": 0.4518,
+      "step": 13255
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270512507061931e-06,
+      "loss": 0.4992,
+      "step": 13256
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.269143991499383e-06,
+      "loss": 0.479,
+      "step": 13257
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.267775509353224e-06,
+      "loss": 0.4715,
+      "step": 13258
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266407060649877e-06,
+      "loss": 0.4498,
+      "step": 13259
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.265038645415764e-06,
+      "loss": 0.4748,
+      "step": 13260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263670263677302e-06,
+      "loss": 0.4568,
+      "step": 13261
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.262301915460909e-06,
+      "loss": 0.4597,
+      "step": 13262
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260933600793003e-06,
+      "loss": 0.4928,
+      "step": 13263
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.259565319700003e-06,
+      "loss": 0.4539,
+      "step": 13264
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258197072208322e-06,
+      "loss": 0.4882,
+      "step": 13265
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256828858344378e-06,
+      "loss": 0.4611,
+      "step": 13266
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.255460678134586e-06,
+      "loss": 0.4985,
+      "step": 13267
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.254092531605361e-06,
+      "loss": 0.5196,
+      "step": 13268
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.252724418783115e-06,
+      "loss": 0.4939,
+      "step": 13269
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25135633969426e-06,
+      "loss": 0.4838,
+      "step": 13270
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.249988294365213e-06,
+      "loss": 0.4668,
+      "step": 13271
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.248620282822386e-06,
+      "loss": 0.4654,
+      "step": 13272
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.247252305092182e-06,
+      "loss": 0.4764,
+      "step": 13273
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.245884361201022e-06,
+      "loss": 0.4583,
+      "step": 13274
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.244516451175307e-06,
+      "loss": 0.4654,
+      "step": 13275
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.243148575041456e-06,
+      "loss": 0.4972,
+      "step": 13276
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.241780732825867e-06,
+      "loss": 0.4566,
+      "step": 13277
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240412924554955e-06,
+      "loss": 0.5023,
+      "step": 13278
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239045150255124e-06,
+      "loss": 0.4722,
+      "step": 13279
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237677409952784e-06,
+      "loss": 0.4669,
+      "step": 13280
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23630970367434e-06,
+      "loss": 0.4562,
+      "step": 13281
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.234942031446191e-06,
+      "loss": 0.4909,
+      "step": 13282
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.233574393294752e-06,
+      "loss": 0.4733,
+      "step": 13283
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23220678924642e-06,
+      "loss": 0.5128,
+      "step": 13284
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2308392193276e-06,
+      "loss": 0.4771,
+      "step": 13285
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.229471683564694e-06,
+      "loss": 0.4591,
+      "step": 13286
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.228104181984108e-06,
+      "loss": 0.4405,
+      "step": 13287
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.226736714612238e-06,
+      "loss": 0.4555,
+      "step": 13288
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.225369281475487e-06,
+      "loss": 0.4725,
+      "step": 13289
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224001882600254e-06,
+      "loss": 0.4796,
+      "step": 13290
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222634518012944e-06,
+      "loss": 0.4416,
+      "step": 13291
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.221267187739947e-06,
+      "loss": 0.4858,
+      "step": 13292
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.219899891807666e-06,
+      "loss": 0.4371,
+      "step": 13293
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.218532630242498e-06,
+      "loss": 0.4578,
+      "step": 13294
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21716540307084e-06,
+      "loss": 0.4874,
+      "step": 13295
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.215798210319089e-06,
+      "loss": 0.479,
+      "step": 13296
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214431052013636e-06,
+      "loss": 0.471,
+      "step": 13297
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.213063928180877e-06,
+      "loss": 0.46,
+      "step": 13298
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.211696838847213e-06,
+      "loss": 0.4669,
+      "step": 13299
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210329784039028e-06,
+      "loss": 0.4813,
+      "step": 13300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.208962763782719e-06,
+      "loss": 0.4891,
+      "step": 13301
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20759577810468e-06,
+      "loss": 0.476,
+      "step": 13302
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2062288270313e-06,
+      "loss": 0.4538,
+      "step": 13303
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20486191058897e-06,
+      "loss": 0.4679,
+      "step": 13304
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.203495028804079e-06,
+      "loss": 0.4839,
+      "step": 13305
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20212818170302e-06,
+      "loss": 0.4679,
+      "step": 13306
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200761369312177e-06,
+      "loss": 0.4581,
+      "step": 13307
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19939459165794e-06,
+      "loss": 0.4612,
+      "step": 13308
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198027848766696e-06,
+      "loss": 0.4625,
+      "step": 13309
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.196661140664836e-06,
+      "loss": 0.4483,
+      "step": 13310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195294467378741e-06,
+      "loss": 0.4796,
+      "step": 13311
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.193927828934798e-06,
+      "loss": 0.4864,
+      "step": 13312
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.192561225359387e-06,
+      "loss": 0.4897,
+      "step": 13313
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.191194656678905e-06,
+      "loss": 0.5072,
+      "step": 13314
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.189828122919721e-06,
+      "loss": 0.1736,
+      "step": 13315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188461624108226e-06,
+      "loss": 0.4491,
+      "step": 13316
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187095160270797e-06,
+      "loss": 0.4983,
+      "step": 13317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185728731433821e-06,
+      "loss": 0.4617,
+      "step": 13318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.184362337623675e-06,
+      "loss": 0.4928,
+      "step": 13319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.182995978866738e-06,
+      "loss": 0.4952,
+      "step": 13320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.181629655189391e-06,
+      "loss": 0.4847,
+      "step": 13321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.180263366618017e-06,
+      "loss": 0.5043,
+      "step": 13322
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178897113178988e-06,
+      "loss": 0.491,
+      "step": 13323
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.177530894898677e-06,
+      "loss": 0.4782,
+      "step": 13324
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.176164711803474e-06,
+      "loss": 0.4988,
+      "step": 13325
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.174798563919745e-06,
+      "loss": 0.4995,
+      "step": 13326
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.173432451273867e-06,
+      "loss": 0.4731,
+      "step": 13327
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.172066373892211e-06,
+      "loss": 0.4484,
+      "step": 13328
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.170700331801162e-06,
+      "loss": 0.4663,
+      "step": 13329
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169334325027084e-06,
+      "loss": 0.452,
+      "step": 13330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16796835359635e-06,
+      "loss": 0.4748,
+      "step": 13331
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16660241753533e-06,
+      "loss": 0.4561,
+      "step": 13332
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.165236516870407e-06,
+      "loss": 0.4692,
+      "step": 13333
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.163870651627938e-06,
+      "loss": 0.4781,
+      "step": 13334
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162504821834296e-06,
+      "loss": 0.4617,
+      "step": 13335
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161139027515856e-06,
+      "loss": 0.4826,
+      "step": 13336
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.159773268698984e-06,
+      "loss": 0.4799,
+      "step": 13337
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.158407545410043e-06,
+      "loss": 0.4688,
+      "step": 13338
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1570418576754e-06,
+      "loss": 0.4552,
+      "step": 13339
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.155676205521428e-06,
+      "loss": 0.4612,
+      "step": 13340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.154310588974493e-06,
+      "loss": 0.4241,
+      "step": 13341
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152945008060953e-06,
+      "loss": 0.4591,
+      "step": 13342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15157946280717e-06,
+      "loss": 0.4553,
+      "step": 13343
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15021395323952e-06,
+      "loss": 0.4713,
+      "step": 13344
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14884847938436e-06,
+      "loss": 0.4567,
+      "step": 13345
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147483041268045e-06,
+      "loss": 0.4606,
+      "step": 13346
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.146117638916945e-06,
+      "loss": 0.4688,
+      "step": 13347
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.144752272357424e-06,
+      "loss": 0.4894,
+      "step": 13348
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.143386941615832e-06,
+      "loss": 0.4525,
+      "step": 13349
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14202164671853e-06,
+      "loss": 0.46,
+      "step": 13350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.140656387691882e-06,
+      "loss": 0.4405,
+      "step": 13351
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13929116456225e-06,
+      "loss": 0.4734,
+      "step": 13352
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.137925977355977e-06,
+      "loss": 0.4782,
+      "step": 13353
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13656082609943e-06,
+      "loss": 0.4614,
+      "step": 13354
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.135195710818966e-06,
+      "loss": 0.4821,
+      "step": 13355
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13383063154094e-06,
+      "loss": 0.4435,
+      "step": 13356
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132465588291697e-06,
+      "loss": 0.4693,
+      "step": 13357
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.1311005810976e-06,
+      "loss": 0.4727,
+      "step": 13358
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.129735609985e-06,
+      "loss": 0.4732,
+      "step": 13359
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128370674980255e-06,
+      "loss": 0.4447,
+      "step": 13360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.127005776109704e-06,
+      "loss": 0.4848,
+      "step": 13361
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125640913399708e-06,
+      "loss": 0.4539,
+      "step": 13362
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.124276086876616e-06,
+      "loss": 0.4797,
+      "step": 13363
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122911296566779e-06,
+      "loss": 0.4495,
+      "step": 13364
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12154654249654e-06,
+      "loss": 0.4671,
+      "step": 13365
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12018182469225e-06,
+      "loss": 0.4448,
+      "step": 13366
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11881714318026e-06,
+      "loss": 0.1781,
+      "step": 13367
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.117452497986916e-06,
+      "loss": 0.4541,
+      "step": 13368
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11608788913856e-06,
+      "loss": 0.4578,
+      "step": 13369
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.114723316661541e-06,
+      "loss": 0.435,
+      "step": 13370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113358780582207e-06,
+      "loss": 0.4465,
+      "step": 13371
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.111994280926893e-06,
+      "loss": 0.4838,
+      "step": 13372
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.110629817721949e-06,
+      "loss": 0.478,
+      "step": 13373
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109265390993717e-06,
+      "loss": 0.4745,
+      "step": 13374
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10790100076854e-06,
+      "loss": 0.5055,
+      "step": 13375
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.106536647072756e-06,
+      "loss": 0.491,
+      "step": 13376
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.105172329932706e-06,
+      "loss": 0.4677,
+      "step": 13377
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.103808049374732e-06,
+      "loss": 0.4502,
+      "step": 13378
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.102443805425176e-06,
+      "loss": 0.4584,
+      "step": 13379
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101079598110368e-06,
+      "loss": 0.4508,
+      "step": 13380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099715427456651e-06,
+      "loss": 0.4668,
+      "step": 13381
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.098351293490365e-06,
+      "loss": 0.4382,
+      "step": 13382
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096987196237842e-06,
+      "loss": 0.4688,
+      "step": 13383
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.095623135725417e-06,
+      "loss": 0.4881,
+      "step": 13384
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094259111979427e-06,
+      "loss": 0.491,
+      "step": 13385
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.092895125026206e-06,
+      "loss": 0.1649,
+      "step": 13386
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.091531174892088e-06,
+      "loss": 0.4469,
+      "step": 13387
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.090167261603406e-06,
+      "loss": 0.4632,
+      "step": 13388
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.088803385186489e-06,
+      "loss": 0.4713,
+      "step": 13389
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.087439545667674e-06,
+      "loss": 0.4742,
+      "step": 13390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.086075743073285e-06,
+      "loss": 0.4431,
+      "step": 13391
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084711977429655e-06,
+      "loss": 0.4573,
+      "step": 13392
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.083348248763115e-06,
+      "loss": 0.4673,
+      "step": 13393
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.081984557099993e-06,
+      "loss": 0.4805,
+      "step": 13394
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.080620902466615e-06,
+      "loss": 0.5232,
+      "step": 13395
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.079257284889308e-06,
+      "loss": 0.4752,
+      "step": 13396
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.0778937043944e-06,
+      "loss": 0.47,
+      "step": 13397
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.076530161008217e-06,
+      "loss": 0.4884,
+      "step": 13398
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07516665475708e-06,
+      "loss": 0.4716,
+      "step": 13399
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07380318566732e-06,
+      "loss": 0.4753,
+      "step": 13400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.072439753765252e-06,
+      "loss": 0.4825,
+      "step": 13401
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07107635907721e-06,
+      "loss": 0.477,
+      "step": 13402
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.069713001629504e-06,
+      "loss": 0.5027,
+      "step": 13403
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.068349681448463e-06,
+      "loss": 0.4694,
+      "step": 13404
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.066986398560405e-06,
+      "loss": 0.4966,
+      "step": 13405
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065623152991652e-06,
+      "loss": 0.4728,
+      "step": 13406
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06425994476852e-06,
+      "loss": 0.4487,
+      "step": 13407
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06289677391733e-06,
+      "loss": 0.4829,
+      "step": 13408
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0615336404644e-06,
+      "loss": 0.4832,
+      "step": 13409
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060170544436045e-06,
+      "loss": 0.477,
+      "step": 13410
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05880748585858e-06,
+      "loss": 0.4458,
+      "step": 13411
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057444464758325e-06,
+      "loss": 0.4872,
+      "step": 13412
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.056081481161594e-06,
+      "loss": 0.4787,
+      "step": 13413
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0547185350947e-06,
+      "loss": 0.4745,
+      "step": 13414
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053355626583954e-06,
+      "loss": 0.4819,
+      "step": 13415
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.051992755655672e-06,
+      "loss": 0.4585,
+      "step": 13416
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.050629922336167e-06,
+      "loss": 0.479,
+      "step": 13417
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.049267126651745e-06,
+      "loss": 0.4657,
+      "step": 13418
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04790436862872e-06,
+      "loss": 0.4784,
+      "step": 13419
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.046541648293402e-06,
+      "loss": 0.4948,
+      "step": 13420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0451789656721e-06,
+      "loss": 0.4827,
+      "step": 13421
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043816320791123e-06,
+      "loss": 0.4483,
+      "step": 13422
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.042453713676777e-06,
+      "loss": 0.4575,
+      "step": 13423
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041091144355367e-06,
+      "loss": 0.4932,
+      "step": 13424
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.039728612853206e-06,
+      "loss": 0.4652,
+      "step": 13425
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.038366119196592e-06,
+      "loss": 0.4928,
+      "step": 13426
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.037003663411833e-06,
+      "loss": 0.4905,
+      "step": 13427
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.035641245525233e-06,
+      "loss": 0.4627,
+      "step": 13428
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034278865563096e-06,
+      "loss": 0.4671,
+      "step": 13429
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03291652355172e-06,
+      "loss": 0.1655,
+      "step": 13430
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031554219517413e-06,
+      "loss": 0.4748,
+      "step": 13431
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030191953486473e-06,
+      "loss": 0.4427,
+      "step": 13432
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0288297254852e-06,
+      "loss": 0.4484,
+      "step": 13433
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.027467535539893e-06,
+      "loss": 0.4594,
+      "step": 13434
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026105383676851e-06,
+      "loss": 0.457,
+      "step": 13435
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.024743269922375e-06,
+      "loss": 0.4465,
+      "step": 13436
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.02338119430276e-06,
+      "loss": 0.4459,
+      "step": 13437
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0220191568443e-06,
+      "loss": 0.4714,
+      "step": 13438
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.020657157573297e-06,
+      "loss": 0.4428,
+      "step": 13439
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.019295196516044e-06,
+      "loss": 0.4644,
+      "step": 13440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01793327369883e-06,
+      "loss": 0.4826,
+      "step": 13441
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.016571389147955e-06,
+      "loss": 0.4702,
+      "step": 13442
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.015209542889706e-06,
+      "loss": 0.4683,
+      "step": 13443
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.013847734950388e-06,
+      "loss": 0.4985,
+      "step": 13444
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012485965356278e-06,
+      "loss": 0.4495,
+      "step": 13445
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011124234133672e-06,
+      "loss": 0.4399,
+      "step": 13446
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.009762541308858e-06,
+      "loss": 0.4512,
+      "step": 13447
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008400886908132e-06,
+      "loss": 0.4753,
+      "step": 13448
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.007039270957778e-06,
+      "loss": 0.4709,
+      "step": 13449
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005677693484077e-06,
+      "loss": 0.4841,
+      "step": 13450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.004316154513327e-06,
+      "loss": 0.4744,
+      "step": 13451
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002954654071813e-06,
+      "loss": 0.4483,
+      "step": 13452
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.001593192185815e-06,
+      "loss": 0.4801,
+      "step": 13453
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.000231768881615e-06,
+      "loss": 0.4662,
+      "step": 13454
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99887038418551e-06,
+      "loss": 0.4882,
+      "step": 13455
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.997509038123771e-06,
+      "loss": 0.4377,
+      "step": 13456
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.996147730722686e-06,
+      "loss": 0.4691,
+      "step": 13457
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.994786462008534e-06,
+      "loss": 0.459,
+      "step": 13458
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.993425232007603e-06,
+      "loss": 0.4478,
+      "step": 13459
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.992064040746163e-06,
+      "loss": 0.463,
+      "step": 13460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.990702888250497e-06,
+      "loss": 0.4576,
+      "step": 13461
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.989341774546889e-06,
+      "loss": 0.4645,
+      "step": 13462
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.987980699661616e-06,
+      "loss": 0.4712,
+      "step": 13463
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.986619663620949e-06,
+      "loss": 0.479,
+      "step": 13464
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.985258666451166e-06,
+      "loss": 0.4796,
+      "step": 13465
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.983897708178547e-06,
+      "loss": 0.4586,
+      "step": 13466
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98253678882937e-06,
+      "loss": 0.4703,
+      "step": 13467
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.9811759084299e-06,
+      "loss": 0.4703,
+      "step": 13468
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979815067006412e-06,
+      "loss": 0.4293,
+      "step": 13469
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978454264585185e-06,
+      "loss": 0.4517,
+      "step": 13470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.977093501192493e-06,
+      "loss": 0.4541,
+      "step": 13471
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975732776854592e-06,
+      "loss": 0.4752,
+      "step": 13472
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.974372091597768e-06,
+      "loss": 0.4888,
+      "step": 13473
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.973011445448283e-06,
+      "loss": 0.4441,
+      "step": 13474
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971650838432414e-06,
+      "loss": 0.4583,
+      "step": 13475
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.970290270576418e-06,
+      "loss": 0.4627,
+      "step": 13476
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.968929741906569e-06,
+      "loss": 0.4605,
+      "step": 13477
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.967569252449137e-06,
+      "loss": 0.4904,
+      "step": 13478
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96620880223038e-06,
+      "loss": 0.4792,
+      "step": 13479
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964848391276565e-06,
+      "loss": 0.4512,
+      "step": 13480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.96348801961396e-06,
+      "loss": 0.4688,
+      "step": 13481
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.962127687268832e-06,
+      "loss": 0.4579,
+      "step": 13482
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960767394267433e-06,
+      "loss": 0.4677,
+      "step": 13483
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.959407140636034e-06,
+      "loss": 0.4759,
+      "step": 13484
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.958046926400893e-06,
+      "loss": 0.4841,
+      "step": 13485
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956686751588277e-06,
+      "loss": 0.4857,
+      "step": 13486
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.955326616224432e-06,
+      "loss": 0.4683,
+      "step": 13487
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95396652033563e-06,
+      "loss": 0.4688,
+      "step": 13488
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952606463948123e-06,
+      "loss": 0.4985,
+      "step": 13489
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.951246447088174e-06,
+      "loss": 0.4657,
+      "step": 13490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.949886469782035e-06,
+      "loss": 0.4474,
+      "step": 13491
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.948526532055963e-06,
+      "loss": 0.492,
+      "step": 13492
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.947166633936215e-06,
+      "loss": 0.4739,
+      "step": 13493
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.945806775449043e-06,
+      "loss": 0.4873,
+      "step": 13494
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.944446956620703e-06,
+      "loss": 0.4789,
+      "step": 13495
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94308717747745e-06,
+      "loss": 0.4838,
+      "step": 13496
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941727438045533e-06,
+      "loss": 0.4614,
+      "step": 13497
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940367738351201e-06,
+      "loss": 0.4586,
+      "step": 13498
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.939008078420707e-06,
+      "loss": 0.466,
+      "step": 13499
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.937648458280305e-06,
+      "loss": 0.4792,
+      "step": 13500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.936288877956243e-06,
+      "loss": 0.4648,
+      "step": 13501
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934929337474764e-06,
+      "loss": 0.4591,
+      "step": 13502
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93356983686212e-06,
+      "loss": 0.4687,
+      "step": 13503
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.932210376144556e-06,
+      "loss": 0.4815,
+      "step": 13504
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930850955348321e-06,
+      "loss": 0.4444,
+      "step": 13505
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.929491574499659e-06,
+      "loss": 0.495,
+      "step": 13506
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.928132233624813e-06,
+      "loss": 0.4608,
+      "step": 13507
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926772932750027e-06,
+      "loss": 0.4723,
+      "step": 13508
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.925413671901548e-06,
+      "loss": 0.4995,
+      "step": 13509
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.924054451105614e-06,
+      "loss": 0.4703,
+      "step": 13510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.922695270388467e-06,
+      "loss": 0.4968,
+      "step": 13511
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.921336129776349e-06,
+      "loss": 0.4871,
+      "step": 13512
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919977029295501e-06,
+      "loss": 0.4896,
+      "step": 13513
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.918617968972158e-06,
+      "loss": 0.4533,
+      "step": 13514
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917258948832563e-06,
+      "loss": 0.4831,
+      "step": 13515
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91589996890295e-06,
+      "loss": 0.4699,
+      "step": 13516
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.914541029209561e-06,
+      "loss": 0.4902,
+      "step": 13517
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.913182129778625e-06,
+      "loss": 0.4684,
+      "step": 13518
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.911823270636384e-06,
+      "loss": 0.456,
+      "step": 13519
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.910464451809068e-06,
+      "loss": 0.4577,
+      "step": 13520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909105673322911e-06,
+      "loss": 0.4749,
+      "step": 13521
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907746935204149e-06,
+      "loss": 0.4409,
+      "step": 13522
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90638823747901e-06,
+      "loss": 0.4549,
+      "step": 13523
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90502958017373e-06,
+      "loss": 0.4388,
+      "step": 13524
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.903670963314536e-06,
+      "loss": 0.4802,
+      "step": 13525
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902312386927659e-06,
+      "loss": 0.4876,
+      "step": 13526
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.900953851039328e-06,
+      "loss": 0.4316,
+      "step": 13527
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899595355675775e-06,
+      "loss": 0.4736,
+      "step": 13528
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89823690086322e-06,
+      "loss": 0.4553,
+      "step": 13529
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896878486627894e-06,
+      "loss": 0.4845,
+      "step": 13530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895520112996024e-06,
+      "loss": 0.4646,
+      "step": 13531
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.894161779993835e-06,
+      "loss": 0.4514,
+      "step": 13532
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892803487647547e-06,
+      "loss": 0.4512,
+      "step": 13533
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.891445235983388e-06,
+      "loss": 0.4522,
+      "step": 13534
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89008702502758e-06,
+      "loss": 0.4621,
+      "step": 13535
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888728854806346e-06,
+      "loss": 0.4608,
+      "step": 13536
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887370725345903e-06,
+      "loss": 0.5175,
+      "step": 13537
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.886012636672474e-06,
+      "loss": 0.4642,
+      "step": 13538
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.884654588812282e-06,
+      "loss": 0.4787,
+      "step": 13539
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88329658179154e-06,
+      "loss": 0.4531,
+      "step": 13540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.881938615636469e-06,
+      "loss": 0.4829,
+      "step": 13541
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.880580690373284e-06,
+      "loss": 0.4635,
+      "step": 13542
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.879222806028209e-06,
+      "loss": 0.4641,
+      "step": 13543
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87786496262745e-06,
+      "loss": 0.4663,
+      "step": 13544
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876507160197226e-06,
+      "loss": 0.4749,
+      "step": 13545
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.875149398763753e-06,
+      "loss": 0.4553,
+      "step": 13546
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873791678353242e-06,
+      "loss": 0.5178,
+      "step": 13547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.872433998991904e-06,
+      "loss": 0.4495,
+      "step": 13548
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.871076360705956e-06,
+      "loss": 0.4477,
+      "step": 13549
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.869718763521604e-06,
+      "loss": 0.4799,
+      "step": 13550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868361207465061e-06,
+      "loss": 0.4757,
+      "step": 13551
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.867003692562533e-06,
+      "loss": 0.4427,
+      "step": 13552
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.865646218840232e-06,
+      "loss": 0.4247,
+      "step": 13553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.864288786324364e-06,
+      "loss": 0.4675,
+      "step": 13554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.862931395041139e-06,
+      "loss": 0.4596,
+      "step": 13555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.861574045016757e-06,
+      "loss": 0.4629,
+      "step": 13556
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86021673627743e-06,
+      "loss": 0.4478,
+      "step": 13557
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.858859468849357e-06,
+      "loss": 0.1827,
+      "step": 13558
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85750224275875e-06,
+      "loss": 0.4861,
+      "step": 13559
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.856145058031802e-06,
+      "loss": 0.4622,
+      "step": 13560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85478791469472e-06,
+      "loss": 0.4639,
+      "step": 13561
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.853430812773706e-06,
+      "loss": 0.4345,
+      "step": 13562
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.852073752294958e-06,
+      "loss": 0.468,
+      "step": 13563
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.850716733284677e-06,
+      "loss": 0.4684,
+      "step": 13564
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849359755769063e-06,
+      "loss": 0.4696,
+      "step": 13565
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.848002819774316e-06,
+      "loss": 0.4482,
+      "step": 13566
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84664592532663e-06,
+      "loss": 0.4614,
+      "step": 13567
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8452890724522e-06,
+      "loss": 0.4956,
+      "step": 13568
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843932261177224e-06,
+      "loss": 0.4599,
+      "step": 13569
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.842575491527903e-06,
+      "loss": 0.4725,
+      "step": 13570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.841218763530423e-06,
+      "loss": 0.4605,
+      "step": 13571
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839862077210979e-06,
+      "loss": 0.449,
+      "step": 13572
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83850543259576e-06,
+      "loss": 0.4887,
+      "step": 13573
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837148829710972e-06,
+      "loss": 0.4677,
+      "step": 13574
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835792268582789e-06,
+      "loss": 0.4538,
+      "step": 13575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83443574923741e-06,
+      "loss": 0.4884,
+      "step": 13576
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833079271701021e-06,
+      "loss": 0.4699,
+      "step": 13577
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.831722835999818e-06,
+      "loss": 0.4623,
+      "step": 13578
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.830366442159978e-06,
+      "loss": 0.4922,
+      "step": 13579
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82901009020769e-06,
+      "loss": 0.4664,
+      "step": 13580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827653780169152e-06,
+      "loss": 0.4658,
+      "step": 13581
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.826297512070533e-06,
+      "loss": 0.4608,
+      "step": 13582
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.824941285938027e-06,
+      "loss": 0.463,
+      "step": 13583
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.823585101797812e-06,
+      "loss": 0.4501,
+      "step": 13584
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822228959676082e-06,
+      "loss": 0.4558,
+      "step": 13585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.820872859599006e-06,
+      "loss": 0.4412,
+      "step": 13586
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81951680159277e-06,
+      "loss": 0.46,
+      "step": 13587
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818160785683555e-06,
+      "loss": 0.4737,
+      "step": 13588
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.816804811897547e-06,
+      "loss": 0.4883,
+      "step": 13589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.815448880260912e-06,
+      "loss": 0.4601,
+      "step": 13590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814092990799833e-06,
+      "loss": 0.4551,
+      "step": 13591
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.812737143540492e-06,
+      "loss": 0.4755,
+      "step": 13592
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.811381338509066e-06,
+      "loss": 0.472,
+      "step": 13593
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810025575731721e-06,
+      "loss": 0.4571,
+      "step": 13594
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808669855234636e-06,
+      "loss": 0.462,
+      "step": 13595
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.807314177043988e-06,
+      "loss": 0.4775,
+      "step": 13596
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.805958541185955e-06,
+      "loss": 0.4459,
+      "step": 13597
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804602947686694e-06,
+      "loss": 0.4468,
+      "step": 13598
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803247396572386e-06,
+      "loss": 0.4769,
+      "step": 13599
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801891887869205e-06,
+      "loss": 0.4483,
+      "step": 13600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800536421603317e-06,
+      "loss": 0.4682,
+      "step": 13601
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.799180997800885e-06,
+      "loss": 0.4717,
+      "step": 13602
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.797825616488086e-06,
+      "loss": 0.4449,
+      "step": 13603
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796470277691087e-06,
+      "loss": 0.4724,
+      "step": 13604
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795114981436048e-06,
+      "loss": 0.4753,
+      "step": 13605
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.793759727749137e-06,
+      "loss": 0.4624,
+      "step": 13606
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.792404516656523e-06,
+      "loss": 0.4877,
+      "step": 13607
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79104934818437e-06,
+      "loss": 0.5103,
+      "step": 13608
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.789694222358833e-06,
+      "loss": 0.4415,
+      "step": 13609
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788339139206083e-06,
+      "loss": 0.4612,
+      "step": 13610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.786984098752278e-06,
+      "loss": 0.4588,
+      "step": 13611
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.785629101023585e-06,
+      "loss": 0.4687,
+      "step": 13612
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784274146046149e-06,
+      "loss": 0.444,
+      "step": 13613
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.782919233846144e-06,
+      "loss": 0.4826,
+      "step": 13614
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78156436444972e-06,
+      "loss": 0.4742,
+      "step": 13615
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780209537883045e-06,
+      "loss": 0.4364,
+      "step": 13616
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77885475417226e-06,
+      "loss": 0.4634,
+      "step": 13617
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.777500013343532e-06,
+      "loss": 0.4556,
+      "step": 13618
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.776145315423011e-06,
+      "loss": 0.4728,
+      "step": 13619
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.4741,
+      "step": 13620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.773436048411217e-06,
+      "loss": 0.4589,
+      "step": 13621
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772081479372246e-06,
+      "loss": 0.4471,
+      "step": 13622
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.770726953346097e-06,
+      "loss": 0.4829,
+      "step": 13623
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76937247035892e-06,
+      "loss": 0.4826,
+      "step": 13624
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.768018030436862e-06,
+      "loss": 0.4589,
+      "step": 13625
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.766663633606076e-06,
+      "loss": 0.4868,
+      "step": 13626
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.765309279892711e-06,
+      "loss": 0.4725,
+      "step": 13627
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76395496932291e-06,
+      "loss": 0.4639,
+      "step": 13628
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.762600701922823e-06,
+      "loss": 0.4831,
+      "step": 13629
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.761246477718595e-06,
+      "loss": 0.4385,
+      "step": 13630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.759892296736375e-06,
+      "loss": 0.4553,
+      "step": 13631
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.758538159002298e-06,
+      "loss": 0.4816,
+      "step": 13632
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.757184064542515e-06,
+      "loss": 0.4585,
+      "step": 13633
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755830013383168e-06,
+      "loss": 0.484,
+      "step": 13634
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.754476005550396e-06,
+      "loss": 0.4577,
+      "step": 13635
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75312204107034e-06,
+      "loss": 0.4441,
+      "step": 13636
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.751768119969142e-06,
+      "loss": 0.444,
+      "step": 13637
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.750414242272939e-06,
+      "loss": 0.4934,
+      "step": 13638
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.749060408007874e-06,
+      "loss": 0.4756,
+      "step": 13639
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.747706617200078e-06,
+      "loss": 0.4738,
+      "step": 13640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.746352869875689e-06,
+      "loss": 0.4687,
+      "step": 13641
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.744999166060847e-06,
+      "loss": 0.4705,
+      "step": 13642
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.743645505781685e-06,
+      "loss": 0.4465,
+      "step": 13643
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742291889064334e-06,
+      "loss": 0.474,
+      "step": 13644
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.740938315934932e-06,
+      "loss": 0.4925,
+      "step": 13645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73958478641961e-06,
+      "loss": 0.4681,
+      "step": 13646
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.738231300544496e-06,
+      "loss": 0.4777,
+      "step": 13647
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.736877858335725e-06,
+      "loss": 0.4819,
+      "step": 13648
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.735524459819426e-06,
+      "loss": 0.5048,
+      "step": 13649
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73417110502173e-06,
+      "loss": 0.4718,
+      "step": 13650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73281779396876e-06,
+      "loss": 0.4688,
+      "step": 13651
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.731464526686647e-06,
+      "loss": 0.4712,
+      "step": 13652
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730111303201518e-06,
+      "loss": 0.4552,
+      "step": 13653
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.728758123539498e-06,
+      "loss": 0.4795,
+      "step": 13654
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72740498772671e-06,
+      "loss": 0.472,
+      "step": 13655
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72605189578928e-06,
+      "loss": 0.4686,
+      "step": 13656
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72469884775333e-06,
+      "loss": 0.4457,
+      "step": 13657
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.723345843644986e-06,
+      "loss": 0.499,
+      "step": 13658
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.721992883490365e-06,
+      "loss": 0.4559,
+      "step": 13659
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.720639967315588e-06,
+      "loss": 0.4452,
+      "step": 13660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719287095146777e-06,
+      "loss": 0.4442,
+      "step": 13661
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.717934267010051e-06,
+      "loss": 0.4555,
+      "step": 13662
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716581482931525e-06,
+      "loss": 0.4857,
+      "step": 13663
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71522874293732e-06,
+      "loss": 0.4582,
+      "step": 13664
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.713876047053548e-06,
+      "loss": 0.4569,
+      "step": 13665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71252339530633e-06,
+      "loss": 0.4566,
+      "step": 13666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.711170787721777e-06,
+      "loss": 0.4749,
+      "step": 13667
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.709818224326003e-06,
+      "loss": 0.4576,
+      "step": 13668
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708465705145123e-06,
+      "loss": 0.4501,
+      "step": 13669
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.707113230205247e-06,
+      "loss": 0.4772,
+      "step": 13670
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.705760799532485e-06,
+      "loss": 0.4559,
+      "step": 13671
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70440841315295e-06,
+      "loss": 0.4763,
+      "step": 13672
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703056071092755e-06,
+      "loss": 0.4493,
+      "step": 13673
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.701703773378e-06,
+      "loss": 0.4462,
+      "step": 13674
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700351520034798e-06,
+      "loss": 0.4664,
+      "step": 13675
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.698999311089256e-06,
+      "loss": 0.4733,
+      "step": 13676
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697647146567484e-06,
+      "loss": 0.4697,
+      "step": 13677
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696295026495578e-06,
+      "loss": 0.4689,
+      "step": 13678
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.694942950899647e-06,
+      "loss": 0.4646,
+      "step": 13679
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.693590919805796e-06,
+      "loss": 0.4531,
+      "step": 13680
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.692238933240127e-06,
+      "loss": 0.468,
+      "step": 13681
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69088699122874e-06,
+      "loss": 0.4832,
+      "step": 13682
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.689535093797737e-06,
+      "loss": 0.4696,
+      "step": 13683
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688183240973219e-06,
+      "loss": 0.4448,
+      "step": 13684
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.686831432781288e-06,
+      "loss": 0.4601,
+      "step": 13685
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685479669248035e-06,
+      "loss": 0.4621,
+      "step": 13686
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.684127950399563e-06,
+      "loss": 0.4726,
+      "step": 13687
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.682776276261969e-06,
+      "loss": 0.4453,
+      "step": 13688
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.681424646861343e-06,
+      "loss": 0.4634,
+      "step": 13689
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680073062223786e-06,
+      "loss": 0.4632,
+      "step": 13690
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67872152237539e-06,
+      "loss": 0.4462,
+      "step": 13691
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67737002734225e-06,
+      "loss": 0.488,
+      "step": 13692
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676018577150454e-06,
+      "loss": 0.461,
+      "step": 13693
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674667171826096e-06,
+      "loss": 0.4645,
+      "step": 13694
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673315811395265e-06,
+      "loss": 0.4787,
+      "step": 13695
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67196449588406e-06,
+      "loss": 0.4415,
+      "step": 13696
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.670613225318558e-06,
+      "loss": 0.4806,
+      "step": 13697
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669261999724849e-06,
+      "loss": 0.4583,
+      "step": 13698
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667910819129021e-06,
+      "loss": 0.4995,
+      "step": 13699
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.666559683557169e-06,
+      "loss": 0.4537,
+      "step": 13700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.665208593035367e-06,
+      "loss": 0.4562,
+      "step": 13701
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.663857547589704e-06,
+      "loss": 0.4692,
+      "step": 13702
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66250654724626e-06,
+      "loss": 0.4512,
+      "step": 13703
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661155592031128e-06,
+      "loss": 0.4723,
+      "step": 13704
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.659804681970378e-06,
+      "loss": 0.4495,
+      "step": 13705
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658453817090095e-06,
+      "loss": 0.4785,
+      "step": 13706
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.657102997416361e-06,
+      "loss": 0.4785,
+      "step": 13707
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.655752222975261e-06,
+      "loss": 0.4795,
+      "step": 13708
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.654401493792863e-06,
+      "loss": 0.4851,
+      "step": 13709
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.653050809895245e-06,
+      "loss": 0.4987,
+      "step": 13710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651700171308493e-06,
+      "loss": 0.4884,
+      "step": 13711
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.650349578058673e-06,
+      "loss": 0.4375,
+      "step": 13712
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.648999030171864e-06,
+      "loss": 0.4693,
+      "step": 13713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.647648527674139e-06,
+      "loss": 0.4509,
+      "step": 13714
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.646298070591578e-06,
+      "loss": 0.4639,
+      "step": 13715
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.644947658950244e-06,
+      "loss": 0.4647,
+      "step": 13716
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.643597292776207e-06,
+      "loss": 0.4594,
+      "step": 13717
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642246972095548e-06,
+      "loss": 0.4762,
+      "step": 13718
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.640896696934334e-06,
+      "loss": 0.469,
+      "step": 13719
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.639546467318629e-06,
+      "loss": 0.4896,
+      "step": 13720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.638196283274497e-06,
+      "loss": 0.4525,
+      "step": 13721
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.636846144828014e-06,
+      "loss": 0.453,
+      "step": 13722
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635496052005249e-06,
+      "loss": 0.4978,
+      "step": 13723
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.634146004832256e-06,
+      "loss": 0.4824,
+      "step": 13724
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632796003335104e-06,
+      "loss": 0.4868,
+      "step": 13725
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631446047539857e-06,
+      "loss": 0.466,
+      "step": 13726
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.630096137472584e-06,
+      "loss": 0.4542,
+      "step": 13727
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628746273159333e-06,
+      "loss": 0.4623,
+      "step": 13728
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.627396454626175e-06,
+      "loss": 0.4653,
+      "step": 13729
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626046681899171e-06,
+      "loss": 0.4887,
+      "step": 13730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.624696955004373e-06,
+      "loss": 0.4909,
+      "step": 13731
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.62334727396784e-06,
+      "loss": 0.463,
+      "step": 13732
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6219976388156344e-06,
+      "loss": 0.4572,
+      "step": 13733
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.620648049573815e-06,
+      "loss": 0.4569,
+      "step": 13734
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.619298506268427e-06,
+      "loss": 0.4497,
+      "step": 13735
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.617949008925531e-06,
+      "loss": 0.4525,
+      "step": 13736
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.616599557571182e-06,
+      "loss": 0.4647,
+      "step": 13737
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615250152231435e-06,
+      "loss": 0.4782,
+      "step": 13738
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.613900792932331e-06,
+      "loss": 0.4742,
+      "step": 13739
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.612551479699933e-06,
+      "loss": 0.4723,
+      "step": 13740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.611202212560284e-06,
+      "loss": 0.4603,
+      "step": 13741
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609852991539442e-06,
+      "loss": 0.4596,
+      "step": 13742
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.608503816663443e-06,
+      "loss": 0.4681,
+      "step": 13743
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.607154687958343e-06,
+      "loss": 0.4215,
+      "step": 13744
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.605805605450186e-06,
+      "loss": 0.4996,
+      "step": 13745
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.604456569165022e-06,
+      "loss": 0.4524,
+      "step": 13746
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.603107579128888e-06,
+      "loss": 0.4874,
+      "step": 13747
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.601758635367834e-06,
+      "loss": 0.4695,
+      "step": 13748
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.600409737907901e-06,
+      "loss": 0.4829,
+      "step": 13749
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599060886775134e-06,
+      "loss": 0.4513,
+      "step": 13750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597712081995569e-06,
+      "loss": 0.4507,
+      "step": 13751
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596363323595248e-06,
+      "loss": 0.482,
+      "step": 13752
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5950146116002184e-06,
+      "loss": 0.4727,
+      "step": 13753
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593665946036505e-06,
+      "loss": 0.4308,
+      "step": 13754
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5923173269301545e-06,
+      "loss": 0.4535,
+      "step": 13755
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590968754307202e-06,
+      "loss": 0.4894,
+      "step": 13756
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589620228193686e-06,
+      "loss": 0.4753,
+      "step": 13757
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5882717486156365e-06,
+      "loss": 0.4823,
+      "step": 13758
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5869233155990904e-06,
+      "loss": 0.4476,
+      "step": 13759
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58557492917008e-06,
+      "loss": 0.4675,
+      "step": 13760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58422658935464e-06,
+      "loss": 0.4481,
+      "step": 13761
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582878296178798e-06,
+      "loss": 0.4567,
+      "step": 13762
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.581530049668587e-06,
+      "loss": 0.4419,
+      "step": 13763
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.580181849850036e-06,
+      "loss": 0.4623,
+      "step": 13764
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.578833696749175e-06,
+      "loss": 0.4753,
+      "step": 13765
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.57748559039203e-06,
+      "loss": 0.4777,
+      "step": 13766
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.576137530804628e-06,
+      "loss": 0.4199,
+      "step": 13767
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5747895180129945e-06,
+      "loss": 0.4564,
+      "step": 13768
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.573441552043159e-06,
+      "loss": 0.4603,
+      "step": 13769
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.572093632921141e-06,
+      "loss": 0.4634,
+      "step": 13770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570745760672964e-06,
+      "loss": 0.4532,
+      "step": 13771
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.569397935324653e-06,
+      "loss": 0.4982,
+      "step": 13772
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5680501569022295e-06,
+      "loss": 0.4804,
+      "step": 13773
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56670242543171e-06,
+      "loss": 0.4559,
+      "step": 13774
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5653547409391175e-06,
+      "loss": 0.4497,
+      "step": 13775
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564007103450472e-06,
+      "loss": 0.4424,
+      "step": 13776
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.562659512991788e-06,
+      "loss": 0.4447,
+      "step": 13777
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561311969589084e-06,
+      "loss": 0.5099,
+      "step": 13778
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.559964473268375e-06,
+      "loss": 0.4712,
+      "step": 13779
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5586170240556796e-06,
+      "loss": 0.4744,
+      "step": 13780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.557269621977008e-06,
+      "loss": 0.4515,
+      "step": 13781
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555922267058375e-06,
+      "loss": 0.4847,
+      "step": 13782
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.554574959325793e-06,
+      "loss": 0.4828,
+      "step": 13783
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.553227698805277e-06,
+      "loss": 0.4962,
+      "step": 13784
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.551880485522831e-06,
+      "loss": 0.4513,
+      "step": 13785
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.550533319504468e-06,
+      "loss": 0.438,
+      "step": 13786
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.549186200776196e-06,
+      "loss": 0.417,
+      "step": 13787
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.547839129364028e-06,
+      "loss": 0.4555,
+      "step": 13788
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.546492105293962e-06,
+      "loss": 0.4648,
+      "step": 13789
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.477,
+      "step": 13790
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.543798199284175e-06,
+      "loss": 0.4538,
+      "step": 13791
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5424513173964645e-06,
+      "loss": 0.4763,
+      "step": 13792
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.541104482954875e-06,
+      "loss": 0.4528,
+      "step": 13793
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.539757695985415e-06,
+      "loss": 0.4979,
+      "step": 13794
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538410956514086e-06,
+      "loss": 0.4723,
+      "step": 13795
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.537064264566884e-06,
+      "loss": 0.4821,
+      "step": 13796
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.53571762016981e-06,
+      "loss": 0.462,
+      "step": 13797
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534371023348866e-06,
+      "loss": 0.4423,
+      "step": 13798
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.533024474130048e-06,
+      "loss": 0.4483,
+      "step": 13799
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5316779725393495e-06,
+      "loss": 0.4813,
+      "step": 13800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.530331518602771e-06,
+      "loss": 0.4673,
+      "step": 13801
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.528985112346306e-06,
+      "loss": 0.4512,
+      "step": 13802
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.527638753795949e-06,
+      "loss": 0.4409,
+      "step": 13803
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.526292442977692e-06,
+      "loss": 0.473,
+      "step": 13804
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.524946179917527e-06,
+      "loss": 0.4626,
+      "step": 13805
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523599964641446e-06,
+      "loss": 0.4555,
+      "step": 13806
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5222537971754425e-06,
+      "loss": 0.4678,
+      "step": 13807
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5209076775455e-06,
+      "loss": 0.467,
+      "step": 13808
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.519561605777611e-06,
+      "loss": 0.4551,
+      "step": 13809
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.518215581897763e-06,
+      "loss": 0.4565,
+      "step": 13810
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516869605931943e-06,
+      "loss": 0.473,
+      "step": 13811
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5155236779061335e-06,
+      "loss": 0.4474,
+      "step": 13812
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514177797846323e-06,
+      "loss": 0.4373,
+      "step": 13813
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.512831965778491e-06,
+      "loss": 0.47,
+      "step": 13814
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51148618172863e-06,
+      "loss": 0.454,
+      "step": 13815
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510140445722711e-06,
+      "loss": 0.4552,
+      "step": 13816
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.508794757786721e-06,
+      "loss": 0.5004,
+      "step": 13817
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.507449117946641e-06,
+      "loss": 0.4607,
+      "step": 13818
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506103526228445e-06,
+      "loss": 0.4649,
+      "step": 13819
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504757982658117e-06,
+      "loss": 0.4456,
+      "step": 13820
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503412487261631e-06,
+      "loss": 0.181,
+      "step": 13821
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.502067040064966e-06,
+      "loss": 0.4823,
+      "step": 13822
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500721641094095e-06,
+      "loss": 0.4271,
+      "step": 13823
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499376290374994e-06,
+      "loss": 0.4383,
+      "step": 13824
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.498030987933635e-06,
+      "loss": 0.4518,
+      "step": 13825
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.496685733795999e-06,
+      "loss": 0.494,
+      "step": 13826
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.495340527988046e-06,
+      "loss": 0.4676,
+      "step": 13827
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.493995370535754e-06,
+      "loss": 0.4614,
+      "step": 13828
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4926502614650865e-06,
+      "loss": 0.4669,
+      "step": 13829
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.491305200802026e-06,
+      "loss": 0.4721,
+      "step": 13830
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.489960188572526e-06,
+      "loss": 0.4902,
+      "step": 13831
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.48861522480256e-06,
+      "loss": 0.46,
+      "step": 13832
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487270309518092e-06,
+      "loss": 0.464,
+      "step": 13833
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4859254427450954e-06,
+      "loss": 0.4618,
+      "step": 13834
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484580624509525e-06,
+      "loss": 0.4732,
+      "step": 13835
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.483235854837344e-06,
+      "loss": 0.4697,
+      "step": 13836
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.481891133754526e-06,
+      "loss": 0.4717,
+      "step": 13837
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.480546461287019e-06,
+      "loss": 0.4445,
+      "step": 13838
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.479201837460792e-06,
+      "loss": 0.4615,
+      "step": 13839
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.477857262301799e-06,
+      "loss": 0.4634,
+      "step": 13840
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.476512735836008e-06,
+      "loss": 0.4655,
+      "step": 13841
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.475168258089367e-06,
+      "loss": 0.4593,
+      "step": 13842
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.473823829087835e-06,
+      "loss": 0.4532,
+      "step": 13843
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.472479448857372e-06,
+      "loss": 0.4868,
+      "step": 13844
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471135117423935e-06,
+      "loss": 0.4854,
+      "step": 13845
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469790834813468e-06,
+      "loss": 0.4782,
+      "step": 13846
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4684466010519285e-06,
+      "loss": 0.4795,
+      "step": 13847
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4671024161652726e-06,
+      "loss": 0.495,
+      "step": 13848
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.465758280179452e-06,
+      "loss": 0.4764,
+      "step": 13849
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.464414193120411e-06,
+      "loss": 0.4514,
+      "step": 13850
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463070155014098e-06,
+      "loss": 0.4652,
+      "step": 13851
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461726165886469e-06,
+      "loss": 0.4594,
+      "step": 13852
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.460382225763471e-06,
+      "loss": 0.4427,
+      "step": 13853
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.45903833467104e-06,
+      "loss": 0.4763,
+      "step": 13854
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457694492635132e-06,
+      "loss": 0.1651,
+      "step": 13855
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456350699681687e-06,
+      "loss": 0.436,
+      "step": 13856
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.455006955836655e-06,
+      "loss": 0.4709,
+      "step": 13857
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453663261125967e-06,
+      "loss": 0.4671,
+      "step": 13858
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.452319615575573e-06,
+      "loss": 0.1776,
+      "step": 13859
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.450976019211416e-06,
+      "loss": 0.4775,
+      "step": 13860
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4496324720594285e-06,
+      "loss": 0.4833,
+      "step": 13861
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.44828897414555e-06,
+      "loss": 0.4651,
+      "step": 13862
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.446945525495723e-06,
+      "loss": 0.4839,
+      "step": 13863
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445602126135889e-06,
+      "loss": 0.4811,
+      "step": 13864
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4442587760919696e-06,
+      "loss": 0.4521,
+      "step": 13865
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442915475389911e-06,
+      "loss": 0.4757,
+      "step": 13866
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.441572224055644e-06,
+      "loss": 0.4358,
+      "step": 13867
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.440229022115107e-06,
+      "loss": 0.4724,
+      "step": 13868
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.43888586959422e-06,
+      "loss": 0.4584,
+      "step": 13869
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4375427665189244e-06,
+      "loss": 0.4632,
+      "step": 13870
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.436199712915149e-06,
+      "loss": 0.4838,
+      "step": 13871
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.434856708808822e-06,
+      "loss": 0.468,
+      "step": 13872
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433513754225872e-06,
+      "loss": 0.4748,
+      "step": 13873
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.432170849192224e-06,
+      "loss": 0.4502,
+      "step": 13874
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.430827993733809e-06,
+      "loss": 0.4911,
+      "step": 13875
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.429485187876551e-06,
+      "loss": 0.4756,
+      "step": 13876
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.428142431646372e-06,
+      "loss": 0.4629,
+      "step": 13877
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426799725069198e-06,
+      "loss": 0.4762,
+      "step": 13878
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.425457068170956e-06,
+      "loss": 0.4513,
+      "step": 13879
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424114460977555e-06,
+      "loss": 0.4444,
+      "step": 13880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.422771903514927e-06,
+      "loss": 0.4578,
+      "step": 13881
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.421429395808987e-06,
+      "loss": 0.4408,
+      "step": 13882
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42008693788566e-06,
+      "loss": 0.4926,
+      "step": 13883
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.418744529770856e-06,
+      "loss": 0.4718,
+      "step": 13884
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.417402171490495e-06,
+      "loss": 0.4816,
+      "step": 13885
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416059863070492e-06,
+      "loss": 0.4756,
+      "step": 13886
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.414717604536765e-06,
+      "loss": 0.4465,
+      "step": 13887
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.413375395915226e-06,
+      "loss": 0.4856,
+      "step": 13888
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.412033237231786e-06,
+      "loss": 0.4466,
+      "step": 13889
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410691128512362e-06,
+      "loss": 0.4834,
+      "step": 13890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.409349069782862e-06,
+      "loss": 0.4662,
+      "step": 13891
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.408007061069194e-06,
+      "loss": 0.1638,
+      "step": 13892
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.406665102397272e-06,
+      "loss": 0.4577,
+      "step": 13893
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4053231937930005e-06,
+      "loss": 0.4662,
+      "step": 13894
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40398133528229e-06,
+      "loss": 0.4542,
+      "step": 13895
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.402639526891044e-06,
+      "loss": 0.4542,
+      "step": 13896
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.401297768645168e-06,
+      "loss": 0.4627,
+      "step": 13897
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.399956060570567e-06,
+      "loss": 0.4658,
+      "step": 13898
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3986144026931475e-06,
+      "loss": 0.463,
+      "step": 13899
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.397272795038806e-06,
+      "loss": 0.4426,
+      "step": 13900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.395931237633447e-06,
+      "loss": 0.4658,
+      "step": 13901
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3945897305029725e-06,
+      "loss": 0.4654,
+      "step": 13902
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.393248273673279e-06,
+      "loss": 0.4651,
+      "step": 13903
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.391906867170266e-06,
+      "loss": 0.4652,
+      "step": 13904
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.390565511019834e-06,
+      "loss": 0.4698,
+      "step": 13905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3892242052478765e-06,
+      "loss": 0.445,
+      "step": 13906
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.38788294988029e-06,
+      "loss": 0.4719,
+      "step": 13907
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3865417449429655e-06,
+      "loss": 0.4743,
+      "step": 13908
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385200590461803e-06,
+      "loss": 0.4451,
+      "step": 13909
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383859486462693e-06,
+      "loss": 0.4831,
+      "step": 13910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3825184329715265e-06,
+      "loss": 0.4557,
+      "step": 13911
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3811774300141914e-06,
+      "loss": 0.4582,
+      "step": 13912
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.379836477616582e-06,
+      "loss": 0.4542,
+      "step": 13913
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.378495575804588e-06,
+      "loss": 0.4524,
+      "step": 13914
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.377154724604091e-06,
+      "loss": 0.451,
+      "step": 13915
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.375813924040982e-06,
+      "loss": 0.4593,
+      "step": 13916
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.374473174141148e-06,
+      "loss": 0.4605,
+      "step": 13917
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.373132474930473e-06,
+      "loss": 0.4651,
+      "step": 13918
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.371791826434838e-06,
+      "loss": 0.1636,
+      "step": 13919
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.370451228680129e-06,
+      "loss": 0.4717,
+      "step": 13920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.369110681692226e-06,
+      "loss": 0.4674,
+      "step": 13921
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367770185497014e-06,
+      "loss": 0.4314,
+      "step": 13922
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.366429740120369e-06,
+      "loss": 0.46,
+      "step": 13923
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.36508934558817e-06,
+      "loss": 0.4575,
+      "step": 13924
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.363749001926298e-06,
+      "loss": 0.4355,
+      "step": 13925
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.362408709160626e-06,
+      "loss": 0.4398,
+      "step": 13926
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.361068467317034e-06,
+      "loss": 0.4974,
+      "step": 13927
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.359728276421394e-06,
+      "loss": 0.4598,
+      "step": 13928
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358388136499583e-06,
+      "loss": 0.4875,
+      "step": 13929
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.357048047577471e-06,
+      "loss": 0.4847,
+      "step": 13930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355708009680932e-06,
+      "loss": 0.4774,
+      "step": 13931
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354368022835836e-06,
+      "loss": 0.4707,
+      "step": 13932
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.353028087068057e-06,
+      "loss": 0.4826,
+      "step": 13933
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.35168820240346e-06,
+      "loss": 0.471,
+      "step": 13934
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.350348368867912e-06,
+      "loss": 0.4595,
+      "step": 13935
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.349008586487285e-06,
+      "loss": 0.4417,
+      "step": 13936
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.347668855287444e-06,
+      "loss": 0.4558,
+      "step": 13937
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34632917529425e-06,
+      "loss": 0.4465,
+      "step": 13938
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.344989546533572e-06,
+      "loss": 0.4781,
+      "step": 13939
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343649969031268e-06,
+      "loss": 0.4758,
+      "step": 13940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34231044281321e-06,
+      "loss": 0.4759,
+      "step": 13941
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34097096790525e-06,
+      "loss": 0.4468,
+      "step": 13942
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.33963154433325e-06,
+      "loss": 0.4502,
+      "step": 13943
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3382921721230736e-06,
+      "loss": 0.4588,
+      "step": 13944
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.336952851300574e-06,
+      "loss": 0.4462,
+      "step": 13945
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3356135818916095e-06,
+      "loss": 0.4617,
+      "step": 13946
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.334274363922039e-06,
+      "loss": 0.4561,
+      "step": 13947
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.332935197417719e-06,
+      "loss": 0.1677,
+      "step": 13948
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.331596082404497e-06,
+      "loss": 0.4494,
+      "step": 13949
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.330257018908231e-06,
+      "loss": 0.4487,
+      "step": 13950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.328918006954772e-06,
+      "loss": 0.4735,
+      "step": 13951
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.327579046569977e-06,
+      "loss": 0.4564,
+      "step": 13952
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3262401377796875e-06,
+      "loss": 0.4832,
+      "step": 13953
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.324901280609757e-06,
+      "loss": 0.4499,
+      "step": 13954
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323562475086033e-06,
+      "loss": 0.4476,
+      "step": 13955
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.32222372123437e-06,
+      "loss": 0.4667,
+      "step": 13956
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.320885019080602e-06,
+      "loss": 0.4509,
+      "step": 13957
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.319546368650581e-06,
+      "loss": 0.4913,
+      "step": 13958
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3182077699701485e-06,
+      "loss": 0.4385,
+      "step": 13959
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.316869223065156e-06,
+      "loss": 0.4579,
+      "step": 13960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315530727961435e-06,
+      "loss": 0.4403,
+      "step": 13961
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.314192284684831e-06,
+      "loss": 0.4506,
+      "step": 13962
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.312853893261187e-06,
+      "loss": 0.4465,
+      "step": 13963
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3115155537163444e-06,
+      "loss": 0.4883,
+      "step": 13964
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310177266076134e-06,
+      "loss": 0.4851,
+      "step": 13965
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.308839030366395e-06,
+      "loss": 0.4809,
+      "step": 13966
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3075008466129704e-06,
+      "loss": 0.4618,
+      "step": 13967
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.306162714841687e-06,
+      "loss": 0.4759,
+      "step": 13968
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.304824635078382e-06,
+      "loss": 0.4547,
+      "step": 13969
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3034866073488885e-06,
+      "loss": 0.4778,
+      "step": 13970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.302148631679047e-06,
+      "loss": 0.4894,
+      "step": 13971
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.300810708094677e-06,
+      "loss": 0.4503,
+      "step": 13972
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.299472836621611e-06,
+      "loss": 0.4747,
+      "step": 13973
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.298135017285684e-06,
+      "loss": 0.4915,
+      "step": 13974
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.296797250112724e-06,
+      "loss": 0.4679,
+      "step": 13975
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295459535128552e-06,
+      "loss": 0.4846,
+      "step": 13976
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2941218723589945e-06,
+      "loss": 0.4716,
+      "step": 13977
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.292784261829884e-06,
+      "loss": 0.4794,
+      "step": 13978
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.291446703567044e-06,
+      "loss": 0.4844,
+      "step": 13979
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.290109197596289e-06,
+      "loss": 0.446,
+      "step": 13980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.28877174394345e-06,
+      "loss": 0.4576,
+      "step": 13981
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.287434342634343e-06,
+      "loss": 0.4572,
+      "step": 13982
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2860969936947966e-06,
+      "loss": 0.4428,
+      "step": 13983
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.284759697150616e-06,
+      "loss": 0.4397,
+      "step": 13984
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.283422453027631e-06,
+      "loss": 0.4722,
+      "step": 13985
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.282085261351659e-06,
+      "loss": 0.4791,
+      "step": 13986
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280748122148506e-06,
+      "loss": 0.4396,
+      "step": 13987
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.279411035443994e-06,
+      "loss": 0.4702,
+      "step": 13988
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.278074001263937e-06,
+      "loss": 0.4869,
+      "step": 13989
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276737019634155e-06,
+      "loss": 0.4767,
+      "step": 13990
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.275400090580444e-06,
+      "loss": 0.4211,
+      "step": 13991
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.274063214128626e-06,
+      "loss": 0.4754,
+      "step": 13992
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.27272639030451e-06,
+      "loss": 0.4478,
+      "step": 13993
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.271389619133908e-06,
+      "loss": 0.4794,
+      "step": 13994
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.270052900642619e-06,
+      "loss": 0.5156,
+      "step": 13995
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268716234856457e-06,
+      "loss": 0.4599,
+      "step": 13996
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267379621801227e-06,
+      "loss": 0.4914,
+      "step": 13997
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.266043061502737e-06,
+      "loss": 0.4658,
+      "step": 13998
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.264706553986782e-06,
+      "loss": 0.4673,
+      "step": 13999
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 0.4541,
+      "step": 14000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2620336974057074e-06,
+      "loss": 0.4384,
+      "step": 14001
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.260697348392191e-06,
+      "loss": 0.4504,
+      "step": 14002
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.25936105226442e-06,
+      "loss": 0.4664,
+      "step": 14003
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.258024809048194e-06,
+      "loss": 0.4961,
+      "step": 14004
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256688618769311e-06,
+      "loss": 0.4521,
+      "step": 14005
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.25535248145357e-06,
+      "loss": 0.4761,
+      "step": 14006
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2540163971267614e-06,
+      "loss": 0.4625,
+      "step": 14007
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.252680365814686e-06,
+      "loss": 0.4418,
+      "step": 14008
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.251344387543136e-06,
+      "loss": 0.4749,
+      "step": 14009
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2500084623379005e-06,
+      "loss": 0.4818,
+      "step": 14010
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.248672590224776e-06,
+      "loss": 0.454,
+      "step": 14011
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.24733677122955e-06,
+      "loss": 0.4864,
+      "step": 14012
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.246001005378015e-06,
+      "loss": 0.4697,
+      "step": 14013
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.244665292695956e-06,
+      "loss": 0.4507,
+      "step": 14014
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243329633209164e-06,
+      "loss": 0.485,
+      "step": 14015
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.241994026943424e-06,
+      "loss": 0.4266,
+      "step": 14016
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.240658473924523e-06,
+      "loss": 0.5097,
+      "step": 14017
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.239322974178244e-06,
+      "loss": 0.4447,
+      "step": 14018
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23798752773037e-06,
+      "loss": 0.4609,
+      "step": 14019
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.236652134606684e-06,
+      "loss": 0.4777,
+      "step": 14020
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.235316794832972e-06,
+      "loss": 0.4587,
+      "step": 14021
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.233981508435006e-06,
+      "loss": 0.4435,
+      "step": 14022
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.232646275438571e-06,
+      "loss": 0.4532,
+      "step": 14023
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.231311095869444e-06,
+      "loss": 0.4751,
+      "step": 14024
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229975969753406e-06,
+      "loss": 0.4757,
+      "step": 14025
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.228640897116226e-06,
+      "loss": 0.4298,
+      "step": 14026
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.227305877983685e-06,
+      "loss": 0.4609,
+      "step": 14027
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.225970912381557e-06,
+      "loss": 0.4835,
+      "step": 14028
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.22463600033561e-06,
+      "loss": 0.4707,
+      "step": 14029
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.223301141871621e-06,
+      "loss": 0.4814,
+      "step": 14030
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.22196633701536e-06,
+      "loss": 0.4741,
+      "step": 14031
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2206315857926006e-06,
+      "loss": 0.4666,
+      "step": 14032
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.219296888229105e-06,
+      "loss": 0.1623,
+      "step": 14033
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2179622443506446e-06,
+      "loss": 0.4297,
+      "step": 14034
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2166276541829864e-06,
+      "loss": 0.4628,
+      "step": 14035
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215293117751899e-06,
+      "loss": 0.4544,
+      "step": 14036
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213958635083143e-06,
+      "loss": 0.4774,
+      "step": 14037
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.212624206202483e-06,
+      "loss": 0.4607,
+      "step": 14038
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.211289831135684e-06,
+      "loss": 0.4551,
+      "step": 14039
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209955509908508e-06,
+      "loss": 0.4553,
+      "step": 14040
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.208621242546714e-06,
+      "loss": 0.4536,
+      "step": 14041
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.207287029076062e-06,
+      "loss": 0.4855,
+      "step": 14042
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20595286952231e-06,
+      "loss": 0.4882,
+      "step": 14043
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2046187639112186e-06,
+      "loss": 0.4755,
+      "step": 14044
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.203284712268541e-06,
+      "loss": 0.4865,
+      "step": 14045
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2019507146200344e-06,
+      "loss": 0.4657,
+      "step": 14046
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2006167709914535e-06,
+      "loss": 0.4792,
+      "step": 14047
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.199282881408551e-06,
+      "loss": 0.4404,
+      "step": 14048
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197949045897082e-06,
+      "loss": 0.4647,
+      "step": 14049
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.196615264482791e-06,
+      "loss": 0.4447,
+      "step": 14050
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1952815371914386e-06,
+      "loss": 0.4397,
+      "step": 14051
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193947864048764e-06,
+      "loss": 0.4787,
+      "step": 14052
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192614245080523e-06,
+      "loss": 0.4639,
+      "step": 14053
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.191280680312458e-06,
+      "loss": 0.1848,
+      "step": 14054
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.18994716977032e-06,
+      "loss": 0.4543,
+      "step": 14055
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1886137134798486e-06,
+      "loss": 0.4795,
+      "step": 14056
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1872803114667914e-06,
+      "loss": 0.4622,
+      "step": 14057
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1859469637568895e-06,
+      "loss": 0.4349,
+      "step": 14058
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.184613670375888e-06,
+      "loss": 0.4726,
+      "step": 14059
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183280431349524e-06,
+      "loss": 0.4582,
+      "step": 14060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.181947246703538e-06,
+      "loss": 0.4679,
+      "step": 14061
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180614116463671e-06,
+      "loss": 0.428,
+      "step": 14062
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179281040655661e-06,
+      "loss": 0.4572,
+      "step": 14063
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.177948019305243e-06,
+      "loss": 0.4687,
+      "step": 14064
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1766150524381515e-06,
+      "loss": 0.4831,
+      "step": 14065
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.175282140080124e-06,
+      "loss": 0.4623,
+      "step": 14066
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173949282256895e-06,
+      "loss": 0.4956,
+      "step": 14067
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.172616478994192e-06,
+      "loss": 0.4454,
+      "step": 14068
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1712837303177495e-06,
+      "loss": 0.4633,
+      "step": 14069
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169951036253296e-06,
+      "loss": 0.4807,
+      "step": 14070
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.168618396826569e-06,
+      "loss": 0.4719,
+      "step": 14071
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167285812063286e-06,
+      "loss": 0.4654,
+      "step": 14072
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1659532819891785e-06,
+      "loss": 0.4384,
+      "step": 14073
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.164620806629976e-06,
+      "loss": 0.4542,
+      "step": 14074
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163288386011397e-06,
+      "loss": 0.4346,
+      "step": 14075
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1619560201591705e-06,
+      "loss": 0.4835,
+      "step": 14076
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160623709099015e-06,
+      "loss": 0.4612,
+      "step": 14077
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.159291452856663e-06,
+      "loss": 0.481,
+      "step": 14078
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1579592514578234e-06,
+      "loss": 0.4604,
+      "step": 14079
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.15662710492822e-06,
+      "loss": 0.4635,
+      "step": 14080
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.155295013293572e-06,
+      "loss": 0.4546,
+      "step": 14081
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153962976579603e-06,
+      "loss": 0.4657,
+      "step": 14082
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.152630994812019e-06,
+      "loss": 0.4723,
+      "step": 14083
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151299068016542e-06,
+      "loss": 0.4446,
+      "step": 14084
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149967196218882e-06,
+      "loss": 0.4744,
+      "step": 14085
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.148635379444763e-06,
+      "loss": 0.464,
+      "step": 14086
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.147303617719885e-06,
+      "loss": 0.461,
+      "step": 14087
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145971911069963e-06,
+      "loss": 0.5006,
+      "step": 14088
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1446402595207124e-06,
+      "loss": 0.4676,
+      "step": 14089
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143308663097841e-06,
+      "loss": 0.468,
+      "step": 14090
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.141977121827051e-06,
+      "loss": 0.4803,
+      "step": 14091
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.140645635734053e-06,
+      "loss": 0.4907,
+      "step": 14092
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.139314204844559e-06,
+      "loss": 0.4595,
+      "step": 14093
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.137982829184263e-06,
+      "loss": 0.4593,
+      "step": 14094
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.136651508778876e-06,
+      "loss": 0.4653,
+      "step": 14095
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.135320243654096e-06,
+      "loss": 0.466,
+      "step": 14096
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1339890338356346e-06,
+      "loss": 0.4876,
+      "step": 14097
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1326578793491806e-06,
+      "loss": 0.4537,
+      "step": 14098
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.131326780220439e-06,
+      "loss": 0.447,
+      "step": 14099
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1299957364751085e-06,
+      "loss": 0.4785,
+      "step": 14100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128664748138892e-06,
+      "loss": 0.4699,
+      "step": 14101
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.127333815237476e-06,
+      "loss": 0.4929,
+      "step": 14102
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1260029377965566e-06,
+      "loss": 0.4639,
+      "step": 14103
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.124672115841835e-06,
+      "loss": 0.4422,
+      "step": 14104
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123341349399006e-06,
+      "loss": 0.4526,
+      "step": 14105
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.122010638493751e-06,
+      "loss": 0.4776,
+      "step": 14106
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.120679983151766e-06,
+      "loss": 0.4792,
+      "step": 14107
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119349383398744e-06,
+      "loss": 0.4784,
+      "step": 14108
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.118018839260376e-06,
+      "loss": 0.4744,
+      "step": 14109
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116688350762338e-06,
+      "loss": 0.456,
+      "step": 14110
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.115357917930328e-06,
+      "loss": 0.1715,
+      "step": 14111
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.11402754079003e-06,
+      "loss": 0.4719,
+      "step": 14112
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1126972193671285e-06,
+      "loss": 0.4768,
+      "step": 14113
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1113669536872996e-06,
+      "loss": 0.4918,
+      "step": 14114
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110036743776234e-06,
+      "loss": 0.4475,
+      "step": 14115
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1087065896596145e-06,
+      "loss": 0.4876,
+      "step": 14116
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1073764913631115e-06,
+      "loss": 0.4801,
+      "step": 14117
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.106046448912412e-06,
+      "loss": 0.4748,
+      "step": 14118
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1047164623331945e-06,
+      "loss": 0.4817,
+      "step": 14119
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.103386531651137e-06,
+      "loss": 0.4657,
+      "step": 14120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102056656891907e-06,
+      "loss": 0.4532,
+      "step": 14121
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100726838081185e-06,
+      "loss": 0.4689,
+      "step": 14122
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0993970752446475e-06,
+      "loss": 0.4697,
+      "step": 14123
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.098067368407968e-06,
+      "loss": 0.4698,
+      "step": 14124
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.096737717596809e-06,
+      "loss": 0.4562,
+      "step": 14125
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0954081228368485e-06,
+      "loss": 0.4588,
+      "step": 14126
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.094078584153757e-06,
+      "loss": 0.4458,
+      "step": 14127
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0927491015732e-06,
+      "loss": 0.4669,
+      "step": 14128
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091419675120843e-06,
+      "loss": 0.4789,
+      "step": 14129
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.090090304822356e-06,
+      "loss": 0.4739,
+      "step": 14130
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.088760990703402e-06,
+      "loss": 0.4675,
+      "step": 14131
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0874317327896494e-06,
+      "loss": 0.4555,
+      "step": 14132
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.086102531106755e-06,
+      "loss": 0.4502,
+      "step": 14133
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.084773385680383e-06,
+      "loss": 0.4728,
+      "step": 14134
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083444296536199e-06,
+      "loss": 0.4554,
+      "step": 14135
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.082115263699854e-06,
+      "loss": 0.4763,
+      "step": 14136
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.080786287197011e-06,
+      "loss": 0.4694,
+      "step": 14137
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079457367053328e-06,
+      "loss": 0.4494,
+      "step": 14138
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.078128503294466e-06,
+      "loss": 0.4541,
+      "step": 14139
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0767996959460714e-06,
+      "loss": 0.4946,
+      "step": 14140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0754709450338045e-06,
+      "loss": 0.4716,
+      "step": 14141
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0741422505833155e-06,
+      "loss": 0.4628,
+      "step": 14142
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0728136126202595e-06,
+      "loss": 0.4754,
+      "step": 14143
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071485031170286e-06,
+      "loss": 0.4571,
+      "step": 14144
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.070156506259044e-06,
+      "loss": 0.4761,
+      "step": 14145
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.068828037912183e-06,
+      "loss": 0.447,
+      "step": 14146
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.067499626155354e-06,
+      "loss": 0.4548,
+      "step": 14147
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0661712710141995e-06,
+      "loss": 0.4613,
+      "step": 14148
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.064842972514365e-06,
+      "loss": 0.4762,
+      "step": 14149
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0635147306814965e-06,
+      "loss": 0.4599,
+      "step": 14150
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.062186545541239e-06,
+      "loss": 0.4669,
+      "step": 14151
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.060858417119233e-06,
+      "loss": 0.4532,
+      "step": 14152
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.059530345441119e-06,
+      "loss": 0.481,
+      "step": 14153
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0582023305325375e-06,
+      "loss": 0.4795,
+      "step": 14154
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056874372419131e-06,
+      "loss": 0.4577,
+      "step": 14155
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.055546471126531e-06,
+      "loss": 0.4328,
+      "step": 14156
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.054218626680379e-06,
+      "loss": 0.4591,
+      "step": 14157
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.052890839106312e-06,
+      "loss": 0.4703,
+      "step": 14158
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05156310842996e-06,
+      "loss": 0.4563,
+      "step": 14159
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0502354346769575e-06,
+      "loss": 0.4878,
+      "step": 14160
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.048907817872939e-06,
+      "loss": 0.4311,
+      "step": 14161
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.047580258043537e-06,
+      "loss": 0.531,
+      "step": 14162
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046252755214378e-06,
+      "loss": 0.4413,
+      "step": 14163
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0449253094110925e-06,
+      "loss": 0.4531,
+      "step": 14164
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.043597920659308e-06,
+      "loss": 0.4835,
+      "step": 14165
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.042270588984656e-06,
+      "loss": 0.4671,
+      "step": 14166
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.040943314412758e-06,
+      "loss": 0.4677,
+      "step": 14167
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.039616096969237e-06,
+      "loss": 0.4776,
+      "step": 14168
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.03828893667972e-06,
+      "loss": 0.4766,
+      "step": 14169
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036961833569831e-06,
+      "loss": 0.4829,
+      "step": 14170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.035634787665186e-06,
+      "loss": 0.4591,
+      "step": 14171
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.034307798991409e-06,
+      "loss": 0.4867,
+      "step": 14172
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0329808675741185e-06,
+      "loss": 0.4644,
+      "step": 14173
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.031653993438935e-06,
+      "loss": 0.4684,
+      "step": 14174
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.030327176611471e-06,
+      "loss": 0.4124,
+      "step": 14175
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.029000417117345e-06,
+      "loss": 0.4474,
+      "step": 14176
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.027673714982173e-06,
+      "loss": 0.4527,
+      "step": 14177
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026347070231565e-06,
+      "loss": 0.4547,
+      "step": 14178
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.025020482891136e-06,
+      "loss": 0.4574,
+      "step": 14179
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.023693952986496e-06,
+      "loss": 0.4711,
+      "step": 14180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0223674805432595e-06,
+      "loss": 0.4743,
+      "step": 14181
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.021041065587031e-06,
+      "loss": 0.4904,
+      "step": 14182
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01971470814342e-06,
+      "loss": 0.4613,
+      "step": 14183
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.018388408238034e-06,
+      "loss": 0.4523,
+      "step": 14184
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01706216589648e-06,
+      "loss": 0.4535,
+      "step": 14185
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01573598114436e-06,
+      "loss": 0.4339,
+      "step": 14186
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01440985400728e-06,
+      "loss": 0.4538,
+      "step": 14187
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0130837845108425e-06,
+      "loss": 0.4767,
+      "step": 14188
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0117577726806495e-06,
+      "loss": 0.4517,
+      "step": 14189
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.010431818542298e-06,
+      "loss": 0.439,
+      "step": 14190
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.00910592212139e-06,
+      "loss": 0.4822,
+      "step": 14191
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0077800834435225e-06,
+      "loss": 0.4606,
+      "step": 14192
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.006454302534296e-06,
+      "loss": 0.4946,
+      "step": 14193
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0051285794193004e-06,
+      "loss": 0.4665,
+      "step": 14194
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0038029141241335e-06,
+      "loss": 0.4952,
+      "step": 14195
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002477306674389e-06,
+      "loss": 0.4776,
+      "step": 14196
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.001151757095663e-06,
+      "loss": 0.4544,
+      "step": 14197
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.999826265413542e-06,
+      "loss": 0.4833,
+      "step": 14198
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9985008316536165e-06,
+      "loss": 0.4677,
+      "step": 14199
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.997175455841478e-06,
+      "loss": 0.4699,
+      "step": 14200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9958501380027134e-06,
+      "loss": 0.4732,
+      "step": 14201
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.994524878162909e-06,
+      "loss": 0.1804,
+      "step": 14202
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993199676347651e-06,
+      "loss": 0.4738,
+      "step": 14203
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.991874532582528e-06,
+      "loss": 0.454,
+      "step": 14204
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.990549446893116e-06,
+      "loss": 0.4557,
+      "step": 14205
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989224419305003e-06,
+      "loss": 0.4775,
+      "step": 14206
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.987899449843767e-06,
+      "loss": 0.4783,
+      "step": 14207
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.986574538534996e-06,
+      "loss": 0.4923,
+      "step": 14208
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9852496854042585e-06,
+      "loss": 0.4586,
+      "step": 14209
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983924890477138e-06,
+      "loss": 0.462,
+      "step": 14210
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.982600153779208e-06,
+      "loss": 0.4506,
+      "step": 14211
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.981275475336056e-06,
+      "loss": 0.456,
+      "step": 14212
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.97995085517324e-06,
+      "loss": 0.4616,
+      "step": 14213
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978626293316342e-06,
+      "loss": 0.4742,
+      "step": 14214
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.977301789790931e-06,
+      "loss": 0.489,
+      "step": 14215
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9759773446225885e-06,
+      "loss": 0.45,
+      "step": 14216
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974652957836872e-06,
+      "loss": 0.4362,
+      "step": 14217
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973328629459351e-06,
+      "loss": 0.4721,
+      "step": 14218
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.972004359515601e-06,
+      "loss": 0.438,
+      "step": 14219
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970680148031191e-06,
+      "loss": 0.416,
+      "step": 14220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.969355995031675e-06,
+      "loss": 0.4471,
+      "step": 14221
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.968031900542622e-06,
+      "loss": 0.4499,
+      "step": 14222
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966707864589602e-06,
+      "loss": 0.4559,
+      "step": 14223
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.965383887198169e-06,
+      "loss": 0.4572,
+      "step": 14224
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9640599683938835e-06,
+      "loss": 0.4807,
+      "step": 14225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9627361082023125e-06,
+      "loss": 0.5094,
+      "step": 14226
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.961412306649015e-06,
+      "loss": 0.4587,
+      "step": 14227
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.96008856375954e-06,
+      "loss": 0.4791,
+      "step": 14228
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958764879559448e-06,
+      "loss": 0.468,
+      "step": 14229
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957441254074297e-06,
+      "loss": 0.5015,
+      "step": 14230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9561176873296446e-06,
+      "loss": 0.4367,
+      "step": 14231
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.954794179351033e-06,
+      "loss": 0.4628,
+      "step": 14232
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.95347073016402e-06,
+      "loss": 0.4691,
+      "step": 14233
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9521473397941575e-06,
+      "loss": 0.4496,
+      "step": 14234
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9508240082669995e-06,
+      "loss": 0.4413,
+      "step": 14235
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.949500735608083e-06,
+      "loss": 0.4631,
+      "step": 14236
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9481775218429646e-06,
+      "loss": 0.459,
+      "step": 14237
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9468543669971875e-06,
+      "loss": 0.4635,
+      "step": 14238
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.945531271096302e-06,
+      "loss": 0.467,
+      "step": 14239
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944208234165843e-06,
+      "loss": 0.4633,
+      "step": 14240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.942885256231358e-06,
+      "loss": 0.4564,
+      "step": 14241
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9415623373183936e-06,
+      "loss": 0.4434,
+      "step": 14242
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940239477452482e-06,
+      "loss": 0.4578,
+      "step": 14243
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.938916676659164e-06,
+      "loss": 0.4587,
+      "step": 14244
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9375939349639825e-06,
+      "loss": 0.4611,
+      "step": 14245
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.936271252392477e-06,
+      "loss": 0.4513,
+      "step": 14246
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.934948628970173e-06,
+      "loss": 0.466,
+      "step": 14247
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933626064722613e-06,
+      "loss": 0.4476,
+      "step": 14248
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.93230355967533e-06,
+      "loss": 0.4582,
+      "step": 14249
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930981113853861e-06,
+      "loss": 0.4668,
+      "step": 14250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9296587272837245e-06,
+      "loss": 0.4887,
+      "step": 14251
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928336399990462e-06,
+      "loss": 0.4557,
+      "step": 14252
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9270141319996e-06,
+      "loss": 0.4388,
+      "step": 14253
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9256919233366686e-06,
+      "loss": 0.4716,
+      "step": 14254
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.924369774027189e-06,
+      "loss": 0.4895,
+      "step": 14255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.923047684096691e-06,
+      "loss": 0.4518,
+      "step": 14256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9217256535706985e-06,
+      "loss": 0.4454,
+      "step": 14257
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9204036824747385e-06,
+      "loss": 0.4361,
+      "step": 14258
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919081770834326e-06,
+      "loss": 0.4645,
+      "step": 14259
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.917759918674988e-06,
+      "loss": 0.4624,
+      "step": 14260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916438126022242e-06,
+      "loss": 0.4935,
+      "step": 14261
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.915116392901609e-06,
+      "loss": 0.4428,
+      "step": 14262
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.913794719338605e-06,
+      "loss": 0.4787,
+      "step": 14263
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.912473105358746e-06,
+      "loss": 0.4532,
+      "step": 14264
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.91115155098755e-06,
+      "loss": 0.4697,
+      "step": 14265
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.4852,
+      "step": 14266
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9085086211731935e-06,
+      "loss": 0.4714,
+      "step": 14267
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.907187245781062e-06,
+      "loss": 0.4345,
+      "step": 14268
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9058659300996425e-06,
+      "loss": 0.4904,
+      "step": 14269
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.904544674154443e-06,
+      "loss": 0.172,
+      "step": 14270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.903223477970974e-06,
+      "loss": 0.4459,
+      "step": 14271
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.901902341574742e-06,
+      "loss": 0.4444,
+      "step": 14272
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900581264991255e-06,
+      "loss": 0.4839,
+      "step": 14273
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.899260248246016e-06,
+      "loss": 0.5049,
+      "step": 14274
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897939291364528e-06,
+      "loss": 0.435,
+      "step": 14275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.896618394372296e-06,
+      "loss": 0.4665,
+      "step": 14276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.895297557294824e-06,
+      "loss": 0.4546,
+      "step": 14277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.893976780157605e-06,
+      "loss": 0.4777,
+      "step": 14278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.892656062986145e-06,
+      "loss": 0.4675,
+      "step": 14279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8913354058059374e-06,
+      "loss": 0.4459,
+      "step": 14280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.890014808642486e-06,
+      "loss": 0.4277,
+      "step": 14281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888694271521279e-06,
+      "loss": 0.4393,
+      "step": 14282
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8873737944678145e-06,
+      "loss": 0.4809,
+      "step": 14283
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.886053377507587e-06,
+      "loss": 0.4511,
+      "step": 14284
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884733020666086e-06,
+      "loss": 0.4454,
+      "step": 14285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8834127239688055e-06,
+      "loss": 0.4701,
+      "step": 14286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882092487441232e-06,
+      "loss": 0.4761,
+      "step": 14287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.880772311108861e-06,
+      "loss": 0.4726,
+      "step": 14288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.879452194997173e-06,
+      "loss": 0.484,
+      "step": 14289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878132139131657e-06,
+      "loss": 0.1886,
+      "step": 14290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.876812143537799e-06,
+      "loss": 0.4654,
+      "step": 14291
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.875492208241084e-06,
+      "loss": 0.4482,
+      "step": 14292
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874172333266993e-06,
+      "loss": 0.4309,
+      "step": 14293
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.872852518641009e-06,
+      "loss": 0.4929,
+      "step": 14294
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871532764388613e-06,
+      "loss": 0.4694,
+      "step": 14295
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870213070535285e-06,
+      "loss": 0.4505,
+      "step": 14296
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868893437106501e-06,
+      "loss": 0.4522,
+      "step": 14297
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.867573864127739e-06,
+      "loss": 0.4412,
+      "step": 14298
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8662543516244775e-06,
+      "loss": 0.4661,
+      "step": 14299
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.864934899622191e-06,
+      "loss": 0.4764,
+      "step": 14300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863615508146348e-06,
+      "loss": 0.1658,
+      "step": 14301
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.862296177222427e-06,
+      "loss": 0.4768,
+      "step": 14302
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.860976906875897e-06,
+      "loss": 0.4525,
+      "step": 14303
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.859657697132229e-06,
+      "loss": 0.4601,
+      "step": 14304
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858338548016892e-06,
+      "loss": 0.4631,
+      "step": 14305
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.857019459555352e-06,
+      "loss": 0.4733,
+      "step": 14306
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.855700431773079e-06,
+      "loss": 0.4623,
+      "step": 14307
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854381464695534e-06,
+      "loss": 0.4534,
+      "step": 14308
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853062558348184e-06,
+      "loss": 0.4701,
+      "step": 14309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851743712756494e-06,
+      "loss": 0.4887,
+      "step": 14310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850424927945925e-06,
+      "loss": 0.4611,
+      "step": 14311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849106203941935e-06,
+      "loss": 0.4768,
+      "step": 14312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.847787540769986e-06,
+      "loss": 0.4519,
+      "step": 14313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.846468938455535e-06,
+      "loss": 0.4816,
+      "step": 14314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.845150397024043e-06,
+      "loss": 0.4626,
+      "step": 14315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.843831916500961e-06,
+      "loss": 0.485,
+      "step": 14316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.842513496911747e-06,
+      "loss": 0.4586,
+      "step": 14317
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841195138281855e-06,
+      "loss": 0.4952,
+      "step": 14318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.839876840636739e-06,
+      "loss": 0.4858,
+      "step": 14319
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.838558604001843e-06,
+      "loss": 0.1729,
+      "step": 14320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837240428402626e-06,
+      "loss": 0.5061,
+      "step": 14321
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.835922313864532e-06,
+      "loss": 0.431,
+      "step": 14322
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834604260413014e-06,
+      "loss": 0.4635,
+      "step": 14323
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.833286268073512e-06,
+      "loss": 0.4657,
+      "step": 14324
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.831968336871475e-06,
+      "loss": 0.457,
+      "step": 14325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.830650466832349e-06,
+      "loss": 0.4261,
+      "step": 14326
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.829332657981573e-06,
+      "loss": 0.4756,
+      "step": 14327
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.828014910344591e-06,
+      "loss": 0.5085,
+      "step": 14328
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.826697223946846e-06,
+      "loss": 0.4679,
+      "step": 14329
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825379598813776e-06,
+      "loss": 0.4497,
+      "step": 14330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.824062034970817e-06,
+      "loss": 0.5001,
+      "step": 14331
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.822744532443409e-06,
+      "loss": 0.4546,
+      "step": 14332
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8214270912569844e-06,
+      "loss": 0.469,
+      "step": 14333
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.820109711436989e-06,
+      "loss": 0.4472,
+      "step": 14334
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.818792393008844e-06,
+      "loss": 0.4748,
+      "step": 14335
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8174751359979866e-06,
+      "loss": 0.4565,
+      "step": 14336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.816157940429844e-06,
+      "loss": 0.451,
+      "step": 14337
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.814840806329859e-06,
+      "loss": 0.4604,
+      "step": 14338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.813523733723446e-06,
+      "loss": 0.4697,
+      "step": 14339
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8122067226360415e-06,
+      "loss": 0.4533,
+      "step": 14340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.810889773093065e-06,
+      "loss": 0.4567,
+      "step": 14341
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.809572885119952e-06,
+      "loss": 0.4443,
+      "step": 14342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808256058742119e-06,
+      "loss": 0.4497,
+      "step": 14343
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806939293984988e-06,
+      "loss": 0.4623,
+      "step": 14344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.805622590873985e-06,
+      "loss": 0.4476,
+      "step": 14345
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.804305949434534e-06,
+      "loss": 0.4768,
+      "step": 14346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.802989369692046e-06,
+      "loss": 0.4775,
+      "step": 14347
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.801672851671941e-06,
+      "loss": 0.4688,
+      "step": 14348
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800356395399645e-06,
+      "loss": 0.4189,
+      "step": 14349
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.799040000900563e-06,
+      "loss": 0.4459,
+      "step": 14350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797723668200113e-06,
+      "loss": 0.4485,
+      "step": 14351
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7964073973237075e-06,
+      "loss": 0.4453,
+      "step": 14352
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7950911882967675e-06,
+      "loss": 0.4574,
+      "step": 14353
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793775041144692e-06,
+      "loss": 0.4496,
+      "step": 14354
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.792458955892895e-06,
+      "loss": 0.5056,
+      "step": 14355
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7911429325667866e-06,
+      "loss": 0.4662,
+      "step": 14356
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.78982697119178e-06,
+      "loss": 0.4701,
+      "step": 14357
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.78851107179327e-06,
+      "loss": 0.4973,
+      "step": 14358
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.787195234396667e-06,
+      "loss": 0.4344,
+      "step": 14359
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.785879459027376e-06,
+      "loss": 0.4683,
+      "step": 14360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7845637457108016e-06,
+      "loss": 0.4921,
+      "step": 14361
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7832480944723385e-06,
+      "loss": 0.4598,
+      "step": 14362
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.781932505337393e-06,
+      "loss": 0.4792,
+      "step": 14363
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.780616978331362e-06,
+      "loss": 0.4467,
+      "step": 14364
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.779301513479648e-06,
+      "loss": 0.4602,
+      "step": 14365
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.777986110807637e-06,
+      "loss": 0.4448,
+      "step": 14366
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.776670770340735e-06,
+      "loss": 0.4737,
+      "step": 14367
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.775355492104329e-06,
+      "loss": 0.4764,
+      "step": 14368
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.774040276123823e-06,
+      "loss": 0.5007,
+      "step": 14369
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772725122424594e-06,
+      "loss": 0.4783,
+      "step": 14370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.771410031032042e-06,
+      "loss": 0.4503,
+      "step": 14371
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.770095001971559e-06,
+      "loss": 0.4624,
+      "step": 14372
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.768780035268523e-06,
+      "loss": 0.4613,
+      "step": 14373
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.76746513094833e-06,
+      "loss": 0.4926,
+      "step": 14374
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7661502890363615e-06,
+      "loss": 0.4693,
+      "step": 14375
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.76483550955801e-06,
+      "loss": 0.4869,
+      "step": 14376
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.763520792538645e-06,
+      "loss": 0.4614,
+      "step": 14377
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7622061380036595e-06,
+      "loss": 0.4798,
+      "step": 14378
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.760891545978431e-06,
+      "loss": 0.4359,
+      "step": 14379
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759577016488343e-06,
+      "loss": 0.4339,
+      "step": 14380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.758262549558767e-06,
+      "loss": 0.5002,
+      "step": 14381
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.756948145215087e-06,
+      "loss": 0.4619,
+      "step": 14382
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755633803482676e-06,
+      "loss": 0.4563,
+      "step": 14383
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.754319524386912e-06,
+      "loss": 0.4438,
+      "step": 14384
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7530053079531664e-06,
+      "loss": 0.4671,
+      "step": 14385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751691154206811e-06,
+      "loss": 0.4371,
+      "step": 14386
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.750377063173219e-06,
+      "loss": 0.4691,
+      "step": 14387
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749063034877761e-06,
+      "loss": 0.4515,
+      "step": 14388
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7477490693458035e-06,
+      "loss": 0.4788,
+      "step": 14389
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746435166602716e-06,
+      "loss": 0.4975,
+      "step": 14390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.745121326673868e-06,
+      "loss": 0.4799,
+      "step": 14391
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74380754958462e-06,
+      "loss": 0.4351,
+      "step": 14392
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.742493835360336e-06,
+      "loss": 0.4356,
+      "step": 14393
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.741180184026383e-06,
+      "loss": 0.4677,
+      "step": 14394
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.739866595608121e-06,
+      "loss": 0.4833,
+      "step": 14395
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7385530701309085e-06,
+      "loss": 0.502,
+      "step": 14396
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737239607620108e-06,
+      "loss": 0.4447,
+      "step": 14397
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.735926208101074e-06,
+      "loss": 0.4786,
+      "step": 14398
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.734612871599169e-06,
+      "loss": 0.4552,
+      "step": 14399
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.733299598139742e-06,
+      "loss": 0.4799,
+      "step": 14400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.73198638774815e-06,
+      "loss": 0.4964,
+      "step": 14401
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7306732404497475e-06,
+      "loss": 0.4771,
+      "step": 14402
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729360156269889e-06,
+      "loss": 0.4782,
+      "step": 14403
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7280471352339185e-06,
+      "loss": 0.1834,
+      "step": 14404
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.726734177367189e-06,
+      "loss": 0.4555,
+      "step": 14405
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.725421282695048e-06,
+      "loss": 0.4613,
+      "step": 14406
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.724108451242848e-06,
+      "loss": 0.4515,
+      "step": 14407
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.722795683035925e-06,
+      "loss": 0.4853,
+      "step": 14408
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.721482978099631e-06,
+      "loss": 0.4851,
+      "step": 14409
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.720170336459308e-06,
+      "loss": 0.4696,
+      "step": 14410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.718857758140298e-06,
+      "loss": 0.4737,
+      "step": 14411
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.717545243167941e-06,
+      "loss": 0.4602,
+      "step": 14412
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.716232791567576e-06,
+      "loss": 0.4747,
+      "step": 14413
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.714920403364546e-06,
+      "loss": 0.4961,
+      "step": 14414
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.713608078584184e-06,
+      "loss": 0.4754,
+      "step": 14415
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.712295817251827e-06,
+      "loss": 0.4622,
+      "step": 14416
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7109836193928104e-06,
+      "loss": 0.4246,
+      "step": 14417
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7096714850324694e-06,
+      "loss": 0.4568,
+      "step": 14418
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7083594141961326e-06,
+      "loss": 0.4653,
+      "step": 14419
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.707047406909135e-06,
+      "loss": 0.4449,
+      "step": 14420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.705735463196803e-06,
+      "loss": 0.4429,
+      "step": 14421
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7044235830844714e-06,
+      "loss": 0.4904,
+      "step": 14422
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.703111766597461e-06,
+      "loss": 0.4702,
+      "step": 14423
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7018000137611e-06,
+      "loss": 0.4685,
+      "step": 14424
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.700488324600715e-06,
+      "loss": 0.1732,
+      "step": 14425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699176699141631e-06,
+      "loss": 0.4434,
+      "step": 14426
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697865137409166e-06,
+      "loss": 0.4804,
+      "step": 14427
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.696553639428645e-06,
+      "loss": 0.4347,
+      "step": 14428
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.695242205225387e-06,
+      "loss": 0.4565,
+      "step": 14429
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693930834824713e-06,
+      "loss": 0.4657,
+      "step": 14430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692619528251938e-06,
+      "loss": 0.476,
+      "step": 14431
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.691308285532379e-06,
+      "loss": 0.4544,
+      "step": 14432
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.689997106691352e-06,
+      "loss": 0.4439,
+      "step": 14433
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.688685991754169e-06,
+      "loss": 0.4615,
+      "step": 14434
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687374940746145e-06,
+      "loss": 0.4481,
+      "step": 14435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.686063953692591e-06,
+      "loss": 0.4782,
+      "step": 14436
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684753030618819e-06,
+      "loss": 0.4451,
+      "step": 14437
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.683442171550134e-06,
+      "loss": 0.4551,
+      "step": 14438
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682131376511845e-06,
+      "loss": 0.4628,
+      "step": 14439
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.68082064552926e-06,
+      "loss": 0.4749,
+      "step": 14440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6795099786276876e-06,
+      "loss": 0.4707,
+      "step": 14441
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.678199375832424e-06,
+      "loss": 0.4757,
+      "step": 14442
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.676888837168777e-06,
+      "loss": 0.4658,
+      "step": 14443
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.675578362662047e-06,
+      "loss": 0.4684,
+      "step": 14444
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.674267952337538e-06,
+      "loss": 0.4732,
+      "step": 14445
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.672957606220543e-06,
+      "loss": 0.4867,
+      "step": 14446
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6716473243363635e-06,
+      "loss": 0.4825,
+      "step": 14447
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670337106710294e-06,
+      "loss": 0.4551,
+      "step": 14448
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.669026953367636e-06,
+      "loss": 0.4295,
+      "step": 14449
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667716864333675e-06,
+      "loss": 0.4598,
+      "step": 14450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.66640683963371e-06,
+      "loss": 0.4324,
+      "step": 14451
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.665096879293027e-06,
+      "loss": 0.4581,
+      "step": 14452
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663786983336928e-06,
+      "loss": 0.4816,
+      "step": 14453
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.662477151790689e-06,
+      "loss": 0.4775,
+      "step": 14454
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.661167384679605e-06,
+      "loss": 0.4729,
+      "step": 14455
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.659857682028963e-06,
+      "loss": 0.4279,
+      "step": 14456
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.658548043864044e-06,
+      "loss": 0.4533,
+      "step": 14457
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657238470210135e-06,
+      "loss": 0.478,
+      "step": 14458
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.655928961092519e-06,
+      "loss": 0.4404,
+      "step": 14459
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.65461951653648e-06,
+      "loss": 0.4596,
+      "step": 14460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653310136567295e-06,
+      "loss": 0.4599,
+      "step": 14461
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.652000821210242e-06,
+      "loss": 0.4518,
+      "step": 14462
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.650691570490602e-06,
+      "loss": 0.4845,
+      "step": 14463
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.649382384433657e-06,
+      "loss": 0.4345,
+      "step": 14464
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.648073263064672e-06,
+      "loss": 0.464,
+      "step": 14465
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.646764206408927e-06,
+      "loss": 0.4554,
+      "step": 14466
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.64545521449169e-06,
+      "loss": 0.447,
+      "step": 14467
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.644146287338244e-06,
+      "loss": 0.4817,
+      "step": 14468
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.642837424973849e-06,
+      "loss": 0.4565,
+      "step": 14469
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6415286274237744e-06,
+      "loss": 0.4562,
+      "step": 14470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.640219894713293e-06,
+      "loss": 0.4482,
+      "step": 14471
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.638911226867676e-06,
+      "loss": 0.4543,
+      "step": 14472
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.637602623912178e-06,
+      "loss": 0.4808,
+      "step": 14473
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.636294085872066e-06,
+      "loss": 0.4649,
+      "step": 14474
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.634985612772611e-06,
+      "loss": 0.5135,
+      "step": 14475
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.633677204639064e-06,
+      "loss": 0.4697,
+      "step": 14476
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.632368861496692e-06,
+      "loss": 0.4669,
+      "step": 14477
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6310605833707495e-06,
+      "loss": 0.4824,
+      "step": 14478
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.629752370286504e-06,
+      "loss": 0.4744,
+      "step": 14479
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.628444222269203e-06,
+      "loss": 0.4837,
+      "step": 14480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627136139344102e-06,
+      "loss": 0.4823,
+      "step": 14481
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.625828121536458e-06,
+      "loss": 0.4791,
+      "step": 14482
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.624520168871531e-06,
+      "loss": 0.4461,
+      "step": 14483
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.62321228137456e-06,
+      "loss": 0.45,
+      "step": 14484
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621904459070798e-06,
+      "loss": 0.4789,
+      "step": 14485
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.620596701985502e-06,
+      "loss": 0.4784,
+      "step": 14486
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.619289010143918e-06,
+      "loss": 0.4495,
+      "step": 14487
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6179813835712855e-06,
+      "loss": 0.459,
+      "step": 14488
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.616673822292853e-06,
+      "loss": 0.4613,
+      "step": 14489
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.615366326333868e-06,
+      "loss": 0.4351,
+      "step": 14490
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.614058895719575e-06,
+      "loss": 0.1863,
+      "step": 14491
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6127515304752056e-06,
+      "loss": 0.4891,
+      "step": 14492
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.611444230626009e-06,
+      "loss": 0.4569,
+      "step": 14493
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.610136996197221e-06,
+      "loss": 0.4709,
+      "step": 14494
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.608829827214085e-06,
+      "loss": 0.4636,
+      "step": 14495
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.607522723701827e-06,
+      "loss": 0.4705,
+      "step": 14496
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6062156856856905e-06,
+      "loss": 0.4313,
+      "step": 14497
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.604908713190909e-06,
+      "loss": 0.4536,
+      "step": 14498
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.60360180624271e-06,
+      "loss": 0.4837,
+      "step": 14499
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6022949648663294e-06,
+      "loss": 0.44,
+      "step": 14500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6009881890869964e-06,
+      "loss": 0.4612,
+      "step": 14501
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.599681478929944e-06,
+      "loss": 0.4416,
+      "step": 14502
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598374834420391e-06,
+      "loss": 0.4328,
+      "step": 14503
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.59706825558357e-06,
+      "loss": 0.4791,
+      "step": 14504
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595761742444705e-06,
+      "loss": 0.4488,
+      "step": 14505
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5944552950290256e-06,
+      "loss": 0.4853,
+      "step": 14506
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.59314891336174e-06,
+      "loss": 0.4541,
+      "step": 14507
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.591842597468081e-06,
+      "loss": 0.4914,
+      "step": 14508
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.590536347373268e-06,
+      "loss": 0.4667,
+      "step": 14509
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.589230163102518e-06,
+      "loss": 0.4619,
+      "step": 14510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.587924044681046e-06,
+      "loss": 0.4736,
+      "step": 14511
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.586617992134072e-06,
+      "loss": 0.4393,
+      "step": 14512
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.58531200548681e-06,
+      "loss": 0.165,
+      "step": 14513
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.584006084764476e-06,
+      "loss": 0.4355,
+      "step": 14514
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582700229992276e-06,
+      "loss": 0.4764,
+      "step": 14515
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.581394441195426e-06,
+      "loss": 0.4416,
+      "step": 14516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.580088718399136e-06,
+      "loss": 0.4537,
+      "step": 14517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.578783061628615e-06,
+      "loss": 0.4503,
+      "step": 14518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577477470909066e-06,
+      "loss": 0.4347,
+      "step": 14519
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.576171946265699e-06,
+      "loss": 0.4562,
+      "step": 14520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5748664877237215e-06,
+      "loss": 0.4862,
+      "step": 14521
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.573561095308329e-06,
+      "loss": 0.4737,
+      "step": 14522
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572255769044731e-06,
+      "loss": 0.44,
+      "step": 14523
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.570950508958125e-06,
+      "loss": 0.4604,
+      "step": 14524
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569645315073713e-06,
+      "loss": 0.5091,
+      "step": 14525
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.568340187416691e-06,
+      "loss": 0.4286,
+      "step": 14526
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5670351260122554e-06,
+      "loss": 0.4641,
+      "step": 14527
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5657301308856055e-06,
+      "loss": 0.1723,
+      "step": 14528
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.564425202061935e-06,
+      "loss": 0.4696,
+      "step": 14529
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563120339566436e-06,
+      "loss": 0.4587,
+      "step": 14530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.561815543424299e-06,
+      "loss": 0.429,
+      "step": 14531
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.560510813660719e-06,
+      "loss": 0.4742,
+      "step": 14532
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.559206150300884e-06,
+      "loss": 0.4448,
+      "step": 14533
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.557901553369979e-06,
+      "loss": 0.4805,
+      "step": 14534
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556597022893194e-06,
+      "loss": 0.4658,
+      "step": 14535
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.555292558895715e-06,
+      "loss": 0.4787,
+      "step": 14536
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.553988161402727e-06,
+      "loss": 0.4377,
+      "step": 14537
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552683830439408e-06,
+      "loss": 0.4645,
+      "step": 14538
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.551379566030944e-06,
+      "loss": 0.4576,
+      "step": 14539
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.550075368202518e-06,
+      "loss": 0.4558,
+      "step": 14540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.548771236979302e-06,
+      "loss": 0.4463,
+      "step": 14541
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5474671723864795e-06,
+      "loss": 0.4423,
+      "step": 14542
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546163174449224e-06,
+      "loss": 0.4515,
+      "step": 14543
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.544859243192715e-06,
+      "loss": 0.4328,
+      "step": 14544
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543555378642121e-06,
+      "loss": 0.438,
+      "step": 14545
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.542251580822618e-06,
+      "loss": 0.4521,
+      "step": 14546
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.540947849759378e-06,
+      "loss": 0.4739,
+      "step": 14547
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.539644185477571e-06,
+      "loss": 0.1669,
+      "step": 14548
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.538340588002366e-06,
+      "loss": 0.4858,
+      "step": 14549
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537037057358928e-06,
+      "loss": 0.4676,
+      "step": 14550
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.535733593572424e-06,
+      "loss": 0.4713,
+      "step": 14551
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.534430196668025e-06,
+      "loss": 0.4727,
+      "step": 14552
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533126866670886e-06,
+      "loss": 0.4365,
+      "step": 14553
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531823603606173e-06,
+      "loss": 0.4803,
+      "step": 14554
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.530520407499049e-06,
+      "loss": 0.4546,
+      "step": 14555
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.529217278374674e-06,
+      "loss": 0.4789,
+      "step": 14556
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.527914216258203e-06,
+      "loss": 0.4554,
+      "step": 14557
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.526611221174796e-06,
+      "loss": 0.4923,
+      "step": 14558
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.525308293149608e-06,
+      "loss": 0.4685,
+      "step": 14559
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524005432207795e-06,
+      "loss": 0.4619,
+      "step": 14560
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522702638374509e-06,
+      "loss": 0.4436,
+      "step": 14561
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.521399911674902e-06,
+      "loss": 0.4576,
+      "step": 14562
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520097252134125e-06,
+      "loss": 0.4827,
+      "step": 14563
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5187946597773284e-06,
+      "loss": 0.4713,
+      "step": 14564
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517492134629659e-06,
+      "loss": 0.4518,
+      "step": 14565
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.516189676716266e-06,
+      "loss": 0.4665,
+      "step": 14566
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.514887286062295e-06,
+      "loss": 0.4678,
+      "step": 14567
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513584962692885e-06,
+      "loss": 0.4742,
+      "step": 14568
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.512282706633184e-06,
+      "loss": 0.46,
+      "step": 14569
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.510980517908334e-06,
+      "loss": 0.4746,
+      "step": 14570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5096783965434755e-06,
+      "loss": 0.4832,
+      "step": 14571
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.508376342563742e-06,
+      "loss": 0.4583,
+      "step": 14572
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507074355994279e-06,
+      "loss": 0.4934,
+      "step": 14573
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.505772436860216e-06,
+      "loss": 0.4713,
+      "step": 14574
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.504470585186696e-06,
+      "loss": 0.4592,
+      "step": 14575
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.503168800998847e-06,
+      "loss": 0.4704,
+      "step": 14576
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501867084321802e-06,
+      "loss": 0.4341,
+      "step": 14577
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.500565435180693e-06,
+      "loss": 0.4948,
+      "step": 14578
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.499263853600655e-06,
+      "loss": 0.4451,
+      "step": 14579
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4979623396068095e-06,
+      "loss": 0.4414,
+      "step": 14580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4966608932242845e-06,
+      "loss": 0.4359,
+      "step": 14581
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.495359514478212e-06,
+      "loss": 0.4715,
+      "step": 14582
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.494058203393711e-06,
+      "loss": 0.4545,
+      "step": 14583
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.492756959995908e-06,
+      "loss": 0.4685,
+      "step": 14584
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.491455784309923e-06,
+      "loss": 0.4519,
+      "step": 14585
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.490154676360879e-06,
+      "loss": 0.171,
+      "step": 14586
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.488853636173894e-06,
+      "loss": 0.4669,
+      "step": 14587
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487552663774086e-06,
+      "loss": 0.4453,
+      "step": 14588
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.486251759186573e-06,
+      "loss": 0.4722,
+      "step": 14589
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484950922436475e-06,
+      "loss": 0.4739,
+      "step": 14590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.483650153548897e-06,
+      "loss": 0.4226,
+      "step": 14591
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.482349452548957e-06,
+      "loss": 0.4446,
+      "step": 14592
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.481048819461765e-06,
+      "loss": 0.458,
+      "step": 14593
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479748254312439e-06,
+      "loss": 0.454,
+      "step": 14594
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4784477571260775e-06,
+      "loss": 0.462,
+      "step": 14595
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477147327927793e-06,
+      "loss": 0.466,
+      "step": 14596
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.47584696674269e-06,
+      "loss": 0.435,
+      "step": 14597
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474546673595881e-06,
+      "loss": 0.4535,
+      "step": 14598
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.473246448512458e-06,
+      "loss": 0.4958,
+      "step": 14599
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4719462915175305e-06,
+      "loss": 0.4555,
+      "step": 14600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4706462026362e-06,
+      "loss": 0.4586,
+      "step": 14601
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.469346181893568e-06,
+      "loss": 0.4425,
+      "step": 14602
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468046229314728e-06,
+      "loss": 0.4563,
+      "step": 14603
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.466746344924777e-06,
+      "loss": 0.4431,
+      "step": 14604
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465446528748818e-06,
+      "loss": 0.4742,
+      "step": 14605
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464146780811938e-06,
+      "loss": 0.468,
+      "step": 14606
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462847101139233e-06,
+      "loss": 0.4562,
+      "step": 14607
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.461547489755795e-06,
+      "loss": 0.4499,
+      "step": 14608
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.46024794668672e-06,
+      "loss": 0.4444,
+      "step": 14609
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458948471957088e-06,
+      "loss": 0.49,
+      "step": 14610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.45764906559199e-06,
+      "loss": 0.4529,
+      "step": 14611
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4563497276165155e-06,
+      "loss": 0.4573,
+      "step": 14612
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4550504580557525e-06,
+      "loss": 0.4674,
+      "step": 14613
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.453751256934777e-06,
+      "loss": 0.4552,
+      "step": 14614
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.452452124278674e-06,
+      "loss": 0.4874,
+      "step": 14615
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.45115306011253e-06,
+      "loss": 0.4685,
+      "step": 14616
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.449854064461423e-06,
+      "loss": 0.4651,
+      "step": 14617
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448555137350427e-06,
+      "loss": 0.4566,
+      "step": 14618
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.447256278804623e-06,
+      "loss": 0.4604,
+      "step": 14619
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445957488849088e-06,
+      "loss": 0.4813,
+      "step": 14620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4446587675089e-06,
+      "loss": 0.4468,
+      "step": 14621
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.443360114809123e-06,
+      "loss": 0.4773,
+      "step": 14622
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442061530774835e-06,
+      "loss": 0.4648,
+      "step": 14623
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.440763015431111e-06,
+      "loss": 0.4241,
+      "step": 14624
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.439464568803014e-06,
+      "loss": 0.4674,
+      "step": 14625
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438166190915609e-06,
+      "loss": 0.441,
+      "step": 14626
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.436867881793971e-06,
+      "loss": 0.4419,
+      "step": 14627
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.435569641463167e-06,
+      "loss": 0.4726,
+      "step": 14628
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.43427146994825e-06,
+      "loss": 0.4463,
+      "step": 14629
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.432973367274292e-06,
+      "loss": 0.4324,
+      "step": 14630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.431675333466352e-06,
+      "loss": 0.469,
+      "step": 14631
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.430377368549494e-06,
+      "loss": 0.448,
+      "step": 14632
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429079472548767e-06,
+      "loss": 0.4578,
+      "step": 14633
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.427781645489237e-06,
+      "loss": 0.4338,
+      "step": 14634
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426483887395958e-06,
+      "loss": 0.4758,
+      "step": 14635
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4251861982939865e-06,
+      "loss": 0.4741,
+      "step": 14636
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.423888578208373e-06,
+      "loss": 0.4766,
+      "step": 14637
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.422591027164169e-06,
+      "loss": 0.4639,
+      "step": 14638
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.421293545186429e-06,
+      "loss": 0.4695,
+      "step": 14639
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.419996132300203e-06,
+      "loss": 0.477,
+      "step": 14640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418698788530535e-06,
+      "loss": 0.4826,
+      "step": 14641
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.417401513902473e-06,
+      "loss": 0.4643,
+      "step": 14642
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.416104308441065e-06,
+      "loss": 0.4517,
+      "step": 14643
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.414807172171356e-06,
+      "loss": 0.4577,
+      "step": 14644
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413510105118383e-06,
+      "loss": 0.4542,
+      "step": 14645
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.412213107307192e-06,
+      "loss": 0.4596,
+      "step": 14646
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.410916178762825e-06,
+      "loss": 0.435,
+      "step": 14647
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409619319510316e-06,
+      "loss": 0.4608,
+      "step": 14648
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.408322529574704e-06,
+      "loss": 0.4661,
+      "step": 14649
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.407025808981028e-06,
+      "loss": 0.4915,
+      "step": 14650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.405729157754321e-06,
+      "loss": 0.4919,
+      "step": 14651
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.404432575919615e-06,
+      "loss": 0.4636,
+      "step": 14652
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403136063501945e-06,
+      "loss": 0.4566,
+      "step": 14653
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.401839620526337e-06,
+      "loss": 0.4884,
+      "step": 14654
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.40054324701783e-06,
+      "loss": 0.4779,
+      "step": 14655
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.399246943001442e-06,
+      "loss": 0.4678,
+      "step": 14656
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.397950708502203e-06,
+      "loss": 0.45,
+      "step": 14657
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.39665454354514e-06,
+      "loss": 0.4967,
+      "step": 14658
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.39535844815528e-06,
+      "loss": 0.4523,
+      "step": 14659
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.394062422357638e-06,
+      "loss": 0.4609,
+      "step": 14660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.39276646617724e-06,
+      "loss": 0.4603,
+      "step": 14661
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.391470579639107e-06,
+      "loss": 0.4698,
+      "step": 14662
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390174762768257e-06,
+      "loss": 0.468,
+      "step": 14663
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.388879015589707e-06,
+      "loss": 0.5021,
+      "step": 14664
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.387583338128471e-06,
+      "loss": 0.436,
+      "step": 14665
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.386287730409566e-06,
+      "loss": 0.458,
+      "step": 14666
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384992192458008e-06,
+      "loss": 0.4738,
+      "step": 14667
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3836967242988035e-06,
+      "loss": 0.4695,
+      "step": 14668
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3824013259569665e-06,
+      "loss": 0.475,
+      "step": 14669
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3811059974575065e-06,
+      "loss": 0.4727,
+      "step": 14670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.379810738825429e-06,
+      "loss": 0.4328,
+      "step": 14671
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.378515550085741e-06,
+      "loss": 0.474,
+      "step": 14672
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37722043126345e-06,
+      "loss": 0.4752,
+      "step": 14673
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.375925382383561e-06,
+      "loss": 0.4907,
+      "step": 14674
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.374630403471072e-06,
+      "loss": 0.4586,
+      "step": 14675
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373335494550987e-06,
+      "loss": 0.4459,
+      "step": 14676
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372040655648306e-06,
+      "loss": 0.4418,
+      "step": 14677
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370745886788028e-06,
+      "loss": 0.4669,
+      "step": 14678
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369451187995148e-06,
+      "loss": 0.4563,
+      "step": 14679
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.368156559294662e-06,
+      "loss": 0.4857,
+      "step": 14680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.366862000711566e-06,
+      "loss": 0.4379,
+      "step": 14681
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365567512270856e-06,
+      "loss": 0.455,
+      "step": 14682
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3642730939975176e-06,
+      "loss": 0.4635,
+      "step": 14683
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.362978745916544e-06,
+      "loss": 0.4434,
+      "step": 14684
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361684468052925e-06,
+      "loss": 0.449,
+      "step": 14685
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.360390260431648e-06,
+      "loss": 0.4645,
+      "step": 14686
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359096123077697e-06,
+      "loss": 0.4497,
+      "step": 14687
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35780205601606e-06,
+      "loss": 0.4746,
+      "step": 14688
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35650805927172e-06,
+      "loss": 0.4478,
+      "step": 14689
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.355214132869658e-06,
+      "loss": 0.4872,
+      "step": 14690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.353920276834855e-06,
+      "loss": 0.4515,
+      "step": 14691
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.352626491192292e-06,
+      "loss": 0.4629,
+      "step": 14692
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351332775966947e-06,
+      "loss": 0.4851,
+      "step": 14693
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350039131183796e-06,
+      "loss": 0.4881,
+      "step": 14694
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348745556867814e-06,
+      "loss": 0.4245,
+      "step": 14695
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.347452053043974e-06,
+      "loss": 0.4947,
+      "step": 14696
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.346158619737255e-06,
+      "loss": 0.4974,
+      "step": 14697
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.344865256972621e-06,
+      "loss": 0.4534,
+      "step": 14698
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3435719647750445e-06,
+      "loss": 0.48,
+      "step": 14699
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3422787431694954e-06,
+      "loss": 0.479,
+      "step": 14700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340985592180942e-06,
+      "loss": 0.4734,
+      "step": 14701
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.339692511834347e-06,
+      "loss": 0.4771,
+      "step": 14702
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.338399502154675e-06,
+      "loss": 0.4487,
+      "step": 14703
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33710656316689e-06,
+      "loss": 0.4784,
+      "step": 14704
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33581369489596e-06,
+      "loss": 0.46,
+      "step": 14705
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3345208973668355e-06,
+      "loss": 0.4659,
+      "step": 14706
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.333228170604481e-06,
+      "loss": 0.4462,
+      "step": 14707
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33193551463385e-06,
+      "loss": 0.4561,
+      "step": 14708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.330642929479909e-06,
+      "loss": 0.4753,
+      "step": 14709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3293504151676026e-06,
+      "loss": 0.4397,
+      "step": 14710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.328057971721887e-06,
+      "loss": 0.4629,
+      "step": 14711
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326765599167719e-06,
+      "loss": 0.4442,
+      "step": 14712
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3254732975300426e-06,
+      "loss": 0.4824,
+      "step": 14713
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324181066833811e-06,
+      "loss": 0.4706,
+      "step": 14714
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3228889071039705e-06,
+      "loss": 0.4453,
+      "step": 14715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321596818365476e-06,
+      "loss": 0.4892,
+      "step": 14716
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320304800643262e-06,
+      "loss": 0.4562,
+      "step": 14717
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.319012853962277e-06,
+      "loss": 0.4776,
+      "step": 14718
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3177209783474616e-06,
+      "loss": 0.4632,
+      "step": 14719
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316429173823765e-06,
+      "loss": 0.4604,
+      "step": 14720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.315137440416117e-06,
+      "loss": 0.4514,
+      "step": 14721
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.313845778149461e-06,
+      "loss": 0.4613,
+      "step": 14722
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.312554187048733e-06,
+      "loss": 0.4459,
+      "step": 14723
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.311262667138875e-06,
+      "loss": 0.4527,
+      "step": 14724
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.30997121844481e-06,
+      "loss": 0.4751,
+      "step": 14725
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3086798409914775e-06,
+      "loss": 0.4897,
+      "step": 14726
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.307388534803809e-06,
+      "loss": 0.4472,
+      "step": 14727
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3060972999067396e-06,
+      "loss": 0.4653,
+      "step": 14728
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.30480613632519e-06,
+      "loss": 0.4981,
+      "step": 14729
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.303515044084087e-06,
+      "loss": 0.4539,
+      "step": 14730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3022240232083676e-06,
+      "loss": 0.4529,
+      "step": 14731
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.300933073722947e-06,
+      "loss": 0.457,
+      "step": 14732
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2996421956527495e-06,
+      "loss": 0.4945,
+      "step": 14733
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.298351389022699e-06,
+      "loss": 0.474,
+      "step": 14734
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2970606538577205e-06,
+      "loss": 0.4598,
+      "step": 14735
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295769990182725e-06,
+      "loss": 0.4794,
+      "step": 14736
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.294479398022635e-06,
+      "loss": 0.4689,
+      "step": 14737
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2931888774023656e-06,
+      "loss": 0.4824,
+      "step": 14738
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2918984283468385e-06,
+      "loss": 0.4512,
+      "step": 14739
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.290608050880957e-06,
+      "loss": 0.4436,
+      "step": 14740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.289317745029637e-06,
+      "loss": 0.4597,
+      "step": 14741
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2880275108177915e-06,
+      "loss": 0.4731,
+      "step": 14742
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2867373482703355e-06,
+      "loss": 0.4385,
+      "step": 14743
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.285447257412163e-06,
+      "loss": 0.4651,
+      "step": 14744
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.284157238268193e-06,
+      "loss": 0.449,
+      "step": 14745
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.282867290863326e-06,
+      "loss": 0.4664,
+      "step": 14746
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.281577415222471e-06,
+      "loss": 0.4635,
+      "step": 14747
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280287611370522e-06,
+      "loss": 0.4637,
+      "step": 14748
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.278997879332385e-06,
+      "loss": 0.4662,
+      "step": 14749
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277708219132961e-06,
+      "loss": 0.4711,
+      "step": 14750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276418630797152e-06,
+      "loss": 0.4778,
+      "step": 14751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.275129114349845e-06,
+      "loss": 0.4448,
+      "step": 14752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.273839669815943e-06,
+      "loss": 0.4652,
+      "step": 14753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.272550297220343e-06,
+      "loss": 0.4611,
+      "step": 14754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271260996587928e-06,
+      "loss": 0.4893,
+      "step": 14755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.269971767943597e-06,
+      "loss": 0.4668,
+      "step": 14756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268682611312239e-06,
+      "loss": 0.4407,
+      "step": 14757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267393526718747e-06,
+      "loss": 0.4607,
+      "step": 14758
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266104514187997e-06,
+      "loss": 0.4583,
+      "step": 14759
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.264815573744884e-06,
+      "loss": 0.4898,
+      "step": 14760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2635267054142915e-06,
+      "loss": 0.4327,
+      "step": 14761
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.262237909221105e-06,
+      "loss": 0.4401,
+      "step": 14762
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.260949185190198e-06,
+      "loss": 0.4752,
+      "step": 14763
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259660533346459e-06,
+      "loss": 0.4545,
+      "step": 14764
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258371953714762e-06,
+      "loss": 0.4193,
+      "step": 14765
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2570834463199905e-06,
+      "loss": 0.4618,
+      "step": 14766
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.255795011187015e-06,
+      "loss": 0.4416,
+      "step": 14767
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.254506648340713e-06,
+      "loss": 0.4577,
+      "step": 14768
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253218357805957e-06,
+      "loss": 0.4676,
+      "step": 14769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.251930139607621e-06,
+      "loss": 0.4422,
+      "step": 14770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250641993770574e-06,
+      "loss": 0.4589,
+      "step": 14771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.249353920319685e-06,
+      "loss": 0.4383,
+      "step": 14772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.248065919279824e-06,
+      "loss": 0.4701,
+      "step": 14773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246777990675855e-06,
+      "loss": 0.4605,
+      "step": 14774
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245490134532644e-06,
+      "loss": 0.4719,
+      "step": 14775
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.244202350875055e-06,
+      "loss": 0.4442,
+      "step": 14776
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242914639727953e-06,
+      "loss": 0.4708,
+      "step": 14777
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241627001116193e-06,
+      "loss": 0.45,
+      "step": 14778
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240339435064638e-06,
+      "loss": 0.4656,
+      "step": 14779
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239051941598147e-06,
+      "loss": 0.435,
+      "step": 14780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.237764520741576e-06,
+      "loss": 0.4849,
+      "step": 14781
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236477172519779e-06,
+      "loss": 0.4599,
+      "step": 14782
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2351898969576095e-06,
+      "loss": 0.4504,
+      "step": 14783
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233902694079923e-06,
+      "loss": 0.4515,
+      "step": 14784
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.23261556391157e-06,
+      "loss": 0.4593,
+      "step": 14785
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.231328506477397e-06,
+      "loss": 0.4749,
+      "step": 14786
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.230041521802254e-06,
+      "loss": 0.4534,
+      "step": 14787
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2287546099109886e-06,
+      "loss": 0.4636,
+      "step": 14788
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.227467770828448e-06,
+      "loss": 0.4649,
+      "step": 14789
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226181004579472e-06,
+      "loss": 0.4501,
+      "step": 14790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.224894311188905e-06,
+      "loss": 0.488,
+      "step": 14791
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22360769068159e-06,
+      "loss": 0.4458,
+      "step": 14792
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.222321143082366e-06,
+      "loss": 0.4497,
+      "step": 14793
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22103466841607e-06,
+      "loss": 0.4466,
+      "step": 14794
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2197482667075405e-06,
+      "loss": 0.4695,
+      "step": 14795
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.218461937981615e-06,
+      "loss": 0.4558,
+      "step": 14796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2171756822631215e-06,
+      "loss": 0.471,
+      "step": 14797
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.215889499576898e-06,
+      "loss": 0.4613,
+      "step": 14798
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.214603389947776e-06,
+      "loss": 0.4672,
+      "step": 14799
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213317353400583e-06,
+      "loss": 0.4731,
+      "step": 14800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.212031389960149e-06,
+      "loss": 0.4529,
+      "step": 14801
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2107454996513e-06,
+      "loss": 0.4502,
+      "step": 14802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209459682498864e-06,
+      "loss": 0.4763,
+      "step": 14803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.208173938527664e-06,
+      "loss": 0.451,
+      "step": 14804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.206888267762522e-06,
+      "loss": 0.4343,
+      "step": 14805
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.205602670228261e-06,
+      "loss": 0.4669,
+      "step": 14806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.204317145949698e-06,
+      "loss": 0.4687,
+      "step": 14807
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.20303169495166e-06,
+      "loss": 0.4638,
+      "step": 14808
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2017463172589545e-06,
+      "loss": 0.4535,
+      "step": 14809
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200461012896401e-06,
+      "loss": 0.4357,
+      "step": 14810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199175781888815e-06,
+      "loss": 0.4509,
+      "step": 14811
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.197890624261012e-06,
+      "loss": 0.4748,
+      "step": 14812
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196605540037797e-06,
+      "loss": 0.4712,
+      "step": 14813
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1953205292439845e-06,
+      "loss": 0.4648,
+      "step": 14814
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194035591904382e-06,
+      "loss": 0.4415,
+      "step": 14815
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1927507280437995e-06,
+      "loss": 0.4735,
+      "step": 14816
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.191465937687038e-06,
+      "loss": 0.4545,
+      "step": 14817
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1901812208589055e-06,
+      "loss": 0.4712,
+      "step": 14818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.188896577584207e-06,
+      "loss": 0.4513,
+      "step": 14819
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.187612007887738e-06,
+      "loss": 0.4566,
+      "step": 14820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186327511794303e-06,
+      "loss": 0.46,
+      "step": 14821
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.185043089328701e-06,
+      "loss": 0.4802,
+      "step": 14822
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.183758740515731e-06,
+      "loss": 0.4825,
+      "step": 14823
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.182474465380183e-06,
+      "loss": 0.4505,
+      "step": 14824
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181190263946856e-06,
+      "loss": 0.4722,
+      "step": 14825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.179906136240541e-06,
+      "loss": 0.4709,
+      "step": 14826
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178622082286034e-06,
+      "loss": 0.4762,
+      "step": 14827
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.177338102108122e-06,
+      "loss": 0.4616,
+      "step": 14828
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176054195731593e-06,
+      "loss": 0.4811,
+      "step": 14829
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.174770363181235e-06,
+      "loss": 0.4483,
+      "step": 14830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173486604481838e-06,
+      "loss": 0.4556,
+      "step": 14831
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.172202919658181e-06,
+      "loss": 0.4472,
+      "step": 14832
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.170919308735048e-06,
+      "loss": 0.4536,
+      "step": 14833
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.169635771737222e-06,
+      "loss": 0.4928,
+      "step": 14834
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.168352308689489e-06,
+      "loss": 0.4544,
+      "step": 14835
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167068919616619e-06,
+      "loss": 0.4576,
+      "step": 14836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.165785604543393e-06,
+      "loss": 0.4481,
+      "step": 14837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.164502363494589e-06,
+      "loss": 0.4448,
+      "step": 14838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1632191964949776e-06,
+      "loss": 0.4846,
+      "step": 14839
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.161936103569333e-06,
+      "loss": 0.4427,
+      "step": 14840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.160653084742429e-06,
+      "loss": 0.4742,
+      "step": 14841
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159370140039036e-06,
+      "loss": 0.4597,
+      "step": 14842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.158087269483921e-06,
+      "loss": 0.4255,
+      "step": 14843
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156804473101852e-06,
+      "loss": 0.4863,
+      "step": 14844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155521750917593e-06,
+      "loss": 0.4539,
+      "step": 14845
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1542391029559165e-06,
+      "loss": 0.4802,
+      "step": 14846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.152956529241576e-06,
+      "loss": 0.4472,
+      "step": 14847
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151674029799337e-06,
+      "loss": 0.467,
+      "step": 14848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1503916046539595e-06,
+      "loss": 0.4447,
+      "step": 14849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149109253830208e-06,
+      "loss": 0.505,
+      "step": 14850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147826977352831e-06,
+      "loss": 0.4511,
+      "step": 14851
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1465447752465855e-06,
+      "loss": 0.4642,
+      "step": 14852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.145262647536232e-06,
+      "loss": 0.4748,
+      "step": 14853
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1439805942465235e-06,
+      "loss": 0.4397,
+      "step": 14854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.142698615402205e-06,
+      "loss": 0.4337,
+      "step": 14855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1414167110280295e-06,
+      "loss": 0.4669,
+      "step": 14856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.140134881148747e-06,
+      "loss": 0.4522,
+      "step": 14857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13885312578911e-06,
+      "loss": 0.1699,
+      "step": 14858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1375714449738555e-06,
+      "loss": 0.4356,
+      "step": 14859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.136289838727727e-06,
+      "loss": 0.4615,
+      "step": 14860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13500830707548e-06,
+      "loss": 0.4661,
+      "step": 14861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1337268500418434e-06,
+      "loss": 0.4705,
+      "step": 14862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.132445467651561e-06,
+      "loss": 0.4435,
+      "step": 14863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.131164159929375e-06,
+      "loss": 0.4398,
+      "step": 14864
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129882926900024e-06,
+      "loss": 0.4617,
+      "step": 14865
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1286017685882344e-06,
+      "loss": 0.4936,
+      "step": 14866
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.127320685018746e-06,
+      "loss": 0.4653,
+      "step": 14867
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126039676216293e-06,
+      "loss": 0.4784,
+      "step": 14868
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.124758742205612e-06,
+      "loss": 0.4501,
+      "step": 14869
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123477883011422e-06,
+      "loss": 0.4851,
+      "step": 14870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1221970986584545e-06,
+      "loss": 0.4661,
+      "step": 14871
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1209163891714405e-06,
+      "loss": 0.4455,
+      "step": 14872
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1196357545751105e-06,
+      "loss": 0.4208,
+      "step": 14873
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.118355194894174e-06,
+      "loss": 0.4443,
+      "step": 14874
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.117074710153366e-06,
+      "loss": 0.4889,
+      "step": 14875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.115794300377403e-06,
+      "loss": 0.4601,
+      "step": 14876
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1145139655910115e-06,
+      "loss": 0.4749,
+      "step": 14877
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.113233705818897e-06,
+      "loss": 0.4473,
+      "step": 14878
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.111953521085787e-06,
+      "loss": 0.4851,
+      "step": 14879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110673411416399e-06,
+      "loss": 0.4498,
+      "step": 14880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.109393376835436e-06,
+      "loss": 0.4338,
+      "step": 14881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.108113417367621e-06,
+      "loss": 0.4617,
+      "step": 14882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.106833533037659e-06,
+      "loss": 0.4631,
+      "step": 14883
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.105553723870267e-06,
+      "loss": 0.4644,
+      "step": 14884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1042739898901416e-06,
+      "loss": 0.4746,
+      "step": 14885
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.102994331122e-06,
+      "loss": 0.4639,
+      "step": 14886
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.101714747590544e-06,
+      "loss": 0.4483,
+      "step": 14887
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.100435239320482e-06,
+      "loss": 0.4298,
+      "step": 14888
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0991558063365075e-06,
+      "loss": 0.4325,
+      "step": 14889
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0978764486633265e-06,
+      "loss": 0.4622,
+      "step": 14890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.096597166325642e-06,
+      "loss": 0.4357,
+      "step": 14891
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.095317959348148e-06,
+      "loss": 0.4618,
+      "step": 14892
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0940388277555415e-06,
+      "loss": 0.4585,
+      "step": 14893
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092759771572518e-06,
+      "loss": 0.4356,
+      "step": 14894
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.091480790823772e-06,
+      "loss": 0.4628,
+      "step": 14895
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.090201885533997e-06,
+      "loss": 0.4452,
+      "step": 14896
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.088923055727881e-06,
+      "loss": 0.5077,
+      "step": 14897
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087644301430115e-06,
+      "loss": 0.4359,
+      "step": 14898
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.086365622665387e-06,
+      "loss": 0.4629,
+      "step": 14899
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0850870194583854e-06,
+      "loss": 0.4671,
+      "step": 14900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.083808491833792e-06,
+      "loss": 0.4864,
+      "step": 14901
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08253003981629e-06,
+      "loss": 0.4708,
+      "step": 14902
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081251663430567e-06,
+      "loss": 0.4582,
+      "step": 14903
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.079973362701298e-06,
+      "loss": 0.4516,
+      "step": 14904
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078695137653164e-06,
+      "loss": 0.4648,
+      "step": 14905
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.077416988310842e-06,
+      "loss": 0.4725,
+      "step": 14906
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076138914699013e-06,
+      "loss": 0.4589,
+      "step": 14907
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.074860916842346e-06,
+      "loss": 0.4577,
+      "step": 14908
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.073582994765516e-06,
+      "loss": 0.469,
+      "step": 14909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072305148493195e-06,
+      "loss": 0.4478,
+      "step": 14910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.071027378050056e-06,
+      "loss": 0.4729,
+      "step": 14911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.069749683460765e-06,
+      "loss": 0.4479,
+      "step": 14912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068472064749991e-06,
+      "loss": 0.4405,
+      "step": 14913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.067194521942397e-06,
+      "loss": 0.4657,
+      "step": 14914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065917055062654e-06,
+      "loss": 0.4993,
+      "step": 14915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064639664135419e-06,
+      "loss": 0.4556,
+      "step": 14916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063362349185355e-06,
+      "loss": 0.4527,
+      "step": 14917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062085110237125e-06,
+      "loss": 0.4677,
+      "step": 14918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0608079473153866e-06,
+      "loss": 0.4425,
+      "step": 14919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0595308604447955e-06,
+      "loss": 0.479,
+      "step": 14920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058253849650006e-06,
+      "loss": 0.4595,
+      "step": 14921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.056976914955679e-06,
+      "loss": 0.4672,
+      "step": 14922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.055700056386461e-06,
+      "loss": 0.4845,
+      "step": 14923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.054423273967005e-06,
+      "loss": 0.4637,
+      "step": 14924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0531465677219615e-06,
+      "loss": 0.4656,
+      "step": 14925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051869937675981e-06,
+      "loss": 0.4471,
+      "step": 14926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.050593383853707e-06,
+      "loss": 0.4811,
+      "step": 14927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.049316906279785e-06,
+      "loss": 0.4324,
+      "step": 14928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.048040504978861e-06,
+      "loss": 0.4862,
+      "step": 14929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0467641799755796e-06,
+      "loss": 0.4783,
+      "step": 14930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0454879312945755e-06,
+      "loss": 0.4794,
+      "step": 14931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.044211758960494e-06,
+      "loss": 0.4522,
+      "step": 14932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.042935662997971e-06,
+      "loss": 0.4905,
+      "step": 14933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.041659643431646e-06,
+      "loss": 0.4714,
+      "step": 14934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.040383700286148e-06,
+      "loss": 0.4778,
+      "step": 14935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.039107833586115e-06,
+      "loss": 0.4639,
+      "step": 14936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0378320433561785e-06,
+      "loss": 0.485,
+      "step": 14937
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.036556329620973e-06,
+      "loss": 0.4534,
+      "step": 14938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.03528069240512e-06,
+      "loss": 0.4271,
+      "step": 14939
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.034005131733253e-06,
+      "loss": 0.4514,
+      "step": 14940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0327296476299955e-06,
+      "loss": 0.463,
+      "step": 14941
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031454240119977e-06,
+      "loss": 0.4392,
+      "step": 14942
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.030178909227815e-06,
+      "loss": 0.4618,
+      "step": 14943
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.028903654978134e-06,
+      "loss": 0.4691,
+      "step": 14944
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027628477395557e-06,
+      "loss": 0.3927,
+      "step": 14945
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.026353376504698e-06,
+      "loss": 0.1865,
+      "step": 14946
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0250783523301755e-06,
+      "loss": 0.4522,
+      "step": 14947
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.023803404896609e-06,
+      "loss": 0.4499,
+      "step": 14948
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.022528534228611e-06,
+      "loss": 0.4476,
+      "step": 14949
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.021253740350793e-06,
+      "loss": 0.4933,
+      "step": 14950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.019979023287767e-06,
+      "loss": 0.4732,
+      "step": 14951
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.018704383064144e-06,
+      "loss": 0.4502,
+      "step": 14952
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.017429819704537e-06,
+      "loss": 0.4669,
+      "step": 14953
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0161553332335445e-06,
+      "loss": 0.4502,
+      "step": 14954
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014880923675774e-06,
+      "loss": 0.4272,
+      "step": 14955
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.013606591055834e-06,
+      "loss": 0.5015,
+      "step": 14956
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0123323353983275e-06,
+      "loss": 0.4703,
+      "step": 14957
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.011058156727849e-06,
+      "loss": 0.4278,
+      "step": 14958
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0097840550690034e-06,
+      "loss": 0.4829,
+      "step": 14959
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.008510030446385e-06,
+      "loss": 0.4515,
+      "step": 14960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007236082884599e-06,
+      "loss": 0.4579,
+      "step": 14961
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.00596221240823e-06,
+      "loss": 0.4567,
+      "step": 14962
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.004688419041877e-06,
+      "loss": 0.4555,
+      "step": 14963
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0034147028101304e-06,
+      "loss": 0.4721,
+      "step": 14964
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.002141063737588e-06,
+      "loss": 0.4694,
+      "step": 14965
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.000867501848828e-06,
+      "loss": 0.4616,
+      "step": 14966
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9995940171684434e-06,
+      "loss": 0.4622,
+      "step": 14967
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.998320609721022e-06,
+      "loss": 0.4508,
+      "step": 14968
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.997047279531145e-06,
+      "loss": 0.46,
+      "step": 14969
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9957740266233975e-06,
+      "loss": 0.4705,
+      "step": 14970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99450085102236e-06,
+      "loss": 0.5071,
+      "step": 14971
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99322775275262e-06,
+      "loss": 0.4507,
+      "step": 14972
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.991954731838746e-06,
+      "loss": 0.4627,
+      "step": 14973
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9906817883053195e-06,
+      "loss": 0.4804,
+      "step": 14974
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989408922176916e-06,
+      "loss": 0.4777,
+      "step": 14975
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.988136133478114e-06,
+      "loss": 0.4517,
+      "step": 14976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9868634222334794e-06,
+      "loss": 0.4469,
+      "step": 14977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9855907884675876e-06,
+      "loss": 0.1805,
+      "step": 14978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9843182322050045e-06,
+      "loss": 0.4468,
+      "step": 14979
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.467,
+      "step": 14980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981773352288054e-06,
+      "loss": 0.1585,
+      "step": 14981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.98050102868281e-06,
+      "loss": 0.4469,
+      "step": 14982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979228782679145e-06,
+      "loss": 0.449,
+      "step": 14983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977956614301623e-06,
+      "loss": 0.4574,
+      "step": 14984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.976684523574795e-06,
+      "loss": 0.4448,
+      "step": 14985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.975412510523224e-06,
+      "loss": 0.456,
+      "step": 14986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9741405751714755e-06,
+      "loss": 0.4477,
+      "step": 14987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.972868717544096e-06,
+      "loss": 0.4582,
+      "step": 14988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.971596937665642e-06,
+      "loss": 0.4596,
+      "step": 14989
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.97032523556067e-06,
+      "loss": 0.4999,
+      "step": 14990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.969053611253735e-06,
+      "loss": 0.4536,
+      "step": 14991
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96778206476938e-06,
+      "loss": 0.4713,
+      "step": 14992
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.966510596132156e-06,
+      "loss": 0.4741,
+      "step": 14993
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965239205366611e-06,
+      "loss": 0.4648,
+      "step": 14994
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9639678924972956e-06,
+      "loss": 0.4548,
+      "step": 14995
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.962696657548747e-06,
+      "loss": 0.4486,
+      "step": 14996
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961425500545508e-06,
+      "loss": 0.4443,
+      "step": 14997
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.960154421512124e-06,
+      "loss": 0.4754,
+      "step": 14998
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9588834204731384e-06,
+      "loss": 0.4526,
+      "step": 14999
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.957612497453078e-06,
+      "loss": 0.4715,
+      "step": 15000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956341652476487e-06,
+      "loss": 0.4369,
+      "step": 15001
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955070885567902e-06,
+      "loss": 0.4358,
+      "step": 15002
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953800196751857e-06,
+      "loss": 0.4751,
+      "step": 15003
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.952529586052876e-06,
+      "loss": 0.4408,
+      "step": 15004
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.951259053495497e-06,
+      "loss": 0.4536,
+      "step": 15005
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.949988599104249e-06,
+      "loss": 0.4675,
+      "step": 15006
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.948718222903662e-06,
+      "loss": 0.4286,
+      "step": 15007
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947447924918254e-06,
+      "loss": 0.4862,
+      "step": 15008
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.946177705172556e-06,
+      "loss": 0.4801,
+      "step": 15009
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944907563691093e-06,
+      "loss": 0.4487,
+      "step": 15010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.943637500498378e-06,
+      "loss": 0.4628,
+      "step": 15011
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942367515618939e-06,
+      "loss": 0.4323,
+      "step": 15012
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9410976090772925e-06,
+      "loss": 0.4565,
+      "step": 15013
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.93982778089796e-06,
+      "loss": 0.4302,
+      "step": 15014
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9385580311054455e-06,
+      "loss": 0.4951,
+      "step": 15015
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.937288359724272e-06,
+      "loss": 0.4804,
+      "step": 15016
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.936018766778953e-06,
+      "loss": 0.4027,
+      "step": 15017
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.934749252293996e-06,
+      "loss": 0.4747,
+      "step": 15018
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9334798162939125e-06,
+      "loss": 0.4728,
+      "step": 15019
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932210458803207e-06,
+      "loss": 0.4346,
+      "step": 15020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9309411798463916e-06,
+      "loss": 0.5022,
+      "step": 15021
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.929671979447968e-06,
+      "loss": 0.4576,
+      "step": 15022
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.92840285763244e-06,
+      "loss": 0.4846,
+      "step": 15023
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.927133814424311e-06,
+      "loss": 0.4874,
+      "step": 15024
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.92586484984808e-06,
+      "loss": 0.4703,
+      "step": 15025
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.924595963928248e-06,
+      "loss": 0.4726,
+      "step": 15026
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923327156689308e-06,
+      "loss": 0.4428,
+      "step": 15027
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.92205842815576e-06,
+      "loss": 0.4343,
+      "step": 15028
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.920789778352098e-06,
+      "loss": 0.4501,
+      "step": 15029
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9195212073028145e-06,
+      "loss": 0.4298,
+      "step": 15030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9182527150324e-06,
+      "loss": 0.4472,
+      "step": 15031
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916984301565345e-06,
+      "loss": 0.4677,
+      "step": 15032
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.915715966926139e-06,
+      "loss": 0.4474,
+      "step": 15033
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.914447711139266e-06,
+      "loss": 0.4719,
+      "step": 15034
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.913179534229213e-06,
+      "loss": 0.4751,
+      "step": 15035
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.911911436220463e-06,
+      "loss": 0.4417,
+      "step": 15036
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910643417137502e-06,
+      "loss": 0.4329,
+      "step": 15037
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9093754770048065e-06,
+      "loss": 0.483,
+      "step": 15038
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.908107615846856e-06,
+      "loss": 0.443,
+      "step": 15039
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9068398336881295e-06,
+      "loss": 0.4795,
+      "step": 15040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.905572130553103e-06,
+      "loss": 0.4632,
+      "step": 15041
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.904304506466252e-06,
+      "loss": 0.4694,
+      "step": 15042
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.903036961452047e-06,
+      "loss": 0.4648,
+      "step": 15043
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.901769495534961e-06,
+      "loss": 0.4574,
+      "step": 15044
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900502108739466e-06,
+      "loss": 0.4539,
+      "step": 15045
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.899234801090027e-06,
+      "loss": 0.4628,
+      "step": 15046
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897967572611113e-06,
+      "loss": 0.4706,
+      "step": 15047
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.896700423327189e-06,
+      "loss": 0.4678,
+      "step": 15048
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8954333532627205e-06,
+      "loss": 0.4763,
+      "step": 15049
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.894166362442167e-06,
+      "loss": 0.4455,
+      "step": 15050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8928994508899905e-06,
+      "loss": 0.4707,
+      "step": 15051
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891632618630654e-06,
+      "loss": 0.4988,
+      "step": 15052
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8903658656886095e-06,
+      "loss": 0.4809,
+      "step": 15053
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.889099192088315e-06,
+      "loss": 0.4537,
+      "step": 15054
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887832597854226e-06,
+      "loss": 0.4485,
+      "step": 15055
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.886566083010799e-06,
+      "loss": 0.4348,
+      "step": 15056
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885299647582481e-06,
+      "loss": 0.4505,
+      "step": 15057
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.884033291593722e-06,
+      "loss": 0.4546,
+      "step": 15058
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.882767015068973e-06,
+      "loss": 0.4436,
+      "step": 15059
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8815008180326825e-06,
+      "loss": 0.4708,
+      "step": 15060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.880234700509293e-06,
+      "loss": 0.4819,
+      "step": 15061
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.878968662523248e-06,
+      "loss": 0.4982,
+      "step": 15062
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.877702704098991e-06,
+      "loss": 0.4444,
+      "step": 15063
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.876436825260967e-06,
+      "loss": 0.4274,
+      "step": 15064
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.875171026033609e-06,
+      "loss": 0.4488,
+      "step": 15065
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873905306441358e-06,
+      "loss": 0.4373,
+      "step": 15066
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87263966650865e-06,
+      "loss": 0.4437,
+      "step": 15067
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8713741062599215e-06,
+      "loss": 0.4673,
+      "step": 15068
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.870108625719601e-06,
+      "loss": 0.4671,
+      "step": 15069
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868843224912126e-06,
+      "loss": 0.4614,
+      "step": 15070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.867577903861925e-06,
+      "loss": 0.4943,
+      "step": 15071
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866312662593423e-06,
+      "loss": 0.4373,
+      "step": 15072
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8650475011310496e-06,
+      "loss": 0.4434,
+      "step": 15073
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8637824194992306e-06,
+      "loss": 0.4489,
+      "step": 15074
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.862517417722393e-06,
+      "loss": 0.4887,
+      "step": 15075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861252495824954e-06,
+      "loss": 0.4405,
+      "step": 15076
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859987653831336e-06,
+      "loss": 0.4608,
+      "step": 15077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.858722891765959e-06,
+      "loss": 0.4934,
+      "step": 15078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.857458209653244e-06,
+      "loss": 0.4575,
+      "step": 15079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856193607517602e-06,
+      "loss": 0.4996,
+      "step": 15080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.854929085383449e-06,
+      "loss": 0.4347,
+      "step": 15081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8536646432751994e-06,
+      "loss": 0.4787,
+      "step": 15082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852400281217268e-06,
+      "loss": 0.4624,
+      "step": 15083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8511359992340585e-06,
+      "loss": 0.4449,
+      "step": 15084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8498717973499845e-06,
+      "loss": 0.468,
+      "step": 15085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.848607675589447e-06,
+      "loss": 0.4393,
+      "step": 15086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.847343633976863e-06,
+      "loss": 0.4581,
+      "step": 15087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846079672536624e-06,
+      "loss": 0.4416,
+      "step": 15088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.844815791293139e-06,
+      "loss": 0.4563,
+      "step": 15089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843551990270805e-06,
+      "loss": 0.4481,
+      "step": 15090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.84228826949403e-06,
+      "loss": 0.452,
+      "step": 15091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8410246289872e-06,
+      "loss": 0.4292,
+      "step": 15092
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8397610687747186e-06,
+      "loss": 0.4396,
+      "step": 15093
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.838497588880977e-06,
+      "loss": 0.4531,
+      "step": 15094
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.83723418933037e-06,
+      "loss": 0.4563,
+      "step": 15095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.83597087014729e-06,
+      "loss": 0.4747,
+      "step": 15096
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8347076313561245e-06,
+      "loss": 0.4535,
+      "step": 15097
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.833444472981268e-06,
+      "loss": 0.4471,
+      "step": 15098
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832181395047099e-06,
+      "loss": 0.4736,
+      "step": 15099
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830918397578005e-06,
+      "loss": 0.4706,
+      "step": 15100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.829655480598373e-06,
+      "loss": 0.4289,
+      "step": 15101
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8283926441325835e-06,
+      "loss": 0.4563,
+      "step": 15102
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827129888205016e-06,
+      "loss": 0.4638,
+      "step": 15103
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825867212840053e-06,
+      "loss": 0.4887,
+      "step": 15104
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.824604618062069e-06,
+      "loss": 0.4681,
+      "step": 15105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.823342103895444e-06,
+      "loss": 0.4815,
+      "step": 15106
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.822079670364547e-06,
+      "loss": 0.459,
+      "step": 15107
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.820817317493752e-06,
+      "loss": 0.4793,
+      "step": 15108
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.819555045307432e-06,
+      "loss": 0.4622,
+      "step": 15109
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.818292853829957e-06,
+      "loss": 0.4495,
+      "step": 15110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.817030743085694e-06,
+      "loss": 0.4521,
+      "step": 15111
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815768713099011e-06,
+      "loss": 0.4269,
+      "step": 15112
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.814506763894272e-06,
+      "loss": 0.486,
+      "step": 15113
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.813244895495845e-06,
+      "loss": 0.4555,
+      "step": 15114
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.811983107928085e-06,
+      "loss": 0.4752,
+      "step": 15115
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.810721401215353e-06,
+      "loss": 0.4802,
+      "step": 15116
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.809459775382014e-06,
+      "loss": 0.4426,
+      "step": 15117
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8081982304524175e-06,
+      "loss": 0.4825,
+      "step": 15118
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806936766450926e-06,
+      "loss": 0.4961,
+      "step": 15119
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8056753834018895e-06,
+      "loss": 0.5048,
+      "step": 15120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.804414081329668e-06,
+      "loss": 0.437,
+      "step": 15121
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8031528602586e-06,
+      "loss": 0.4305,
+      "step": 15122
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801891720213044e-06,
+      "loss": 0.4585,
+      "step": 15123
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.800630661217344e-06,
+      "loss": 0.4587,
+      "step": 15124
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.799369683295848e-06,
+      "loss": 0.429,
+      "step": 15125
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.798108786472901e-06,
+      "loss": 0.5018,
+      "step": 15126
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796847970772847e-06,
+      "loss": 0.4746,
+      "step": 15127
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795587236220026e-06,
+      "loss": 0.447,
+      "step": 15128
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.794326582838782e-06,
+      "loss": 0.4571,
+      "step": 15129
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793066010653447e-06,
+      "loss": 0.442,
+      "step": 15130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.79180551968836e-06,
+      "loss": 0.4702,
+      "step": 15131
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.790545109967858e-06,
+      "loss": 0.4558,
+      "step": 15132
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7892847815162754e-06,
+      "loss": 0.4868,
+      "step": 15133
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788024534357943e-06,
+      "loss": 0.4801,
+      "step": 15134
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786764368517191e-06,
+      "loss": 0.4733,
+      "step": 15135
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.785504284018355e-06,
+      "loss": 0.4614,
+      "step": 15136
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784244280885753e-06,
+      "loss": 0.4863,
+      "step": 15137
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.782984359143713e-06,
+      "loss": 0.45,
+      "step": 15138
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.781724518816563e-06,
+      "loss": 0.4827,
+      "step": 15139
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.780464759928623e-06,
+      "loss": 0.4204,
+      "step": 15140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.779205082504218e-06,
+      "loss": 0.4717,
+      "step": 15141
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.777945486567662e-06,
+      "loss": 0.4585,
+      "step": 15142
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.776685972143279e-06,
+      "loss": 0.4639,
+      "step": 15143
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.775426539255384e-06,
+      "loss": 0.4488,
+      "step": 15144
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77416718792829e-06,
+      "loss": 0.4942,
+      "step": 15145
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.772907918186309e-06,
+      "loss": 0.4589,
+      "step": 15146
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.771648730053756e-06,
+      "loss": 0.4579,
+      "step": 15147
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7703896235549396e-06,
+      "loss": 0.4837,
+      "step": 15148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.769130598714169e-06,
+      "loss": 0.4505,
+      "step": 15149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7678716555557515e-06,
+      "loss": 0.4716,
+      "step": 15150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7666127941039915e-06,
+      "loss": 0.4608,
+      "step": 15151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7653540143831996e-06,
+      "loss": 0.4652,
+      "step": 15152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7640953164176685e-06,
+      "loss": 0.4583,
+      "step": 15153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.762836700231702e-06,
+      "loss": 0.4713,
+      "step": 15154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761578165849596e-06,
+      "loss": 0.4529,
+      "step": 15155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7603197132956615e-06,
+      "loss": 0.4762,
+      "step": 15156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759061342594181e-06,
+      "loss": 0.4704,
+      "step": 15157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.757803053769453e-06,
+      "loss": 0.4714,
+      "step": 15158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.756544846845775e-06,
+      "loss": 0.4602,
+      "step": 15159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7552867218474296e-06,
+      "loss": 0.4728,
+      "step": 15160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.754028678798711e-06,
+      "loss": 0.4688,
+      "step": 15161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.752770717723904e-06,
+      "loss": 0.4462,
+      "step": 15162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.751512838647306e-06,
+      "loss": 0.4421,
+      "step": 15163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.75025504159319e-06,
+      "loss": 0.4369,
+      "step": 15164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748997326585844e-06,
+      "loss": 0.4301,
+      "step": 15165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.74773969364955e-06,
+      "loss": 0.4585,
+      "step": 15166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746482142808592e-06,
+      "loss": 0.4781,
+      "step": 15167
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.745224674087239e-06,
+      "loss": 0.4607,
+      "step": 15168
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.743967287509775e-06,
+      "loss": 0.445,
+      "step": 15169
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7427099831004695e-06,
+      "loss": 0.4398,
+      "step": 15170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7414527608836076e-06,
+      "loss": 0.185,
+      "step": 15171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740195620883453e-06,
+      "loss": 0.4844,
+      "step": 15172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.738938563124276e-06,
+      "loss": 0.4528,
+      "step": 15173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73768158763035e-06,
+      "loss": 0.458,
+      "step": 15174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736424694425943e-06,
+      "loss": 0.4277,
+      "step": 15175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735167883535314e-06,
+      "loss": 0.4935,
+      "step": 15176
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.733911154982728e-06,
+      "loss": 0.4629,
+      "step": 15177
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.732654508792461e-06,
+      "loss": 0.4784,
+      "step": 15178
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.731397944988759e-06,
+      "loss": 0.461,
+      "step": 15179
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.730141463595887e-06,
+      "loss": 0.4812,
+      "step": 15180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.728885064638104e-06,
+      "loss": 0.4591,
+      "step": 15181
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7276287481396695e-06,
+      "loss": 0.4439,
+      "step": 15182
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.726372514124831e-06,
+      "loss": 0.4303,
+      "step": 15183
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.72511636261784e-06,
+      "loss": 0.4535,
+      "step": 15184
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7238602936429575e-06,
+      "loss": 0.4462,
+      "step": 15185
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.722604307224433e-06,
+      "loss": 0.4337,
+      "step": 15186
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.721348403386506e-06,
+      "loss": 0.4369,
+      "step": 15187
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.72009258215343e-06,
+      "loss": 0.4609,
+      "step": 15188
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.718836843549449e-06,
+      "loss": 0.487,
+      "step": 15189
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.71758118759881e-06,
+      "loss": 0.4331,
+      "step": 15190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.716325614325743e-06,
+      "loss": 0.1762,
+      "step": 15191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7150701237545e-06,
+      "loss": 0.473,
+      "step": 15192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.713814715909318e-06,
+      "loss": 0.4656,
+      "step": 15193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.712559390814436e-06,
+      "loss": 0.4465,
+      "step": 15194
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.711304148494084e-06,
+      "loss": 0.4494,
+      "step": 15195
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.710048988972499e-06,
+      "loss": 0.4488,
+      "step": 15196
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.708793912273911e-06,
+      "loss": 0.4724,
+      "step": 15197
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7075389184225576e-06,
+      "loss": 0.4481,
+      "step": 15198
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.706284007442656e-06,
+      "loss": 0.4695,
+      "step": 15199
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.705029179358445e-06,
+      "loss": 0.4635,
+      "step": 15200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7037744341941515e-06,
+      "loss": 0.4464,
+      "step": 15201
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.702519771973991e-06,
+      "loss": 0.4571,
+      "step": 15202
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70126519272219e-06,
+      "loss": 0.4611,
+      "step": 15203
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.700010696462972e-06,
+      "loss": 0.4337,
+      "step": 15204
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698756283220559e-06,
+      "loss": 0.462,
+      "step": 15205
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.697501953019156e-06,
+      "loss": 0.4322,
+      "step": 15206
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.696247705882993e-06,
+      "loss": 0.4551,
+      "step": 15207
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694993541836282e-06,
+      "loss": 0.4802,
+      "step": 15208
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6937394609032385e-06,
+      "loss": 0.4469,
+      "step": 15209
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692485463108066e-06,
+      "loss": 0.4744,
+      "step": 15210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.691231548474978e-06,
+      "loss": 0.4807,
+      "step": 15211
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689977717028186e-06,
+      "loss": 0.4762,
+      "step": 15212
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688723968791894e-06,
+      "loss": 0.4652,
+      "step": 15213
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687470303790307e-06,
+      "loss": 0.4547,
+      "step": 15214
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6862167220476305e-06,
+      "loss": 0.4579,
+      "step": 15215
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.684963223588064e-06,
+      "loss": 0.4832,
+      "step": 15216
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683709808435816e-06,
+      "loss": 0.4672,
+      "step": 15217
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6824564766150724e-06,
+      "loss": 0.461,
+      "step": 15218
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6812032281500385e-06,
+      "loss": 0.4707,
+      "step": 15219
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679950063064906e-06,
+      "loss": 0.4542,
+      "step": 15220
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.678696981383873e-06,
+      "loss": 0.4662,
+      "step": 15221
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.677443983131129e-06,
+      "loss": 0.4809,
+      "step": 15222
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.676191068330865e-06,
+      "loss": 0.4623,
+      "step": 15223
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6749382370072745e-06,
+      "loss": 0.4668,
+      "step": 15224
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6736854891845394e-06,
+      "loss": 0.4729,
+      "step": 15225
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.672432824886846e-06,
+      "loss": 0.4514,
+      "step": 15226
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.671180244138379e-06,
+      "loss": 0.4373,
+      "step": 15227
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.669927746963323e-06,
+      "loss": 0.4497,
+      "step": 15228
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.668675333385858e-06,
+      "loss": 0.4966,
+      "step": 15229
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667423003430164e-06,
+      "loss": 0.4682,
+      "step": 15230
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.666170757120417e-06,
+      "loss": 0.4457,
+      "step": 15231
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6649185944808e-06,
+      "loss": 0.4566,
+      "step": 15232
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6636665155354776e-06,
+      "loss": 0.4526,
+      "step": 15233
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.662414520308627e-06,
+      "loss": 0.4406,
+      "step": 15234
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.442,
+      "step": 15235
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.659910781107027e-06,
+      "loss": 0.4649,
+      "step": 15236
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.658659037180615e-06,
+      "loss": 0.4531,
+      "step": 15237
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65740737706935e-06,
+      "loss": 0.4292,
+      "step": 15238
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6561558007974e-06,
+      "loss": 0.5109,
+      "step": 15239
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6549043083889285e-06,
+      "loss": 0.4883,
+      "step": 15240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.653652899868092e-06,
+      "loss": 0.452,
+      "step": 15241
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.652401575259052e-06,
+      "loss": 0.4737,
+      "step": 15242
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651150334585969e-06,
+      "loss": 0.4691,
+      "step": 15243
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.649899177872999e-06,
+      "loss": 0.4868,
+      "step": 15244
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648648105144299e-06,
+      "loss": 0.4323,
+      "step": 15245
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.647397116424018e-06,
+      "loss": 0.4586,
+      "step": 15246
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6461462117363165e-06,
+      "loss": 0.4398,
+      "step": 15247
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.644895391105334e-06,
+      "loss": 0.474,
+      "step": 15248
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.643644654555225e-06,
+      "loss": 0.4775,
+      "step": 15249
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642394002110137e-06,
+      "loss": 0.474,
+      "step": 15250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6411434337942115e-06,
+      "loss": 0.4707,
+      "step": 15251
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639892949631597e-06,
+      "loss": 0.4178,
+      "step": 15252
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638642549646433e-06,
+      "loss": 0.4705,
+      "step": 15253
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.637392233862862e-06,
+      "loss": 0.4348,
+      "step": 15254
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.636142002305025e-06,
+      "loss": 0.4782,
+      "step": 15255
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.634891854997053e-06,
+      "loss": 0.4665,
+      "step": 15256
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6336417919630835e-06,
+      "loss": 0.4609,
+      "step": 15257
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632391813227253e-06,
+      "loss": 0.4455,
+      "step": 15258
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.631141918813693e-06,
+      "loss": 0.464,
+      "step": 15259
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629892108746533e-06,
+      "loss": 0.467,
+      "step": 15260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6286423830499045e-06,
+      "loss": 0.4677,
+      "step": 15261
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.627392741747934e-06,
+      "loss": 0.4489,
+      "step": 15262
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.62614318486475e-06,
+      "loss": 0.4738,
+      "step": 15263
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6248937124244704e-06,
+      "loss": 0.4561,
+      "step": 15264
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.623644324451223e-06,
+      "loss": 0.4591,
+      "step": 15265
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622395020969127e-06,
+      "loss": 0.4498,
+      "step": 15266
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.621145802002302e-06,
+      "loss": 0.4789,
+      "step": 15267
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.619896667574865e-06,
+      "loss": 0.4705,
+      "step": 15268
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.618647617710935e-06,
+      "loss": 0.4495,
+      "step": 15269
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.617398652434628e-06,
+      "loss": 0.4719,
+      "step": 15270
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61614977177005e-06,
+      "loss": 0.4485,
+      "step": 15271
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.614900975741314e-06,
+      "loss": 0.4772,
+      "step": 15272
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.613652264372533e-06,
+      "loss": 0.4681,
+      "step": 15273
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.612403637687813e-06,
+      "loss": 0.4469,
+      "step": 15274
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61115509571126e-06,
+      "loss": 0.4701,
+      "step": 15275
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6099066384669794e-06,
+      "loss": 0.4744,
+      "step": 15276
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.608658265979076e-06,
+      "loss": 0.4586,
+      "step": 15277
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.607409978271651e-06,
+      "loss": 0.4388,
+      "step": 15278
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.606161775368801e-06,
+      "loss": 0.4684,
+      "step": 15279
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604913657294625e-06,
+      "loss": 0.4537,
+      "step": 15280
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.603665624073216e-06,
+      "loss": 0.4491,
+      "step": 15281
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.602417675728682e-06,
+      "loss": 0.4494,
+      "step": 15282
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.601169812285103e-06,
+      "loss": 0.451,
+      "step": 15283
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599922033766575e-06,
+      "loss": 0.4305,
+      "step": 15284
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598674340197192e-06,
+      "loss": 0.475,
+      "step": 15285
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.597426731601035e-06,
+      "loss": 0.4422,
+      "step": 15286
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.596179208002193e-06,
+      "loss": 0.4742,
+      "step": 15287
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.594931769424749e-06,
+      "loss": 0.4955,
+      "step": 15288
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5936844158927975e-06,
+      "loss": 0.477,
+      "step": 15289
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592437147430407e-06,
+      "loss": 0.4337,
+      "step": 15290
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.591189964061664e-06,
+      "loss": 0.4419,
+      "step": 15291
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.589942865810646e-06,
+      "loss": 0.4307,
+      "step": 15292
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588695852701433e-06,
+      "loss": 0.457,
+      "step": 15293
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.587448924758093e-06,
+      "loss": 0.4934,
+      "step": 15294
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586202082004703e-06,
+      "loss": 0.4784,
+      "step": 15295
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.584955324465333e-06,
+      "loss": 0.4571,
+      "step": 15296
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.583708652164062e-06,
+      "loss": 0.4594,
+      "step": 15297
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.582462065124949e-06,
+      "loss": 0.4706,
+      "step": 15298
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.581215563372063e-06,
+      "loss": 0.4413,
+      "step": 15299
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579969146929472e-06,
+      "loss": 0.4636,
+      "step": 15300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.578722815821241e-06,
+      "loss": 0.468,
+      "step": 15301
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.577476570071425e-06,
+      "loss": 0.4603,
+      "step": 15302
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576230409704084e-06,
+      "loss": 0.4534,
+      "step": 15303
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.574984334743287e-06,
+      "loss": 0.4956,
+      "step": 15304
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573738345213087e-06,
+      "loss": 0.4581,
+      "step": 15305
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.572492441137536e-06,
+      "loss": 0.4381,
+      "step": 15306
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.571246622540688e-06,
+      "loss": 0.455,
+      "step": 15307
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570000889446601e-06,
+      "loss": 0.4555,
+      "step": 15308
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.568755241879318e-06,
+      "loss": 0.4523,
+      "step": 15309
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567509679862886e-06,
+      "loss": 0.4585,
+      "step": 15310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.566264203421361e-06,
+      "loss": 0.4421,
+      "step": 15311
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565018812578789e-06,
+      "loss": 0.4906,
+      "step": 15312
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.563773507359204e-06,
+      "loss": 0.4545,
+      "step": 15313
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.562528287786655e-06,
+      "loss": 0.4761,
+      "step": 15314
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5612831538851806e-06,
+      "loss": 0.4672,
+      "step": 15315
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.560038105678825e-06,
+      "loss": 0.4725,
+      "step": 15316
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.558793143191612e-06,
+      "loss": 0.4825,
+      "step": 15317
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55754826644759e-06,
+      "loss": 0.455,
+      "step": 15318
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55630347547079e-06,
+      "loss": 0.4494,
+      "step": 15319
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5550587702852465e-06,
+      "loss": 0.5319,
+      "step": 15320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.553814150914983e-06,
+      "loss": 0.455,
+      "step": 15321
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552569617384031e-06,
+      "loss": 0.4501,
+      "step": 15322
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.551325169716422e-06,
+      "loss": 0.4604,
+      "step": 15323
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55008080793618e-06,
+      "loss": 0.462,
+      "step": 15324
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548836532067321e-06,
+      "loss": 0.4695,
+      "step": 15325
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.547592342133877e-06,
+      "loss": 0.1609,
+      "step": 15326
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.546348238159871e-06,
+      "loss": 0.4692,
+      "step": 15327
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545104220169313e-06,
+      "loss": 0.4608,
+      "step": 15328
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543860288186222e-06,
+      "loss": 0.4642,
+      "step": 15329
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 0.4602,
+      "step": 15330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541372682338517e-06,
+      "loss": 0.4939,
+      "step": 15331
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.540129008521919e-06,
+      "loss": 0.4576,
+      "step": 15332
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538885420808846e-06,
+      "loss": 0.4613,
+      "step": 15333
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.537641919223304e-06,
+      "loss": 0.4553,
+      "step": 15334
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.536398503789305e-06,
+      "loss": 0.4714,
+      "step": 15335
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.535155174530846e-06,
+      "loss": 0.4652,
+      "step": 15336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.533911931471936e-06,
+      "loss": 0.4508,
+      "step": 15337
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532668774636576e-06,
+      "loss": 0.4578,
+      "step": 15338
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.531425704048767e-06,
+      "loss": 0.4551,
+      "step": 15339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.530182719732509e-06,
+      "loss": 0.4369,
+      "step": 15340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528939821711798e-06,
+      "loss": 0.4447,
+      "step": 15341
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.527697010010633e-06,
+      "loss": 0.4098,
+      "step": 15342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526454284653007e-06,
+      "loss": 0.4552,
+      "step": 15343
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.525211645662909e-06,
+      "loss": 0.4526,
+      "step": 15344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52396909306433e-06,
+      "loss": 0.4741,
+      "step": 15345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.522726626881262e-06,
+      "loss": 0.4803,
+      "step": 15346
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.521484247137691e-06,
+      "loss": 0.4283,
+      "step": 15347
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.520241953857602e-06,
+      "loss": 0.4764,
+      "step": 15348
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.518999747064981e-06,
+      "loss": 0.477,
+      "step": 15349
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517757626783813e-06,
+      "loss": 0.4724,
+      "step": 15350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.516515593038072e-06,
+      "loss": 0.4403,
+      "step": 15351
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5152736458517394e-06,
+      "loss": 0.4555,
+      "step": 15352
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514031785248794e-06,
+      "loss": 0.4576,
+      "step": 15353
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.512790011253211e-06,
+      "loss": 0.4549,
+      "step": 15354
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511548323888964e-06,
+      "loss": 0.4658,
+      "step": 15355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510306723180025e-06,
+      "loss": 0.4515,
+      "step": 15356
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.509065209150367e-06,
+      "loss": 0.4583,
+      "step": 15357
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5078237818239596e-06,
+      "loss": 0.4407,
+      "step": 15358
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.506582441224766e-06,
+      "loss": 0.4917,
+      "step": 15359
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.505341187376753e-06,
+      "loss": 0.4817,
+      "step": 15360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.504100020303885e-06,
+      "loss": 0.4511,
+      "step": 15361
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502858940030125e-06,
+      "loss": 0.4371,
+      "step": 15362
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501617946579434e-06,
+      "loss": 0.432,
+      "step": 15363
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500377039975769e-06,
+      "loss": 0.488,
+      "step": 15364
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49913622024309e-06,
+      "loss": 0.4554,
+      "step": 15365
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.497895487405355e-06,
+      "loss": 0.4379,
+      "step": 15366
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49665484148651e-06,
+      "loss": 0.4382,
+      "step": 15367
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.495414282510511e-06,
+      "loss": 0.4451,
+      "step": 15368
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494173810501308e-06,
+      "loss": 0.4562,
+      "step": 15369
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.492933425482854e-06,
+      "loss": 0.1739,
+      "step": 15370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.491693127479092e-06,
+      "loss": 0.4505,
+      "step": 15371
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490452916513968e-06,
+      "loss": 0.4572,
+      "step": 15372
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4892127926114304e-06,
+      "loss": 0.4449,
+      "step": 15373
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.487972755795414e-06,
+      "loss": 0.4689,
+      "step": 15374
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486732806089863e-06,
+      "loss": 0.4646,
+      "step": 15375
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.485492943518715e-06,
+      "loss": 0.4556,
+      "step": 15376
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.484253168105911e-06,
+      "loss": 0.4584,
+      "step": 15377
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.483013479875381e-06,
+      "loss": 0.4443,
+      "step": 15378
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.481773878851062e-06,
+      "loss": 0.4526,
+      "step": 15379
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480534365056886e-06,
+      "loss": 0.4291,
+      "step": 15380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479294938516786e-06,
+      "loss": 0.4667,
+      "step": 15381
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4780555992546855e-06,
+      "loss": 0.4963,
+      "step": 15382
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.476816347294512e-06,
+      "loss": 0.4525,
+      "step": 15383
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.475577182660192e-06,
+      "loss": 0.4488,
+      "step": 15384
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.474338105375651e-06,
+      "loss": 0.4352,
+      "step": 15385
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.473099115464811e-06,
+      "loss": 0.4668,
+      "step": 15386
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471860212951588e-06,
+      "loss": 0.4554,
+      "step": 15387
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4706213978599055e-06,
+      "loss": 0.4596,
+      "step": 15388
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.469382670213681e-06,
+      "loss": 0.4626,
+      "step": 15389
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468144030036824e-06,
+      "loss": 0.4627,
+      "step": 15390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466905477353251e-06,
+      "loss": 0.4575,
+      "step": 15391
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.465667012186875e-06,
+      "loss": 0.459,
+      "step": 15392
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.464428634561604e-06,
+      "loss": 0.4674,
+      "step": 15393
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463190344501348e-06,
+      "loss": 0.4594,
+      "step": 15394
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4619521420300136e-06,
+      "loss": 0.4427,
+      "step": 15395
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.46071402717151e-06,
+      "loss": 0.4887,
+      "step": 15396
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.459475999949731e-06,
+      "loss": 0.4716,
+      "step": 15397
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.458238060388586e-06,
+      "loss": 0.4423,
+      "step": 15398
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.45700020851197e-06,
+      "loss": 0.4278,
+      "step": 15399
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455762444343785e-06,
+      "loss": 0.442,
+      "step": 15400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.454524767907927e-06,
+      "loss": 0.45,
+      "step": 15401
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453287179228291e-06,
+      "loss": 0.4663,
+      "step": 15402
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.452049678328768e-06,
+      "loss": 0.4838,
+      "step": 15403
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.450812265233256e-06,
+      "loss": 0.487,
+      "step": 15404
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.449574939965637e-06,
+      "loss": 0.4453,
+      "step": 15405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.448337702549802e-06,
+      "loss": 0.4603,
+      "step": 15406
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.447100553009633e-06,
+      "loss": 0.4436,
+      "step": 15407
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.445863491369028e-06,
+      "loss": 0.4529,
+      "step": 15408
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.444626517651858e-06,
+      "loss": 0.4646,
+      "step": 15409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4433896318820065e-06,
+      "loss": 0.4978,
+      "step": 15410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4421528340833565e-06,
+      "loss": 0.451,
+      "step": 15411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4409161242797855e-06,
+      "loss": 0.4782,
+      "step": 15412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4396795024951655e-06,
+      "loss": 0.495,
+      "step": 15413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438442968753374e-06,
+      "loss": 0.4796,
+      "step": 15414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.437206523078285e-06,
+      "loss": 0.4421,
+      "step": 15415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.435970165493767e-06,
+      "loss": 0.4279,
+      "step": 15416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4347338960236915e-06,
+      "loss": 0.4461,
+      "step": 15417
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.433497714691924e-06,
+      "loss": 0.4633,
+      "step": 15418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432261621522339e-06,
+      "loss": 0.4395,
+      "step": 15419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.431025616538787e-06,
+      "loss": 0.4653,
+      "step": 15420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.42978969976514e-06,
+      "loss": 0.174,
+      "step": 15421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4285538712252515e-06,
+      "loss": 0.4514,
+      "step": 15422
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.427318130942994e-06,
+      "loss": 0.4764,
+      "step": 15423
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.426082478942214e-06,
+      "loss": 0.4357,
+      "step": 15424
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.42484691524677e-06,
+      "loss": 0.4433,
+      "step": 15425
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.423611439880516e-06,
+      "loss": 0.4614,
+      "step": 15426
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.422376052867308e-06,
+      "loss": 0.4536,
+      "step": 15427
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.421140754230992e-06,
+      "loss": 0.4528,
+      "step": 15428
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419905543995413e-06,
+      "loss": 0.4357,
+      "step": 15429
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418670422184428e-06,
+      "loss": 0.4567,
+      "step": 15430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.417435388821884e-06,
+      "loss": 0.4744,
+      "step": 15431
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4162004439316125e-06,
+      "loss": 0.4497,
+      "step": 15432
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.414965587537465e-06,
+      "loss": 0.4531,
+      "step": 15433
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413730819663283e-06,
+      "loss": 0.4765,
+      "step": 15434
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412496140332898e-06,
+      "loss": 0.4307,
+      "step": 15435
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.411261549570147e-06,
+      "loss": 0.4499,
+      "step": 15436
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.410027047398874e-06,
+      "loss": 0.4402,
+      "step": 15437
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.408792633842911e-06,
+      "loss": 0.4504,
+      "step": 15438
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.407558308926083e-06,
+      "loss": 0.4641,
+      "step": 15439
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.406324072672224e-06,
+      "loss": 0.4368,
+      "step": 15440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405089925105163e-06,
+      "loss": 0.4303,
+      "step": 15441
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40385586624873e-06,
+      "loss": 0.4778,
+      "step": 15442
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402621896126745e-06,
+      "loss": 0.4536,
+      "step": 15443
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401388014763027e-06,
+      "loss": 0.4771,
+      "step": 15444
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400154222181407e-06,
+      "loss": 0.452,
+      "step": 15445
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.398920518405706e-06,
+      "loss": 0.4897,
+      "step": 15446
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3976869034597335e-06,
+      "loss": 0.4356,
+      "step": 15447
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396453377367312e-06,
+      "loss": 0.4689,
+      "step": 15448
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.395219940152252e-06,
+      "loss": 0.46,
+      "step": 15449
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.393986591838376e-06,
+      "loss": 0.4377,
+      "step": 15450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.392753332449479e-06,
+      "loss": 0.4439,
+      "step": 15451
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3915201620093835e-06,
+      "loss": 0.4452,
+      "step": 15452
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.390287080541895e-06,
+      "loss": 0.475,
+      "step": 15453
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.389054088070822e-06,
+      "loss": 0.4523,
+      "step": 15454
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387821184619964e-06,
+      "loss": 0.5064,
+      "step": 15455
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.386588370213124e-06,
+      "loss": 0.4637,
+      "step": 15456
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.385355644874108e-06,
+      "loss": 0.4445,
+      "step": 15457
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384123008626705e-06,
+      "loss": 0.4375,
+      "step": 15458
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382890461494723e-06,
+      "loss": 0.467,
+      "step": 15459
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381658003501955e-06,
+      "loss": 0.1642,
+      "step": 15460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.380425634672199e-06,
+      "loss": 0.447,
+      "step": 15461
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3791933550292385e-06,
+      "loss": 0.4446,
+      "step": 15462
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3779611645968696e-06,
+      "loss": 0.4432,
+      "step": 15463
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.37672906339888e-06,
+      "loss": 0.4747,
+      "step": 15464
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.375497051459058e-06,
+      "loss": 0.4455,
+      "step": 15465
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.374265128801189e-06,
+      "loss": 0.4321,
+      "step": 15466
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.373033295449057e-06,
+      "loss": 0.4712,
+      "step": 15467
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.371801551426443e-06,
+      "loss": 0.4722,
+      "step": 15468
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.370569896757132e-06,
+      "loss": 0.4685,
+      "step": 15469
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369338331464895e-06,
+      "loss": 0.4512,
+      "step": 15470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.368106855573514e-06,
+      "loss": 0.4657,
+      "step": 15471
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.36687546910676e-06,
+      "loss": 0.4512,
+      "step": 15472
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.365644172088412e-06,
+      "loss": 0.4507,
+      "step": 15473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.364412964542239e-06,
+      "loss": 0.4753,
+      "step": 15474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.363181846492012e-06,
+      "loss": 0.428,
+      "step": 15475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361950817961501e-06,
+      "loss": 0.4529,
+      "step": 15476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360719878974466e-06,
+      "loss": 0.4474,
+      "step": 15477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.359489029554676e-06,
+      "loss": 0.4564,
+      "step": 15478
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3582582697258955e-06,
+      "loss": 0.4609,
+      "step": 15479
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3570275995118844e-06,
+      "loss": 0.4101,
+      "step": 15480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.355797018936401e-06,
+      "loss": 0.4313,
+      "step": 15481
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354566528023207e-06,
+      "loss": 0.499,
+      "step": 15482
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353336126796056e-06,
+      "loss": 0.4735,
+      "step": 15483
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.352105815278706e-06,
+      "loss": 0.4704,
+      "step": 15484
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.350875593494902e-06,
+      "loss": 0.4808,
+      "step": 15485
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3496454614684e-06,
+      "loss": 0.4744,
+      "step": 15486
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34841541922295e-06,
+      "loss": 0.4649,
+      "step": 15487
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.347185466782297e-06,
+      "loss": 0.4476,
+      "step": 15488
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.345955604170189e-06,
+      "loss": 0.4888,
+      "step": 15489
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344725831410369e-06,
+      "loss": 0.48,
+      "step": 15490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.343496148526579e-06,
+      "loss": 0.4802,
+      "step": 15491
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.342266555542565e-06,
+      "loss": 0.434,
+      "step": 15492
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.341037052482056e-06,
+      "loss": 0.4889,
+      "step": 15493
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.339807639368796e-06,
+      "loss": 0.4471,
+      "step": 15494
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.338578316226517e-06,
+      "loss": 0.4537,
+      "step": 15495
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337349083078955e-06,
+      "loss": 0.4483,
+      "step": 15496
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.336119939949842e-06,
+      "loss": 0.4644,
+      "step": 15497
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.334890886862906e-06,
+      "loss": 0.4431,
+      "step": 15498
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333661923841882e-06,
+      "loss": 0.4605,
+      "step": 15499
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.332433050910486e-06,
+      "loss": 0.4395,
+      "step": 15500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.331204268092449e-06,
+      "loss": 0.4647,
+      "step": 15501
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.329975575411494e-06,
+      "loss": 0.4448,
+      "step": 15502
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.328746972891343e-06,
+      "loss": 0.4533,
+      "step": 15503
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.327518460555714e-06,
+      "loss": 0.4786,
+      "step": 15504
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.326290038428326e-06,
+      "loss": 0.4507,
+      "step": 15505
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3250617065328945e-06,
+      "loss": 0.4749,
+      "step": 15506
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32383346489314e-06,
+      "loss": 0.4584,
+      "step": 15507
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.322605313532765e-06,
+      "loss": 0.4322,
+      "step": 15508
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.321377252475486e-06,
+      "loss": 0.4581,
+      "step": 15509
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3201492817450125e-06,
+      "loss": 0.4488,
+      "step": 15510
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.318921401365052e-06,
+      "loss": 0.4516,
+      "step": 15511
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.317693611359307e-06,
+      "loss": 0.4539,
+      "step": 15512
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.316465911751486e-06,
+      "loss": 0.4895,
+      "step": 15513
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.315238302565288e-06,
+      "loss": 0.44,
+      "step": 15514
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3140107838244215e-06,
+      "loss": 0.4685,
+      "step": 15515
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.312783355552575e-06,
+      "loss": 0.4664,
+      "step": 15516
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311556017773447e-06,
+      "loss": 0.425,
+      "step": 15517
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310328770510738e-06,
+      "loss": 0.4424,
+      "step": 15518
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3091016137881366e-06,
+      "loss": 0.4358,
+      "step": 15519
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.307874547629339e-06,
+      "loss": 0.4511,
+      "step": 15520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.306647572058032e-06,
+      "loss": 0.4555,
+      "step": 15521
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305420687097909e-06,
+      "loss": 0.44,
+      "step": 15522
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3041938927726485e-06,
+      "loss": 0.4576,
+      "step": 15523
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.302967189105941e-06,
+      "loss": 0.4537,
+      "step": 15524
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.301740576121468e-06,
+      "loss": 0.4622,
+      "step": 15525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30051405384291e-06,
+      "loss": 0.464,
+      "step": 15526
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.299287622293948e-06,
+      "loss": 0.4487,
+      "step": 15527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.29806128149826e-06,
+      "loss": 0.4626,
+      "step": 15528
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296835031479521e-06,
+      "loss": 0.4795,
+      "step": 15529
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.295608872261411e-06,
+      "loss": 0.4516,
+      "step": 15530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.294382803867591e-06,
+      "loss": 0.4746,
+      "step": 15531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.293156826321739e-06,
+      "loss": 0.4636,
+      "step": 15532
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.29193093964752e-06,
+      "loss": 0.462,
+      "step": 15533
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.290705143868613e-06,
+      "loss": 0.4506,
+      "step": 15534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2894794390086715e-06,
+      "loss": 0.4656,
+      "step": 15535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.288253825091361e-06,
+      "loss": 0.4607,
+      "step": 15536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.287028302140348e-06,
+      "loss": 0.4285,
+      "step": 15537
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285802870179293e-06,
+      "loss": 0.4679,
+      "step": 15538
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284577529231847e-06,
+      "loss": 0.4492,
+      "step": 15539
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283352279321673e-06,
+      "loss": 0.3415,
+      "step": 15540
+    }
+  ],
+  "max_steps": 23310,
+  "num_train_epochs": 3,
+  "total_flos": 5.799335862045403e+20,
+  "trial_name": null,
+  "trial_params": null
+}